diff --git a/llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp b/llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
index f16351fac9e2e..44cad8a862141 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPULegalizerInfo.cpp
@@ -1342,13 +1342,30 @@ AMDGPULegalizerInfo::AMDGPULegalizerInfo(const GCNSubtarget &ST_,
       .scalarize(0);
 
     if (ST.hasVOP3PInsts()) {
-      getActionDefinitionsBuilder({G_SMIN, G_SMAX, G_UMIN, G_UMAX, G_ABS})
-        .legalFor({S32, S16, V2S16})
-        .clampMaxNumElements(0, S16, 2)
-        .minScalar(0, S16)
-        .widenScalarToNextPow2(0)
-        .scalarize(0)
-        .lower();
+      getActionDefinitionsBuilder(G_ABS)
+          .legalFor({S32, S16, V2S16})
+          .clampMaxNumElements(0, S16, 2)
+          .minScalar(0, S16)
+          .widenScalarToNextPow2(0)
+          .scalarize(0)
+          .lower();
+      if (ST.hasIntMinMax64()) {
+        getActionDefinitionsBuilder({G_SMIN, G_SMAX, G_UMIN, G_UMAX})
+            .legalFor({S32, S16, S64, V2S16})
+            .clampMaxNumElements(0, S16, 2)
+            .minScalar(0, S16)
+            .widenScalarToNextPow2(0)
+            .scalarize(0)
+            .lower();
+      } else {
+        getActionDefinitionsBuilder({G_SMIN, G_SMAX, G_UMIN, G_UMAX})
+            .legalFor({S32, S16, V2S16})
+            .clampMaxNumElements(0, S16, 2)
+            .minScalar(0, S16)
+            .widenScalarToNextPow2(0)
+            .scalarize(0)
+            .lower();
+      }
     } else {
       getActionDefinitionsBuilder({G_SMIN, G_SMAX, G_UMIN, G_UMAX, G_ABS})
         .legalFor({S32, S16})
diff --git a/llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp b/llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp
index 306443d25a74f..6c40eb5aa7e6c 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp
@@ -4009,10 +4009,6 @@ AMDGPURegisterBankInfo::getInstrMapping(const MachineInstr &MI) const {
   case AMDGPU::G_SADDE:
   case AMDGPU::G_USUBE:
   case AMDGPU::G_SSUBE:
-  case AMDGPU::G_SMIN:
-  case AMDGPU::G_SMAX:
-  case AMDGPU::G_UMIN:
-  case AMDGPU::G_UMAX:
   case AMDGPU::G_ABS:
   case AMDGPU::G_SHUFFLE_VECTOR:
   case AMDGPU::G_SBFX:
@@ -4022,6 +4018,18 @@ AMDGPURegisterBankInfo::getInstrMapping(const MachineInstr &MI) const {
     if (isSALUMapping(MI))
       return getDefaultMappingSOP(MI);
     return getDefaultMappingVOP(MI);
+  case AMDGPU::G_SMIN:
+  case AMDGPU::G_SMAX:
+  case AMDGPU::G_UMIN:
+  case AMDGPU::G_UMAX:
+    if (isSALUMapping(MI)) {
+      // There are no scalar 64-bit min and max, use vector instruction instead.
+      if (MRI.getType(MI.getOperand(0).getReg()).getSizeInBits() == 64 &&
+          Subtarget.hasIntMinMax64())
+        return getDefaultMappingVOP(MI);
+      return getDefaultMappingSOP(MI);
+    }
+    return getDefaultMappingVOP(MI);
   case AMDGPU::G_FADD:
   case AMDGPU::G_FSUB:
   case AMDGPU::G_FMUL:
diff --git a/llvm/lib/Target/AMDGPU/GCNSubtarget.h b/llvm/lib/Target/AMDGPU/GCNSubtarget.h
index b824c66931288..fba1c5a0f05e1 100644
--- a/llvm/lib/Target/AMDGPU/GCNSubtarget.h
+++ b/llvm/lib/Target/AMDGPU/GCNSubtarget.h
@@ -1532,6 +1532,9 @@ class GCNSubtarget final : public AMDGPUGenSubtargetInfo,
   // instructions.
   bool hasMadU64U32NoCarry() const { return GFX1250Insts; }
 
+  // \returns true if the target has V_{MIN|MAX}_{I|U}64 instructions.
+  bool hasIntMinMax64() const { return GFX1250Insts; }
+
   // \returns true if the target has V_PK_ADD_{MIN|MAX}_{I|U}16 instructions.
   bool hasPkAddMinMaxInsts() const { return GFX1250Insts; }
 
diff --git a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
index fbaf9bc452790..579ca96a76200 100644
--- a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
+++ b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
@@ -909,6 +909,10 @@ SITargetLowering::SITargetLowering(const TargetMachine &TM,
                        Custom);
   }
 
+  if (Subtarget->hasIntMinMax64())
+    setOperationAction({ISD::SMIN, ISD::UMIN, ISD::SMAX, ISD::UMAX}, MVT::i64,
+                       Legal);
+
   setOperationAction(ISD::INTRINSIC_WO_CHAIN,
                      {MVT::Other, MVT::f32, MVT::v4f32, MVT::i16, MVT::f16,
                       MVT::bf16, MVT::v2i16, MVT::v2f16, MVT::v2bf16, MVT::i128,
diff --git a/llvm/lib/Target/AMDGPU/VOP3Instructions.td b/llvm/lib/Target/AMDGPU/VOP3Instructions.td
index 2fef4f029951b..5586dd872fef5 100644
--- a/llvm/lib/Target/AMDGPU/VOP3Instructions.td
+++ b/llvm/lib/Target/AMDGPU/VOP3Instructions.td
@@ -202,6 +202,13 @@ defm V_MAXIMUM_F64 : VOP3Inst <"v_maximum_f64", VOP3_Profile<VOP_F64_F64_F64>, f
 } // End SchedRW = [WriteDoubleAdd]
 } // End SubtargetPredicate = HasIEEEMinimumMaximumInsts, ReadsModeReg = 0, AddedComplexity = 1
 
+let SubtargetPredicate = isGFX1250Plus, SchedRW = [WriteDoubleAdd] in {
+defm V_MAX_I64 : VOP3Inst <"v_max_i64", VOP_I64_I64_I64_DPP, smax>;
+defm V_MAX_U64 : VOP3Inst <"v_max_u64", VOP_I64_I64_I64_DPP, umax>;
+defm V_MIN_I64 : VOP3Inst <"v_min_i64", VOP_I64_I64_I64_DPP, smin>;
+defm V_MIN_U64 : VOP3Inst <"v_min_u64", VOP_I64_I64_I64_DPP, umin>;
+} // End SubtargetPredicate = isGFX1250Plus, SchedRW = [WriteDoubleAdd]
+
 } // End isReMaterializable = 1
 
 let Uses = [MODE, VCC, EXEC] in {
@@ -1810,6 +1817,10 @@ defm V_BITOP3_B32         : VOP3_Real_BITOP3_gfx1250<0x234>;
 defm V_MAD_U32 : VOP3Only_Realtriple_gfx1250<0x235>;
 defm V_MAD_NC_U64_U32 : VOP3Only_Realtriple_gfx1250<0x2fa>;
 defm V_MAD_NC_I64_I32 : VOP3Only_Realtriple_gfx1250<0x2fb>;
+defm V_MIN_U64 : VOP3Only_Realtriple_gfx1250<0x318>;
+defm V_MAX_U64 : VOP3Only_Realtriple_gfx1250<0x319>;
+defm V_MIN_I64 : VOP3Only_Realtriple_gfx1250<0x31a>;
+defm V_MAX_I64 : VOP3Only_Realtriple_gfx1250<0x31b>;
 
 defm V_CVT_PK_FP8_F32  : VOP3Only_Realtriple_t16_and_fake16_gfx12<0x369, "v_cvt_pk_fp8_f32">;
 defm V_CVT_PK_BF8_F32  : VOP3Only_Realtriple_t16_and_fake16_gfx12<0x36a, "v_cvt_pk_bf8_f32">;
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/minmaxabs-i64.ll b/llvm/test/CodeGen/AMDGPU/GlobalISel/minmaxabs-i64.ll
new file mode 100644
index 0000000000000..43c8f46f98cfc
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/minmaxabs-i64.ll
@@ -0,0 +1,192 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -global-isel -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx1250 < %s | FileCheck %s
+
+declare i64 @llvm.umin.i64(i64, i64)
+declare i64 @llvm.umax.i64(i64, i64)
+declare i64 @llvm.smin.i64(i64, i64)
+declare i64 @llvm.smax.i64(i64, i64)
+declare i64 @llvm.abs.i64(i64, i1)
+
+declare <4 x i64> @llvm.umin.v4i64(<4 x i64>, <4 x i64>)
+declare <4 x i64> @llvm.umax.v4i64(<4 x i64>, <4 x i64>)
+declare <4 x i64> @llvm.smin.v4i64(<4 x i64>, <4 x i64>)
+declare <4 x i64> @llvm.smax.v4i64(<4 x i64>, <4 x i64>)
+
+define i64 @test_umin_i64(i64 %a, i64 %b) {
+; CHECK-LABEL: test_umin_i64:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_wait_loadcnt_dscnt 0x0
+; CHECK-NEXT:    s_wait_kmcnt 0x0
+; CHECK-NEXT:    v_min_u64 v[0:1], v[0:1], v[2:3]
+; CHECK-NEXT:    s_set_pc_i64 s[30:31]
+  %r = call i64 @llvm.umin.i64(i64 %a, i64 %b)
+  ret i64 %r
+}
+
+define i64 @test_umax_i64(i64 %a, i64 %b) {
+; CHECK-LABEL: test_umax_i64:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_wait_loadcnt_dscnt 0x0
+; CHECK-NEXT:    s_wait_kmcnt 0x0
+; CHECK-NEXT:    v_max_u64 v[0:1], v[0:1], v[2:3]
+; CHECK-NEXT:    s_set_pc_i64 s[30:31]
+  %r = call i64 @llvm.umax.i64(i64 %a, i64 %b)
+  ret i64 %r
+}
+
+define i64 @test_smin_i64(i64 %a, i64 %b) {
+; CHECK-LABEL: test_smin_i64:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_wait_loadcnt_dscnt 0x0
+; CHECK-NEXT:    s_wait_kmcnt 0x0
+; CHECK-NEXT:    v_min_i64 v[0:1], v[0:1], v[2:3]
+; CHECK-NEXT:    s_set_pc_i64 s[30:31]
+  %r = call i64 @llvm.smin.i64(i64 %a, i64 %b)
+  ret i64 %r
+}
+
+define i64 @test_smax_i64(i64 %a, i64 %b) {
+; CHECK-LABEL: test_smax_i64:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_wait_loadcnt_dscnt 0x0
+; CHECK-NEXT:    s_wait_kmcnt 0x0
+; CHECK-NEXT:    v_max_i64 v[0:1], v[0:1], v[2:3]
+; CHECK-NEXT:    s_set_pc_i64 s[30:31]
+  %r = call i64 @llvm.smax.i64(i64 %a, i64 %b)
+  ret i64 %r
+}
+
+define <4 x i64> @test_umin_v4i64(<4 x i64> %a, <4 x i64> %b) {
+; CHECK-LABEL: test_umin_v4i64:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_wait_loadcnt_dscnt 0x0
+; CHECK-NEXT:    s_wait_kmcnt 0x0
+; CHECK-NEXT:    v_min_u64 v[0:1], v[0:1], v[8:9]
+; CHECK-NEXT:    v_min_u64 v[2:3], v[2:3], v[10:11]
+; CHECK-NEXT:    v_min_u64 v[4:5], v[4:5], v[12:13]
+; CHECK-NEXT:    v_min_u64 v[6:7], v[6:7], v[14:15]
+; CHECK-NEXT:    s_set_pc_i64 s[30:31]
+  %r = call <4 x i64> @llvm.umin.v4i64(<4 x i64> %a, <4 x i64> %b)
+  ret <4 x i64> %r
+}
+
+define <4 x i64> @test_umax_v4i64(<4 x i64> %a, <4 x i64> %b) {
+; CHECK-LABEL: test_umax_v4i64:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_wait_loadcnt_dscnt 0x0
+; CHECK-NEXT:    s_wait_kmcnt 0x0
+; CHECK-NEXT:    v_max_u64 v[0:1], v[0:1], v[8:9]
+; CHECK-NEXT:    v_max_u64 v[2:3], v[2:3], v[10:11]
+; CHECK-NEXT:    v_max_u64 v[4:5], v[4:5], v[12:13]
+; CHECK-NEXT:    v_max_u64 v[6:7], v[6:7], v[14:15]
+; CHECK-NEXT:    s_set_pc_i64 s[30:31]
+  %r = call <4 x i64> @llvm.umax.v4i64(<4 x i64> %a, <4 x i64> %b)
+  ret <4 x i64> %r
+}
+
+define <4 x i64> @test_smin_v4i64(<4 x i64> %a, <4 x i64> %b) {
+; CHECK-LABEL: test_smin_v4i64:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_wait_loadcnt_dscnt 0x0
+; CHECK-NEXT:    s_wait_kmcnt 0x0
+; CHECK-NEXT:    v_min_i64 v[0:1], v[0:1], v[8:9]
+; CHECK-NEXT:    v_min_i64 v[2:3], v[2:3], v[10:11]
+; CHECK-NEXT:    v_min_i64 v[4:5], v[4:5], v[12:13]
+; CHECK-NEXT:    v_min_i64 v[6:7], v[6:7], v[14:15]
+; CHECK-NEXT:    s_set_pc_i64 s[30:31]
+  %r = call <4 x i64> @llvm.smin.v4i64(<4 x i64> %a, <4 x i64> %b)
+  ret <4 x i64> %r
+}
+
+define <4 x i64> @test_smax_v4i64(<4 x i64> %a, <4 x i64> %b) {
+; CHECK-LABEL: test_smax_v4i64:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_wait_loadcnt_dscnt 0x0
+; CHECK-NEXT:    s_wait_kmcnt 0x0
+; CHECK-NEXT:    v_max_i64 v[0:1], v[0:1], v[8:9]
+; CHECK-NEXT:    v_max_i64 v[2:3], v[2:3], v[10:11]
+; CHECK-NEXT:    v_max_i64 v[4:5], v[4:5], v[12:13]
+; CHECK-NEXT:    v_max_i64 v[6:7], v[6:7], v[14:15]
+; CHECK-NEXT:    s_set_pc_i64 s[30:31]
+  %r = call <4 x i64> @llvm.smax.v4i64(<4 x i64> %a, <4 x i64> %b)
+  ret <4 x i64> %r
+}
+
+define i64 @test_abs_i64(i64 %a) {
+; CHECK-LABEL: test_abs_i64:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_wait_loadcnt_dscnt 0x0
+; CHECK-NEXT:    s_wait_kmcnt 0x0
+; CHECK-NEXT:    v_ashrrev_i32_e32 v2, 31, v1
+; CHECK-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1)
+; CHECK-NEXT:    v_mov_b32_e32 v3, v2
+; CHECK-NEXT:    v_add_nc_u64_e32 v[0:1], v[0:1], v[2:3]
+; CHECK-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2)
+; CHECK-NEXT:    v_xor_b32_e32 v0, v0, v2
+; CHECK-NEXT:    v_xor_b32_e32 v1, v1, v2
+; CHECK-NEXT:    s_set_pc_i64 s[30:31]
+  %r = call i64 @llvm.abs.i64(i64 %a, i1 0)
+  ret i64 %r
+}
+
+define amdgpu_ps i64 @test_umin_i64_s(i64 inreg %a, i64 inreg %b) {
+; CHECK-LABEL: test_umin_i64_s:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    v_min_u64 v[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2)
+; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
+; CHECK-NEXT:    v_readfirstlane_b32 s1, v1
+; CHECK-NEXT:    ; return to shader part epilog
+  %r = call i64 @llvm.umin.i64(i64 %a, i64 %b)
+  ret i64 %r
+}
+
+define amdgpu_ps i64 @test_umax_i64_s(i64 inreg %a, i64 inreg %b) {
+; CHECK-LABEL: test_umax_i64_s:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    v_max_u64 v[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2)
+; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
+; CHECK-NEXT:    v_readfirstlane_b32 s1, v1
+; CHECK-NEXT:    ; return to shader part epilog
+  %r = call i64 @llvm.umax.i64(i64 %a, i64 %b)
+  ret i64 %r
+}
+
+define amdgpu_ps i64 @test_smin_i64_s(i64 inreg %a, i64 inreg %b) {
+; CHECK-LABEL: test_smin_i64_s:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    v_min_i64 v[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2)
+; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
+; CHECK-NEXT:    v_readfirstlane_b32 s1, v1
+; CHECK-NEXT:    ; return to shader part epilog
+  %r = call i64 @llvm.smin.i64(i64 %a, i64 %b)
+  ret i64 %r
+}
+
+define amdgpu_ps i64 @test_smax_i64_s(i64 inreg %a, i64 inreg %b) {
+; CHECK-LABEL: test_smax_i64_s:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    v_max_i64 v[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2)
+; CHECK-NEXT:    v_readfirstlane_b32 s0, v0
+; CHECK-NEXT:    v_readfirstlane_b32 s1, v1
+; CHECK-NEXT:    ; return to shader part epilog
+  %r = call i64 @llvm.smax.i64(i64 %a, i64 %b)
+  ret i64 %r
+}
+
+define amdgpu_ps i64 @test_abs_i64_s(i64 inreg %a) {
+; CHECK-LABEL: test_abs_i64_s:
+; CHECK:       ; %bb.0:
+; CHECK-NEXT:    s_ashr_i32 s2, s1, 31
+; CHECK-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; CHECK-NEXT:    s_mov_b32 s3, s2
+; CHECK-NEXT:    s_add_nc_u64 s[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; CHECK-NEXT:    s_xor_b64 s[0:1], s[0:1], s[2:3]
+; CHECK-NEXT:    ; return to shader part epilog
+  %r = call i64 @llvm.abs.i64(i64 %a, i1 0)
+  ret i64 %r
+}
diff --git a/llvm/test/CodeGen/AMDGPU/flat-saddr-atomics.ll b/llvm/test/CodeGen/AMDGPU/flat-saddr-atomics.ll
index bd4ee037a7c6c..2ff66c9b9017a 100644
--- a/llvm/test/CodeGen/AMDGPU/flat-saddr-atomics.ll
+++ b/llvm/test/CodeGen/AMDGPU/flat-saddr-atomics.ll
@@ -281,7 +281,7 @@ define amdgpu_ps <2 x float> @flat_xchg_saddr_i64_rtn(ptr inreg %sbase, i32 %vof
 ; GFX1250-SDAG-NEXT:    v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5]
 ; GFX1250-SDAG-NEXT:    v_cndmask_b32_e32 v4, -1, v4, vcc_lo
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
@@ -322,7 +322,7 @@ define amdgpu_ps <2 x float> @flat_xchg_saddr_i64_rtn(ptr inreg %sbase, i32 %vof
 ; GFX1250-GISEL-NEXT:    v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7]
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e32 v2, -1, v6, vcc_lo
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[4:5], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[4:5], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
@@ -370,7 +370,7 @@ define amdgpu_ps <2 x float> @flat_xchg_saddr_i64_rtn_neg128(ptr inreg %sbase, i
 ; GFX1250-SDAG-NEXT:    v_cmp_ne_u64_e32 vcc_lo, 0, v[4:5]
 ; GFX1250-SDAG-NEXT:    v_cndmask_b32_e32 v4, -1, v4, vcc_lo
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
@@ -414,7 +414,7 @@ define amdgpu_ps <2 x float> @flat_xchg_saddr_i64_rtn_neg128(ptr inreg %sbase, i
 ; GFX1250-GISEL-NEXT:    v_cmp_ne_u64_e32 vcc_lo, 0, v[6:7]
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e32 v2, -1, v6, vcc_lo
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[4:5], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[4:5], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
@@ -458,7 +458,7 @@ define amdgpu_ps void @flat_xchg_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-SDAG-NEXT:    s_delay_alu instid0(VALU_DEP_2)
 ; GFX1250-SDAG-NEXT:    v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1]
 ; GFX1250-SDAG-NEXT:    v_cndmask_b32_e32 v0, -1, v0, vcc_lo
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v0, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v0, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_xchg_saddr_i64_nortn:
@@ -491,7 +491,7 @@ define amdgpu_ps void @flat_xchg_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-GISEL-NEXT:  .LBB12_4: ; %atomicrmw.private
 ; GFX1250-GISEL-NEXT:    v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3]
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e32 v0, -1, v2, vcc_lo
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v0, v[4:5], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v0, v[4:5], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -531,7 +531,7 @@ define amdgpu_ps void @flat_xchg_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v
 ; GFX1250-SDAG-NEXT:  .LBB13_4: ; %atomicrmw.private
 ; GFX1250-SDAG-NEXT:    v_cmp_ne_u64_e32 vcc_lo, 0, v[0:1]
 ; GFX1250-SDAG-NEXT:    v_cndmask_b32_e32 v0, -1, v0, vcc_lo
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v0, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v0, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_xchg_saddr_i64_nortn_neg128:
@@ -567,7 +567,7 @@ define amdgpu_ps void @flat_xchg_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v
 ; GFX1250-GISEL-NEXT:  .LBB13_4: ; %atomicrmw.private
 ; GFX1250-GISEL-NEXT:    v_cmp_ne_u64_e32 vcc_lo, 0, v[2:3]
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e32 v0, -1, v2, vcc_lo
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v0, v[4:5], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v0, v[4:5], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -673,7 +673,7 @@ define amdgpu_ps <2 x float> @flat_add_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_add_nc_u64_e32 v[2:3], v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB18_5
@@ -715,7 +715,7 @@ define amdgpu_ps <2 x float> @flat_add_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v6, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_add_nc_u64_e32 v[2:3], v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB18_5
@@ -764,7 +764,7 @@ define amdgpu_ps <2 x float> @flat_add_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_add_nc_u64_e32 v[2:3], v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB19_5
@@ -809,7 +809,7 @@ define amdgpu_ps <2 x float> @flat_add_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v6, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_add_nc_u64_e32 v[2:3], v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB19_5
@@ -855,7 +855,7 @@ define amdgpu_ps void @flat_add_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_add_nc_u64_e32 v[0:1], v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_add_saddr_i64_nortn:
@@ -891,7 +891,7 @@ define amdgpu_ps void @flat_add_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_add_nc_u64_e32 v[0:1], v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -934,7 +934,7 @@ define amdgpu_ps void @flat_add_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_add_nc_u64_e32 v[0:1], v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_add_saddr_i64_nortn_neg128:
@@ -973,7 +973,7 @@ define amdgpu_ps void @flat_add_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_add_nc_u64_e32 v[0:1], v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -1079,7 +1079,7 @@ define amdgpu_ps <2 x float> @flat_sub_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_sub_nc_u64_e32 v[2:3], v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB26_5
@@ -1121,7 +1121,7 @@ define amdgpu_ps <2 x float> @flat_sub_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v6, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_sub_nc_u64_e32 v[2:3], v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB26_5
@@ -1170,7 +1170,7 @@ define amdgpu_ps <2 x float> @flat_sub_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_sub_nc_u64_e32 v[2:3], v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB27_5
@@ -1215,7 +1215,7 @@ define amdgpu_ps <2 x float> @flat_sub_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v6, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_sub_nc_u64_e32 v[2:3], v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB27_5
@@ -1261,7 +1261,7 @@ define amdgpu_ps void @flat_sub_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_sub_nc_u64_e32 v[0:1], v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_sub_saddr_i64_nortn:
@@ -1297,7 +1297,7 @@ define amdgpu_ps void @flat_sub_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_sub_nc_u64_e32 v[0:1], v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -1340,7 +1340,7 @@ define amdgpu_ps void @flat_sub_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_sub_nc_u64_e32 v[0:1], v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_sub_saddr_i64_nortn_neg128:
@@ -1379,7 +1379,7 @@ define amdgpu_ps void @flat_sub_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_sub_nc_u64_e32 v[0:1], v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -1486,7 +1486,7 @@ define amdgpu_ps <2 x float> @flat_and_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_and_b32_e32 v3, v1, v3
 ; GFX1250-SDAG-NEXT:    v_and_b32_e32 v2, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB34_5
@@ -1529,7 +1529,7 @@ define amdgpu_ps <2 x float> @flat_and_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_and_b32_e32 v2, v0, v4
 ; GFX1250-GISEL-NEXT:    v_and_b32_e32 v3, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB34_5
@@ -1579,7 +1579,7 @@ define amdgpu_ps <2 x float> @flat_and_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_and_b32_e32 v3, v1, v3
 ; GFX1250-SDAG-NEXT:    v_and_b32_e32 v2, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB35_5
@@ -1625,7 +1625,7 @@ define amdgpu_ps <2 x float> @flat_and_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_and_b32_e32 v2, v0, v4
 ; GFX1250-GISEL-NEXT:    v_and_b32_e32 v3, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB35_5
@@ -1672,7 +1672,7 @@ define amdgpu_ps void @flat_and_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_and_b32_e32 v1, v1, v3
 ; GFX1250-SDAG-NEXT:    v_and_b32_e32 v0, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_and_saddr_i64_nortn:
@@ -1709,7 +1709,7 @@ define amdgpu_ps void @flat_and_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_and_b32_e32 v0, v0, v4
 ; GFX1250-GISEL-NEXT:    v_and_b32_e32 v1, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -1753,7 +1753,7 @@ define amdgpu_ps void @flat_and_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_and_b32_e32 v1, v1, v3
 ; GFX1250-SDAG-NEXT:    v_and_b32_e32 v0, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_and_saddr_i64_nortn_neg128:
@@ -1793,7 +1793,7 @@ define amdgpu_ps void @flat_and_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_and_b32_e32 v0, v0, v4
 ; GFX1250-GISEL-NEXT:    v_and_b32_e32 v1, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -1900,7 +1900,7 @@ define amdgpu_ps <2 x float> @flat_or_saddr_i64_rtn(ptr inreg %sbase, i32 %voffs
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_or_b32_e32 v3, v1, v3
 ; GFX1250-SDAG-NEXT:    v_or_b32_e32 v2, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB42_5
@@ -1943,7 +1943,7 @@ define amdgpu_ps <2 x float> @flat_or_saddr_i64_rtn(ptr inreg %sbase, i32 %voffs
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_or_b32_e32 v2, v0, v4
 ; GFX1250-GISEL-NEXT:    v_or_b32_e32 v3, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB42_5
@@ -1993,7 +1993,7 @@ define amdgpu_ps <2 x float> @flat_or_saddr_i64_rtn_neg128(ptr inreg %sbase, i32
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_or_b32_e32 v3, v1, v3
 ; GFX1250-SDAG-NEXT:    v_or_b32_e32 v2, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB43_5
@@ -2039,7 +2039,7 @@ define amdgpu_ps <2 x float> @flat_or_saddr_i64_rtn_neg128(ptr inreg %sbase, i32
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_or_b32_e32 v2, v0, v4
 ; GFX1250-GISEL-NEXT:    v_or_b32_e32 v3, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB43_5
@@ -2086,7 +2086,7 @@ define amdgpu_ps void @flat_or_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, i
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_or_b32_e32 v1, v1, v3
 ; GFX1250-SDAG-NEXT:    v_or_b32_e32 v0, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_or_saddr_i64_nortn:
@@ -2123,7 +2123,7 @@ define amdgpu_ps void @flat_or_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset, i
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_or_b32_e32 v0, v0, v4
 ; GFX1250-GISEL-NEXT:    v_or_b32_e32 v1, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -2167,7 +2167,7 @@ define amdgpu_ps void @flat_or_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vof
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_or_b32_e32 v1, v1, v3
 ; GFX1250-SDAG-NEXT:    v_or_b32_e32 v0, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_or_saddr_i64_nortn_neg128:
@@ -2207,7 +2207,7 @@ define amdgpu_ps void @flat_or_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vof
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_or_b32_e32 v0, v0, v4
 ; GFX1250-GISEL-NEXT:    v_or_b32_e32 v1, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -2314,7 +2314,7 @@ define amdgpu_ps <2 x float> @flat_xor_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_xor_b32_e32 v3, v1, v3
 ; GFX1250-SDAG-NEXT:    v_xor_b32_e32 v2, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB50_5
@@ -2357,7 +2357,7 @@ define amdgpu_ps <2 x float> @flat_xor_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_xor_b32_e32 v2, v0, v4
 ; GFX1250-GISEL-NEXT:    v_xor_b32_e32 v3, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB50_5
@@ -2407,7 +2407,7 @@ define amdgpu_ps <2 x float> @flat_xor_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_xor_b32_e32 v3, v1, v3
 ; GFX1250-SDAG-NEXT:    v_xor_b32_e32 v2, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB51_5
@@ -2453,7 +2453,7 @@ define amdgpu_ps <2 x float> @flat_xor_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_xor_b32_e32 v2, v0, v4
 ; GFX1250-GISEL-NEXT:    v_xor_b32_e32 v3, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB51_5
@@ -2500,7 +2500,7 @@ define amdgpu_ps void @flat_xor_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_xor_b32_e32 v1, v1, v3
 ; GFX1250-SDAG-NEXT:    v_xor_b32_e32 v0, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_xor_saddr_i64_nortn:
@@ -2537,7 +2537,7 @@ define amdgpu_ps void @flat_xor_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_xor_b32_e32 v0, v0, v4
 ; GFX1250-GISEL-NEXT:    v_xor_b32_e32 v1, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -2581,7 +2581,7 @@ define amdgpu_ps void @flat_xor_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_xor_b32_e32 v1, v1, v3
 ; GFX1250-SDAG-NEXT:    v_xor_b32_e32 v0, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_xor_saddr_i64_nortn_neg128:
@@ -2621,7 +2621,7 @@ define amdgpu_ps void @flat_xor_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_xor_b32_e32 v0, v0, v4
 ; GFX1250-GISEL-NEXT:    v_xor_b32_e32 v1, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -2721,9 +2721,8 @@ define amdgpu_ps <2 x float> @flat_max_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_gt_i64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, v3, v1 :: v_dual_cndmask_b32 v2, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    v_max_i64 v[2:3], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB58_5
@@ -2765,9 +2764,8 @@ define amdgpu_ps <2 x float> @flat_max_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v6, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_gt_i64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v2, v4, v0 :: v_dual_cndmask_b32 v3, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    v_max_i64 v[2:3], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB58_5
@@ -2816,9 +2814,8 @@ define amdgpu_ps <2 x float> @flat_max_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_gt_i64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, v3, v1 :: v_dual_cndmask_b32 v2, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    v_max_i64 v[2:3], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB59_5
@@ -2863,9 +2860,8 @@ define amdgpu_ps <2 x float> @flat_max_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v6, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_gt_i64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v2, v4, v0 :: v_dual_cndmask_b32 v3, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    v_max_i64 v[2:3], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB59_5
@@ -2909,9 +2905,8 @@ define amdgpu_ps void @flat_max_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-SDAG-NEXT:    v_cndmask_b32_e32 v4, -1, v0, vcc_lo
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_gt_i64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, v3, v1 :: v_dual_cndmask_b32 v0, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    v_max_i64 v[0:1], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_max_saddr_i64_nortn:
@@ -2945,9 +2940,8 @@ define amdgpu_ps void @flat_max_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e32 v2, -1, v2, vcc_lo
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_gt_i64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v0, v4, v0 :: v_dual_cndmask_b32 v1, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    v_max_i64 v[0:1], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -2988,9 +2982,8 @@ define amdgpu_ps void @flat_max_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-SDAG-NEXT:    v_cndmask_b32_e32 v4, -1, v0, vcc_lo
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_gt_i64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, v3, v1 :: v_dual_cndmask_b32 v0, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    v_max_i64 v[0:1], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_max_saddr_i64_nortn_neg128:
@@ -3027,9 +3020,8 @@ define amdgpu_ps void @flat_max_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e32 v2, -1, v2, vcc_lo
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_gt_i64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v0, v4, v0 :: v_dual_cndmask_b32 v1, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    v_max_i64 v[0:1], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -3129,9 +3121,8 @@ define amdgpu_ps <2 x float> @flat_min_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_le_i64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, v3, v1 :: v_dual_cndmask_b32 v2, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    v_min_i64 v[2:3], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB66_5
@@ -3173,9 +3164,8 @@ define amdgpu_ps <2 x float> @flat_min_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v6, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_lt_i64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v2, v4, v0 :: v_dual_cndmask_b32 v3, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    v_min_i64 v[2:3], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB66_5
@@ -3224,9 +3214,8 @@ define amdgpu_ps <2 x float> @flat_min_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_le_i64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, v3, v1 :: v_dual_cndmask_b32 v2, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    v_min_i64 v[2:3], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB67_5
@@ -3271,9 +3260,8 @@ define amdgpu_ps <2 x float> @flat_min_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v6, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_lt_i64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v2, v4, v0 :: v_dual_cndmask_b32 v3, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    v_min_i64 v[2:3], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB67_5
@@ -3317,9 +3305,8 @@ define amdgpu_ps void @flat_min_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-SDAG-NEXT:    v_cndmask_b32_e32 v4, -1, v0, vcc_lo
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_le_i64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, v3, v1 :: v_dual_cndmask_b32 v0, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    v_min_i64 v[0:1], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_min_saddr_i64_nortn:
@@ -3353,9 +3340,8 @@ define amdgpu_ps void @flat_min_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e32 v2, -1, v2, vcc_lo
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_lt_i64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v0, v4, v0 :: v_dual_cndmask_b32 v1, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    v_min_i64 v[0:1], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -3396,9 +3382,8 @@ define amdgpu_ps void @flat_min_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-SDAG-NEXT:    v_cndmask_b32_e32 v4, -1, v0, vcc_lo
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_le_i64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, v3, v1 :: v_dual_cndmask_b32 v0, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    v_min_i64 v[0:1], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_min_saddr_i64_nortn_neg128:
@@ -3435,9 +3420,8 @@ define amdgpu_ps void @flat_min_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e32 v2, -1, v2, vcc_lo
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_lt_i64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v0, v4, v0 :: v_dual_cndmask_b32 v1, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    v_min_i64 v[0:1], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -3537,9 +3521,8 @@ define amdgpu_ps <2 x float> @flat_umax_saddr_i64_rtn(ptr inreg %sbase, i32 %vof
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_gt_u64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, v3, v1 :: v_dual_cndmask_b32 v2, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    v_max_u64 v[2:3], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB74_5
@@ -3581,9 +3564,8 @@ define amdgpu_ps <2 x float> @flat_umax_saddr_i64_rtn(ptr inreg %sbase, i32 %vof
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v6, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_gt_u64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v2, v4, v0 :: v_dual_cndmask_b32 v3, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    v_max_u64 v[2:3], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB74_5
@@ -3632,9 +3614,8 @@ define amdgpu_ps <2 x float> @flat_umax_saddr_i64_rtn_neg128(ptr inreg %sbase, i
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_gt_u64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, v3, v1 :: v_dual_cndmask_b32 v2, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    v_max_u64 v[2:3], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB75_5
@@ -3679,9 +3660,8 @@ define amdgpu_ps <2 x float> @flat_umax_saddr_i64_rtn_neg128(ptr inreg %sbase, i
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v6, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_gt_u64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v2, v4, v0 :: v_dual_cndmask_b32 v3, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    v_max_u64 v[2:3], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB75_5
@@ -3725,9 +3705,8 @@ define amdgpu_ps void @flat_umax_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-SDAG-NEXT:    v_cndmask_b32_e32 v4, -1, v0, vcc_lo
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_gt_u64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, v3, v1 :: v_dual_cndmask_b32 v0, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    v_max_u64 v[0:1], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_umax_saddr_i64_nortn:
@@ -3761,9 +3740,8 @@ define amdgpu_ps void @flat_umax_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e32 v2, -1, v2, vcc_lo
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_gt_u64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v0, v4, v0 :: v_dual_cndmask_b32 v1, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    v_max_u64 v[0:1], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -3804,9 +3782,8 @@ define amdgpu_ps void @flat_umax_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v
 ; GFX1250-SDAG-NEXT:    v_cndmask_b32_e32 v4, -1, v0, vcc_lo
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_gt_u64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, v3, v1 :: v_dual_cndmask_b32 v0, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    v_max_u64 v[0:1], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_umax_saddr_i64_nortn_neg128:
@@ -3843,9 +3820,8 @@ define amdgpu_ps void @flat_umax_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e32 v2, -1, v2, vcc_lo
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_gt_u64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v0, v4, v0 :: v_dual_cndmask_b32 v1, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    v_max_u64 v[0:1], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -3945,9 +3921,8 @@ define amdgpu_ps <2 x float> @flat_umin_saddr_i64_rtn(ptr inreg %sbase, i32 %vof
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_le_u64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, v3, v1 :: v_dual_cndmask_b32 v2, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    v_min_u64 v[2:3], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB82_5
@@ -3989,9 +3964,8 @@ define amdgpu_ps <2 x float> @flat_umin_saddr_i64_rtn(ptr inreg %sbase, i32 %vof
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v6, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_lt_u64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v2, v4, v0 :: v_dual_cndmask_b32 v3, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    v_min_u64 v[2:3], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB82_5
@@ -4040,9 +4014,8 @@ define amdgpu_ps <2 x float> @flat_umin_saddr_i64_rtn_neg128(ptr inreg %sbase, i
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_le_u64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, v3, v1 :: v_dual_cndmask_b32 v2, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-SDAG-NEXT:    v_min_u64 v[2:3], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB83_5
@@ -4087,9 +4060,8 @@ define amdgpu_ps <2 x float> @flat_umin_saddr_i64_rtn_neg128(ptr inreg %sbase, i
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v6, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_lt_u64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v2, v4, v0 :: v_dual_cndmask_b32 v3, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    v_min_u64 v[2:3], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB83_5
@@ -4133,9 +4105,8 @@ define amdgpu_ps void @flat_umin_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-SDAG-NEXT:    v_cndmask_b32_e32 v4, -1, v0, vcc_lo
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_le_u64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, v3, v1 :: v_dual_cndmask_b32 v0, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    v_min_u64 v[0:1], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_umin_saddr_i64_nortn:
@@ -4169,9 +4140,8 @@ define amdgpu_ps void @flat_umin_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e32 v2, -1, v2, vcc_lo
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_lt_u64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v0, v4, v0 :: v_dual_cndmask_b32 v1, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    v_min_u64 v[0:1], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -4212,9 +4182,8 @@ define amdgpu_ps void @flat_umin_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v
 ; GFX1250-SDAG-NEXT:    v_cndmask_b32_e32 v4, -1, v0, vcc_lo
 ; GFX1250-SDAG-NEXT:    scratch_load_b64 v[0:1], v4, off
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-SDAG-NEXT:    v_cmp_le_u64_e32 vcc_lo, v[0:1], v[2:3]
-; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, v3, v1 :: v_dual_cndmask_b32 v0, v2, v0
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    v_min_u64 v[0:1], v[0:1], v[2:3]
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_umin_saddr_i64_nortn_neg128:
@@ -4251,9 +4220,8 @@ define amdgpu_ps void @flat_umin_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %v
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e32 v2, -1, v2, vcc_lo
 ; GFX1250-GISEL-NEXT:    scratch_load_b64 v[0:1], v2, off
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
-; GFX1250-GISEL-NEXT:    v_cmp_lt_u64_e32 vcc_lo, v[0:1], v[4:5]
-; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v0, v4, v0 :: v_dual_cndmask_b32 v1, v5, v1
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    v_min_u64 v[0:1], v[0:1], v[4:5]
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -4377,7 +4345,7 @@ define amdgpu_ps <2 x float> @flat_cmpxchg_saddr_i64_rtn(ptr inreg %sbase, i32 %
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_cmp_eq_u64_e32 vcc_lo, v[0:1], v[6:7]
 ; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, v1, v5 :: v_dual_cndmask_b32 v2, v0, v4
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v8, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v8, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB90_5
@@ -4423,7 +4391,7 @@ define amdgpu_ps <2 x float> @flat_cmpxchg_saddr_i64_rtn(ptr inreg %sbase, i32 %
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_cmp_eq_u64_e32 vcc_lo, v[0:1], v[8:9]
 ; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v2, v0, v6 :: v_dual_cndmask_b32 v3, v1, v7
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB90_5
@@ -4477,7 +4445,7 @@ define amdgpu_ps <2 x float> @flat_cmpxchg_saddr_i64_rtn_neg128(ptr inreg %sbase
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_cmp_eq_u64_e32 vcc_lo, v[0:1], v[6:7]
 ; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, v1, v5 :: v_dual_cndmask_b32 v2, v0, v4
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v8, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v8, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB91_5
@@ -4526,7 +4494,7 @@ define amdgpu_ps <2 x float> @flat_cmpxchg_saddr_i64_rtn_neg128(ptr inreg %sbase
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_cmp_eq_u64_e32 vcc_lo, v[0:1], v[8:9]
 ; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v2, v0, v6 :: v_dual_cndmask_b32 v3, v1, v7
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v4, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v4, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB91_5
@@ -4577,7 +4545,7 @@ define amdgpu_ps void @flat_cmpxchg_saddr_i64_nortn(ptr inreg %sbase, i32 %voffs
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_cmp_eq_u64_e32 vcc_lo, v[0:1], v[6:7]
 ; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, v1, v5 :: v_dual_cndmask_b32 v0, v0, v4
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_cmpxchg_saddr_i64_nortn:
@@ -4617,7 +4585,7 @@ define amdgpu_ps void @flat_cmpxchg_saddr_i64_nortn(ptr inreg %sbase, i32 %voffs
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_cmp_eq_u64_e32 vcc_lo, v[0:1], v[8:9]
 ; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v0, v0, v6 :: v_dual_cndmask_b32 v1, v1, v7
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -4664,7 +4632,7 @@ define amdgpu_ps void @flat_cmpxchg_saddr_i64_nortn_neg128(ptr inreg %sbase, i32
 ; GFX1250-SDAG-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-SDAG-NEXT:    v_cmp_eq_u64_e32 vcc_lo, v[0:1], v[6:7]
 ; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, v1, v5 :: v_dual_cndmask_b32 v0, v0, v4
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_cmpxchg_saddr_i64_nortn_neg128:
@@ -4707,7 +4675,7 @@ define amdgpu_ps void @flat_cmpxchg_saddr_i64_nortn_neg128(ptr inreg %sbase, i32
 ; GFX1250-GISEL-NEXT:    s_wait_loadcnt 0x0
 ; GFX1250-GISEL-NEXT:    v_cmp_eq_u64_e32 vcc_lo, v[0:1], v[8:9]
 ; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v0, v0, v6 :: v_dual_cndmask_b32 v1, v1, v7
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -4808,7 +4776,7 @@ define amdgpu_ps <2 x float> @flat_inc_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-SDAG-NEXT:    v_cmp_lt_u64_e32 vcc_lo, v[0:1], v[2:3]
 ; GFX1250-SDAG-NEXT:    s_delay_alu instid0(VALU_DEP_2)
 ; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, 0, v5 :: v_dual_cndmask_b32 v2, 0, v4
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB98_5
@@ -4854,7 +4822,7 @@ define amdgpu_ps <2 x float> @flat_inc_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3)
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e64 v2, v2, 0, vcc_lo
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e64 v3, v3, 0, vcc_lo
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB98_5
@@ -4906,7 +4874,7 @@ define amdgpu_ps <2 x float> @flat_inc_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-SDAG-NEXT:    v_cmp_lt_u64_e32 vcc_lo, v[0:1], v[2:3]
 ; GFX1250-SDAG-NEXT:    s_delay_alu instid0(VALU_DEP_2)
 ; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, 0, v5 :: v_dual_cndmask_b32 v2, 0, v4
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-SDAG-NEXT:    s_branch .LBB99_5
@@ -4955,7 +4923,7 @@ define amdgpu_ps <2 x float> @flat_inc_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3)
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e64 v2, v2, 0, vcc_lo
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e64 v3, v3, 0, vcc_lo
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s0
 ; GFX1250-GISEL-NEXT:    s_branch .LBB99_5
@@ -5002,7 +4970,7 @@ define amdgpu_ps void @flat_inc_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-SDAG-NEXT:    v_cmp_lt_u64_e32 vcc_lo, v[0:1], v[2:3]
 ; GFX1250-SDAG-NEXT:    s_delay_alu instid0(VALU_DEP_2)
 ; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, 0, v5 :: v_dual_cndmask_b32 v0, 0, v4
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v6, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v6, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_inc_saddr_i64_nortn:
@@ -5040,7 +5008,7 @@ define amdgpu_ps void @flat_inc_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3)
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e64 v0, v2, 0, vcc_lo
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e64 v1, v3, 0, vcc_lo
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -5084,7 +5052,7 @@ define amdgpu_ps void @flat_inc_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-SDAG-NEXT:    v_cmp_lt_u64_e32 vcc_lo, v[0:1], v[2:3]
 ; GFX1250-SDAG-NEXT:    s_delay_alu instid0(VALU_DEP_2)
 ; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, 0, v5 :: v_dual_cndmask_b32 v0, 0, v4
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v6, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v6, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_inc_saddr_i64_nortn_neg128:
@@ -5125,7 +5093,7 @@ define amdgpu_ps void @flat_inc_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_3)
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e64 v0, v2, 0, vcc_lo
 ; GFX1250-GISEL-NEXT:    v_cndmask_b32_e64 v1, v3, 0, vcc_lo
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -5229,7 +5197,7 @@ define amdgpu_ps <2 x float> @flat_dec_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-SDAG-NEXT:    s_or_b32 vcc_lo, vcc_lo, s0
 ; GFX1250-SDAG-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, v5, v3 :: v_dual_cndmask_b32 v2, v4, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s1
 ; GFX1250-SDAG-NEXT:    s_branch .LBB106_5
@@ -5276,7 +5244,7 @@ define amdgpu_ps <2 x float> @flat_dec_saddr_i64_rtn(ptr inreg %sbase, i32 %voff
 ; GFX1250-GISEL-NEXT:    s_or_b32 vcc_lo, vcc_lo, s0
 ; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v2, v2, v4 :: v_dual_cndmask_b32 v3, v3, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s1
 ; GFX1250-GISEL-NEXT:    s_branch .LBB106_5
@@ -5330,7 +5298,7 @@ define amdgpu_ps <2 x float> @flat_dec_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-SDAG-NEXT:    s_or_b32 vcc_lo, vcc_lo, s0
 ; GFX1250-SDAG-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v3, v5, v3 :: v_dual_cndmask_b32 v2, v4, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-SDAG-NEXT:    s_or_b32 exec_lo, exec_lo, s1
 ; GFX1250-SDAG-NEXT:    s_branch .LBB107_5
@@ -5380,7 +5348,7 @@ define amdgpu_ps <2 x float> @flat_dec_saddr_i64_rtn_neg128(ptr inreg %sbase, i3
 ; GFX1250-GISEL-NEXT:    s_or_b32 vcc_lo, vcc_lo, s0
 ; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v2, v2, v4 :: v_dual_cndmask_b32 v3, v3, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v6, v[2:3], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_wait_xcnt 0x0
 ; GFX1250-GISEL-NEXT:    s_or_b32 exec_lo, exec_lo, s1
 ; GFX1250-GISEL-NEXT:    s_branch .LBB107_5
@@ -5429,7 +5397,7 @@ define amdgpu_ps void @flat_dec_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-SDAG-NEXT:    s_or_b32 vcc_lo, vcc_lo, s0
 ; GFX1250-SDAG-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, v1, v3 :: v_dual_cndmask_b32 v0, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_dec_saddr_i64_nortn:
@@ -5468,7 +5436,7 @@ define amdgpu_ps void @flat_dec_saddr_i64_nortn(ptr inreg %sbase, i32 %voffset,
 ; GFX1250-GISEL-NEXT:    s_or_b32 vcc_lo, vcc_lo, s0
 ; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v0, v0, v4 :: v_dual_cndmask_b32 v1, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
@@ -5514,7 +5482,7 @@ define amdgpu_ps void @flat_dec_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-SDAG-NEXT:    s_or_b32 vcc_lo, vcc_lo, s0
 ; GFX1250-SDAG-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1250-SDAG-NEXT:    v_dual_cndmask_b32 v1, v1, v3 :: v_dual_cndmask_b32 v0, v0, v2
-; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off
+; GFX1250-SDAG-NEXT:    scratch_store_b64 v4, v[0:1], off scope:SCOPE_SE
 ; GFX1250-SDAG-NEXT:    s_endpgm
 ;
 ; GFX1250-GISEL-LABEL: flat_dec_saddr_i64_nortn_neg128:
@@ -5556,7 +5524,7 @@ define amdgpu_ps void @flat_dec_saddr_i64_nortn_neg128(ptr inreg %sbase, i32 %vo
 ; GFX1250-GISEL-NEXT:    s_or_b32 vcc_lo, vcc_lo, s0
 ; GFX1250-GISEL-NEXT:    s_delay_alu instid0(VALU_DEP_1)
 ; GFX1250-GISEL-NEXT:    v_dual_cndmask_b32 v0, v0, v4 :: v_dual_cndmask_b32 v1, v1, v5
-; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off
+; GFX1250-GISEL-NEXT:    scratch_store_b64 v2, v[0:1], off scope:SCOPE_SE
 ; GFX1250-GISEL-NEXT:    s_endpgm
   %zext.offset = zext i32 %voffset to i64
   %gep0 = getelementptr inbounds i8, ptr %sbase, i64 %zext.offset
diff --git a/llvm/test/CodeGen/AMDGPU/max.ll b/llvm/test/CodeGen/AMDGPU/max.ll
index b9b29b78d5c2a..fef9a9ae07fb1 100644
--- a/llvm/test/CodeGen/AMDGPU/max.ll
+++ b/llvm/test/CodeGen/AMDGPU/max.ll
@@ -1,5 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 3
 ; RUN: llc -mtriple=amdgcn -mcpu=pitcairn < %s | FileCheck -enable-var-scope -check-prefix=SI %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx1250 --amdgpu-enable-vopd=0 < %s | FileCheck -enable-var-scope -check-prefix=GFX1250 %s
 ; RUN: llc -mtriple=r600 -mcpu=cypress < %s | FileCheck -enable-var-scope -check-prefix=EG %s
 
 define amdgpu_kernel void @v_test_imax_sge_i32(ptr addrspace(1) %out, ptr addrspace(1) %aptr, ptr addrspace(1) %bptr) nounwind {
@@ -24,6 +25,23 @@ define amdgpu_kernel void @v_test_imax_sge_i32(ptr addrspace(1) %out, ptr addrsp
 ; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: v_test_imax_sge_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x34
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    v_mov_b32_e32 v1, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    global_load_b32 v0, v0, s[0:1] scale_offset
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_load_b32 s2, s[2:3], 0x0
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_max_i32_e32 v0, s2, v0
+; GFX1250-NEXT:    global_store_b32 v1, v0, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: v_test_imax_sge_i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 3, @10, KC0[CB0:0-32], KC1[]
@@ -80,6 +98,26 @@ define amdgpu_kernel void @v_test_imax_sge_v4i32(ptr addrspace(1) %out, ptr addr
 ; SI-NEXT:    buffer_store_dwordx4 v[0:3], off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: v_test_imax_sge_v4i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x34
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    v_mov_b32_e32 v4, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    global_load_b128 v[0:3], v0, s[0:1] scale_offset
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_load_b128 s[4:7], s[2:3], 0x0
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_max_i32_e32 v3, s7, v3
+; GFX1250-NEXT:    v_max_i32_e32 v2, s6, v2
+; GFX1250-NEXT:    v_max_i32_e32 v1, s5, v1
+; GFX1250-NEXT:    v_max_i32_e32 v0, s4, v0
+; GFX1250-NEXT:    global_store_b128 v4, v[0:3], s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: v_test_imax_sge_v4i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 3, @10, KC0[CB0:0-32], KC1[]
@@ -127,6 +165,17 @@ define amdgpu_kernel void @s_test_imax_sge_i32(ptr addrspace(1) %out, i32 %a, i3
 ; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: s_test_imax_sge_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_max_i32 s2, s2, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: s_test_imax_sge_i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 2, @4, KC0[CB0:0-32], KC1[]
@@ -156,6 +205,17 @@ define amdgpu_kernel void @s_test_imax_sge_imm_i32(ptr addrspace(1) %out, i32 %a
 ; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: s_test_imax_sge_imm_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b96 s[0:2], s[4:5], 0x24
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_max_i32 s2, s2, 9
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: s_test_imax_sge_imm_i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 2, @4, KC0[CB0:0-32], KC1[]
@@ -195,6 +255,23 @@ define amdgpu_kernel void @v_test_imax_sge_i8(ptr addrspace(1) %out, ptr addrspa
 ; SI-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: v_test_imax_sge_i8:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x34
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_load_i8 s2, s[2:3], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_i8 s3, s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_max_i32 s2, s2, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b8 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: v_test_imax_sge_i8:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 0, @12, KC0[CB0:0-32], KC1[]
@@ -250,6 +327,17 @@ define amdgpu_kernel void @s_test_imax_sgt_imm_i32(ptr addrspace(1) %out, i32 %a
 ; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: s_test_imax_sgt_imm_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b96 s[0:2], s[4:5], 0x24
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_max_i32 s2, s2, 9
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: s_test_imax_sgt_imm_i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 2, @4, KC0[CB0:0-32], KC1[]
@@ -282,6 +370,18 @@ define amdgpu_kernel void @s_test_imax_sgt_imm_v2i32(ptr addrspace(1) %out, <2 x
 ; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: s_test_imax_sgt_imm_v2i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_max_i32 s2, s2, 9
+; GFX1250-NEXT:    s_max_i32 s3, s3, 9
+; GFX1250-NEXT:    v_mov_b32_e32 v0, s2
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s3
+; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: s_test_imax_sgt_imm_v2i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 4, @4, KC0[CB0:0-32], KC1[]
@@ -322,6 +422,23 @@ define amdgpu_kernel void @v_test_imax_sgt_i32(ptr addrspace(1) %out, ptr addrsp
 ; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: v_test_imax_sgt_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x34
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    v_mov_b32_e32 v1, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    global_load_b32 v0, v0, s[0:1] scale_offset
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_load_b32 s2, s[2:3], 0x0
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_max_i32_e32 v0, s2, v0
+; GFX1250-NEXT:    global_store_b32 v1, v0, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: v_test_imax_sgt_i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 3, @10, KC0[CB0:0-32], KC1[]
@@ -366,6 +483,17 @@ define amdgpu_kernel void @s_test_imax_sgt_i32(ptr addrspace(1) %out, i32 %a, i3
 ; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: s_test_imax_sgt_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_max_i32 s2, s2, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: s_test_imax_sgt_i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 2, @4, KC0[CB0:0-32], KC1[]
@@ -404,6 +532,23 @@ define amdgpu_kernel void @v_test_umax_uge_i32(ptr addrspace(1) %out, ptr addrsp
 ; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: v_test_umax_uge_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x34
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    v_mov_b32_e32 v1, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    global_load_b32 v0, v0, s[0:1] scale_offset
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_load_b32 s2, s[2:3], 0x0
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_max_u32_e32 v0, s2, v0
+; GFX1250-NEXT:    global_store_b32 v1, v0, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: v_test_umax_uge_i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 3, @10, KC0[CB0:0-32], KC1[]
@@ -448,6 +593,17 @@ define amdgpu_kernel void @s_test_umax_uge_i32(ptr addrspace(1) %out, i32 %a, i3
 ; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: s_test_umax_uge_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_max_u32 s2, s2, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: s_test_umax_uge_i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 2, @4, KC0[CB0:0-32], KC1[]
@@ -483,6 +639,22 @@ define amdgpu_kernel void @s_test_umax_uge_v3i32(ptr addrspace(1) %out, <3 x i32
 ; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: s_test_umax_uge_v3i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    s_load_b256 s[8:15], s[4:5], 0x34
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x24
+; GFX1250-NEXT:    v_mov_b32_e32 v3, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_max_u32 s2, s10, s14
+; GFX1250-NEXT:    s_max_u32 s3, s8, s12
+; GFX1250-NEXT:    s_max_u32 s4, s9, s13
+; GFX1250-NEXT:    v_mov_b32_e32 v0, s3
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s4
+; GFX1250-NEXT:    v_mov_b32_e32 v2, s2
+; GFX1250-NEXT:    global_store_b96 v3, v[0:2], s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: s_test_umax_uge_v3i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 7, @4, KC0[CB0:0-32], KC1[]
@@ -527,6 +699,23 @@ define amdgpu_kernel void @v_test_umax_uge_i8(ptr addrspace(1) %out, ptr addrspa
 ; SI-NEXT:    buffer_store_byte v0, off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: v_test_umax_uge_i8:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x34
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_load_u8 s2, s[2:3], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_u8 s3, s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_max_u32 s2, s2, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b8 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: v_test_umax_uge_i8:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 1, @10, KC0[CB0:0-32], KC1[]
@@ -581,6 +770,22 @@ define amdgpu_kernel void @v_test_umax_ugt_i32(ptr addrspace(1) %out, ptr addrsp
 ; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: v_test_umax_ugt_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x34
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    v_mov_b32_e32 v1, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    global_load_b32 v0, v0, s[0:1] scale_offset
+; GFX1250-NEXT:    s_load_b32 s2, s[0:1], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x24
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_max_u32_e32 v0, s2, v0
+; GFX1250-NEXT:    global_store_b32 v1, v0, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: v_test_umax_ugt_i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 3, @10, KC0[CB0:0-32], KC1[]
@@ -625,6 +830,17 @@ define amdgpu_kernel void @s_test_umax_ugt_i32(ptr addrspace(1) %out, i32 %a, i3
 ; SI-NEXT:    buffer_store_dword v0, off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: s_test_umax_ugt_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_max_u32 s2, s2, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: s_test_umax_ugt_i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 2, @4, KC0[CB0:0-32], KC1[]
@@ -657,6 +873,18 @@ define amdgpu_kernel void @s_test_umax_ugt_imm_v2i32(ptr addrspace(1) %out, <2 x
 ; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: s_test_umax_ugt_imm_v2i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_max_u32 s2, s2, 15
+; GFX1250-NEXT:    s_max_u32 s3, s3, 23
+; GFX1250-NEXT:    v_mov_b32_e32 v0, s2
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s3
+; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: s_test_umax_ugt_imm_v2i32:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 4, @4, KC0[CB0:0-32], KC1[]
@@ -693,6 +921,22 @@ define amdgpu_kernel void @simplify_demanded_bits_test_umax_ugt_i16(ptr addrspac
 ; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: simplify_demanded_bits_test_umax_ugt_i16:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_clause 0x2
+; GFX1250-NEXT:    s_load_b32 s2, s[4:5], 0x4c
+; GFX1250-NEXT:    s_load_b32 s3, s[4:5], 0x70
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x24
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_and_b32 s2, s2, 0xffff
+; GFX1250-NEXT:    s_and_b32 s3, s3, 0xffff
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-NEXT:    s_max_u32 s2, s2, s3
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: simplify_demanded_bits_test_umax_ugt_i16:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 0, @10, KC0[], KC1[]
@@ -740,6 +984,22 @@ define amdgpu_kernel void @simplify_demanded_bits_test_max_slt_i16(ptr addrspace
 ; SI-NEXT:    buffer_store_dword v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: simplify_demanded_bits_test_max_slt_i16:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_clause 0x2
+; GFX1250-NEXT:    s_load_b32 s2, s[4:5], 0x4c
+; GFX1250-NEXT:    s_load_b32 s3, s[4:5], 0x70
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x24
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_sext_i32_i16 s2, s2
+; GFX1250-NEXT:    s_sext_i32_i16 s3, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-NEXT:    s_max_i32 s2, s2, s3
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: simplify_demanded_bits_test_max_slt_i16:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 0, @10, KC0[], KC1[]
@@ -786,6 +1046,22 @@ define amdgpu_kernel void @s_test_imax_sge_i16(ptr addrspace(1) %out, [8 x i32],
 ; SI-NEXT:    buffer_store_short v0, off, s[0:3], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: s_test_imax_sge_i16:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_clause 0x2
+; GFX1250-NEXT:    s_load_b32 s2, s[4:5], 0x70
+; GFX1250-NEXT:    s_load_b32 s3, s[4:5], 0x4c
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x24
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_sext_i32_i16 s2, s2
+; GFX1250-NEXT:    s_sext_i32_i16 s3, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-NEXT:    s_max_i32 s2, s3, s2
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b16 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: s_test_imax_sge_i16:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 0, @10, KC0[], KC1[]
@@ -844,6 +1120,17 @@ define amdgpu_kernel void @test_umax_ugt_i64(ptr addrspace(1) %out, i64 %a, i64
 ; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: test_umax_ugt_i64:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x34
+; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_max_u64 v[0:1], s[2:3], s[4:5]
+; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: test_umax_ugt_i64:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 7, @4, KC0[CB0:0-32], KC1[]
@@ -886,6 +1173,17 @@ define amdgpu_kernel void @test_umax_uge_i64(ptr addrspace(1) %out, i64 %a, i64
 ; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: test_umax_uge_i64:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x34
+; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_max_u64 v[0:1], s[2:3], s[4:5]
+; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: test_umax_uge_i64:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 7, @4, KC0[CB0:0-32], KC1[]
@@ -928,6 +1226,17 @@ define amdgpu_kernel void @test_imax_sgt_i64(ptr addrspace(1) %out, i64 %a, i64
 ; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: test_imax_sgt_i64:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x34
+; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_max_i64 v[0:1], s[2:3], s[4:5]
+; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: test_imax_sgt_i64:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 7, @4, KC0[CB0:0-32], KC1[]
@@ -970,6 +1279,17 @@ define amdgpu_kernel void @test_imax_sge_i64(ptr addrspace(1) %out, i64 %a, i64
 ; SI-NEXT:    buffer_store_dwordx2 v[0:1], off, s[4:7], 0
 ; SI-NEXT:    s_endpgm
 ;
+; GFX1250-LABEL: test_imax_sge_i64:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x34
+; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_max_i64 v[0:1], s[2:3], s[4:5]
+; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
+; GFX1250-NEXT:    s_endpgm
+;
 ; EG-LABEL: test_imax_sge_i64:
 ; EG:       ; %bb.0:
 ; EG-NEXT:    ALU 7, @4, KC0[CB0:0-32], KC1[]
diff --git a/llvm/test/CodeGen/AMDGPU/min.ll b/llvm/test/CodeGen/AMDGPU/min.ll
index bf2ddc178ac63..721f9741ba418 100644
--- a/llvm/test/CodeGen/AMDGPU/min.ll
+++ b/llvm/test/CodeGen/AMDGPU/min.ll
@@ -6,6 +6,10 @@
 ; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1010 < %s | FileCheck --check-prefix=GFX10 %s
 ; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1100 -mattr=+real-true16 -amdgpu-enable-vopd=0 < %s | FileCheck --check-prefixes=GFX11,GFX11-TRUE16 %s
 ; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1100 -mattr=-real-true16 -amdgpu-enable-vopd=0 < %s | FileCheck --check-prefixes=GFX11,GFX11-FAKE16 %s
+; TODO: FIXME-TRUE16 - Enable this llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1250 -mattr=+real-true16 -amdgpu-enable-vopd=0 < %s | FileCheck --check-prefixes=GFX1250,GFX1250-TRUE16 %s
+; Crashing on v_test_imin_slt_i16
+; LLVM ERROR: Cannot select: 0x5f895f65b050: i16,ch = load<(load (s16) from %ir.b.gep, addrspace 1)>
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1250 -mattr=-real-true16 -amdgpu-enable-vopd=0 < %s | FileCheck --check-prefixes=GFX1250,GFX1250-FAKE16 %s
 
 define amdgpu_kernel void @v_test_imin_sle_i32(ptr addrspace(1) %out, ptr addrspace(1) %a.ptr, ptr addrspace(1) %b.ptr) #0 {
 ; EG-LABEL: v_test_imin_sle_i32:
@@ -124,6 +128,21 @@ define amdgpu_kernel void @v_test_imin_sle_i32(ptr addrspace(1) %out, ptr addrsp
 ; GFX11-NEXT:    v_min_i32_e32 v1, v1, v2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: v_test_imin_sle_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_load_b32 v1, v0, s[2:3] scale_offset
+; GFX1250-NEXT:    global_load_b32 v2, v0, s[4:5] scale_offset
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    v_min_i32_e32 v1, v1, v2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1] scale_offset
+; GFX1250-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %a.gep = getelementptr inbounds i32, ptr addrspace(1) %a.ptr, i32 %tid
   %b.gep = getelementptr inbounds i32, ptr addrspace(1) %b.ptr, i32 %tid
@@ -206,6 +225,17 @@ define amdgpu_kernel void @s_test_imin_sle_i32(ptr addrspace(1) %out, i32 %a, i3
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_imin_sle_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_min_i32 s2, s2, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp sle i32 %a, %b
   %val = select i1 %cmp, i32 %a, i32 %b
   store i32 %val, ptr addrspace(1) %out, align 4
@@ -282,6 +312,17 @@ define amdgpu_kernel void @s_test_imin_sle_v1i32(ptr addrspace(1) %out, <1 x i32
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_imin_sle_v1i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_min_i32 s2, s2, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp sle <1 x i32> %a, %b
   %val = select <1 x i1> %cmp, <1 x i32> %a, <1 x i32> %b
   store <1 x i32> %val, ptr addrspace(1) %out
@@ -397,6 +438,24 @@ define amdgpu_kernel void @s_test_imin_sle_v4i32(ptr addrspace(1) %out, <4 x i32
 ; GFX11-NEXT:    v_mov_b32_e32 v3, s2
 ; GFX11-NEXT:    global_store_b128 v4, v[0:3], s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_imin_sle_v4i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    s_load_b256 s[8:15], s[4:5], 0x10
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v4, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_min_i32 s2, s11, s15
+; GFX1250-NEXT:    s_min_i32 s3, s10, s14
+; GFX1250-NEXT:    s_min_i32 s4, s8, s12
+; GFX1250-NEXT:    s_min_i32 s5, s9, s13
+; GFX1250-NEXT:    v_mov_b32_e32 v0, s4
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s5
+; GFX1250-NEXT:    v_mov_b32_e32 v2, s3
+; GFX1250-NEXT:    v_mov_b32_e32 v3, s2
+; GFX1250-NEXT:    global_store_b128 v4, v[0:3], s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp sle <4 x i32> %a, %b
   %val = select <4 x i1> %cmp, <4 x i32> %a, <4 x i32> %b
   store <4 x i32> %val, ptr addrspace(1) %out
@@ -514,6 +573,22 @@ define amdgpu_kernel void @s_test_imin_sle_i8(ptr addrspace(1) %out, [8 x i32],
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX11-NEXT:    global_store_b8 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_imin_sle_i8:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_clause 0x2
+; GFX1250-NEXT:    s_load_b32 s2, s[4:5], 0x4c
+; GFX1250-NEXT:    s_load_b32 s3, s[4:5], 0x28
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_sext_i32_i8 s2, s2
+; GFX1250-NEXT:    s_sext_i32_i8 s3, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-NEXT:    s_min_i32 s2, s3, s2
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b8 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp sle i8 %a, %b
   %val = select i1 %cmp, i8 %a, i8 %b
   store i8 %val, ptr addrspace(1) %out
@@ -753,6 +828,42 @@ define amdgpu_kernel void @s_test_imin_sle_v4i8(ptr addrspace(1) %out, [8 x i32]
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_imin_sle_v4i8:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_clause 0x2
+; GFX1250-NEXT:    s_load_b32 s2, s[4:5], 0x28
+; GFX1250-NEXT:    s_load_b32 s3, s[4:5], 0x4c
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_sext_i32_i16 s5, s2
+; GFX1250-NEXT:    s_sext_i32_i16 s7, s3
+; GFX1250-NEXT:    s_ashr_i32 s4, s2, 24
+; GFX1250-NEXT:    s_ashr_i32 s6, s3, 24
+; GFX1250-NEXT:    s_sext_i32_i8 s8, s3
+; GFX1250-NEXT:    s_sext_i32_i8 s9, s2
+; GFX1250-NEXT:    s_bfe_i32 s3, s3, 0x80010
+; GFX1250-NEXT:    s_bfe_i32 s2, s2, 0x80010
+; GFX1250-NEXT:    s_ashr_i32 s7, s7, 8
+; GFX1250-NEXT:    s_ashr_i32 s5, s5, 8
+; GFX1250-NEXT:    s_min_i32 s8, s9, s8
+; GFX1250-NEXT:    s_min_i32 s4, s4, s6
+; GFX1250-NEXT:    s_min_i32 s2, s2, s3
+; GFX1250-NEXT:    s_min_i32 s3, s5, s7
+; GFX1250-NEXT:    s_and_b32 s5, s8, 0xff
+; GFX1250-NEXT:    s_lshl_b32 s4, s4, 8
+; GFX1250-NEXT:    s_lshl_b32 s3, s3, 8
+; GFX1250-NEXT:    s_and_b32 s2, s2, 0xff
+; GFX1250-NEXT:    s_or_b32 s3, s5, s3
+; GFX1250-NEXT:    s_or_b32 s2, s2, s4
+; GFX1250-NEXT:    s_and_b32 s3, s3, 0xffff
+; GFX1250-NEXT:    s_lshl_b32 s2, s2, 16
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-NEXT:    s_or_b32 s2, s3, s2
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp sle <4 x i8> %a, %b
   %val = select <4 x i1> %cmp, <4 x i8> %a, <4 x i8> %b
   store <4 x i8> %val, ptr addrspace(1) %out
@@ -862,6 +973,15 @@ define amdgpu_kernel void @s_test_imin_sle_v2i16(ptr addrspace(1) %out, <2 x i16
 ; GFX11-NEXT:    v_pk_min_i16 v1, s2, s3
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_imin_sle_v2i16:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_pk_min_i16 v1, s2, s3
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp sle <2 x i16> %a, %b
   %val = select <2 x i1> %cmp, <2 x i16> %a, <2 x i16> %b
   store <2 x i16> %val, ptr addrspace(1) %out
@@ -998,6 +1118,18 @@ define amdgpu_kernel void @s_test_imin_sle_v4i16(ptr addrspace(1) %out, <4 x i16
 ; GFX11-NEXT:    v_pk_min_i16 v0, s0, s2
 ; GFX11-NEXT:    global_store_b64 v2, v[0:1], s[4:5]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_imin_sle_v4i16:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x8
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_pk_min_i16 v1, s1, s3
+; GFX1250-NEXT:    v_pk_min_i16 v0, s0, s2
+; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[4:5]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp sle <4 x i16> %a, %b
   %val = select <4 x i1> %cmp, <4 x i16> %a, <4 x i16> %b
   store <4 x i16> %val, ptr addrspace(1) %out
@@ -1121,6 +1253,21 @@ define amdgpu_kernel void @v_test_imin_slt_i32(ptr addrspace(1) %out, ptr addrsp
 ; GFX11-NEXT:    v_min_i32_e32 v1, v1, v2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: v_test_imin_slt_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_load_b32 v1, v0, s[2:3] scale_offset
+; GFX1250-NEXT:    global_load_b32 v2, v0, s[4:5] scale_offset
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    v_min_i32_e32 v1, v1, v2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1] scale_offset
+; GFX1250-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %a.gep = getelementptr inbounds i32, ptr addrspace(1) %aptr, i32 %tid
   %b.gep = getelementptr inbounds i32, ptr addrspace(1) %bptr, i32 %tid
@@ -1283,6 +1430,21 @@ define amdgpu_kernel void @v_test_imin_slt_i16(ptr addrspace(1) %out, ptr addrsp
 ; GFX11-FAKE16-NEXT:    v_min_i16 v1, v1, v2
 ; GFX11-FAKE16-NEXT:    global_store_b16 v0, v1, s[0:1]
 ; GFX11-FAKE16-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: v_test_imin_slt_i16:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_load_u16 v1, v0, s[2:3] scale_offset
+; GFX1250-NEXT:    global_load_u16 v2, v0, s[4:5] scale_offset
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    v_min_i16 v1, v1, v2
+; GFX1250-NEXT:    global_store_b16 v0, v1, s[0:1] scale_offset
+; GFX1250-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %a.gep = getelementptr inbounds i16, ptr addrspace(1) %aptr, i32 %tid
   %b.gep = getelementptr inbounds i16, ptr addrspace(1) %bptr, i32 %tid
@@ -1366,6 +1528,17 @@ define amdgpu_kernel void @s_test_imin_slt_i32(ptr addrspace(1) %out, i32 %a, i3
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_imin_slt_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_min_i32 s2, s2, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp slt i32 %a, %b
   %val = select i1 %cmp, i32 %a, i32 %b
   store i32 %val, ptr addrspace(1) %out, align 4
@@ -1459,6 +1632,20 @@ define amdgpu_kernel void @s_test_imin_slt_v2i32(ptr addrspace(1) %out, <2 x i32
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s1
 ; GFX11-NEXT:    global_store_b64 v2, v[0:1], s[4:5]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_imin_slt_v2i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x8
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_min_i32 s0, s0, s2
+; GFX1250-NEXT:    s_min_i32 s1, s1, s3
+; GFX1250-NEXT:    v_mov_b32_e32 v0, s0
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s1
+; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[4:5]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp slt <2 x i32> %a, %b
   %val = select <2 x i1> %cmp, <2 x i32> %a, <2 x i32> %b
   store <2 x i32> %val, ptr addrspace(1) %out
@@ -1542,6 +1729,17 @@ define amdgpu_kernel void @s_test_imin_slt_imm_i32(ptr addrspace(1) %out, i32 %a
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_imin_slt_imm_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b96 s[0:2], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_min_i32 s2, s2, 8
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp slt i32 %a, 8
   %val = select i1 %cmp, i32 %a, i32 8
   store i32 %val, ptr addrspace(1) %out, align 4
@@ -1625,6 +1823,17 @@ define amdgpu_kernel void @s_test_imin_sle_imm_i32(ptr addrspace(1) %out, i32 %a
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_imin_sle_imm_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b96 s[0:2], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_min_i32 s2, s2, 8
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp sle i32 %a, 8
   %val = select i1 %cmp, i32 %a, i32 8
   store i32 %val, ptr addrspace(1) %out, align 4
@@ -1748,6 +1957,21 @@ define amdgpu_kernel void @v_test_umin_ule_i32(ptr addrspace(1) %out, ptr addrsp
 ; GFX11-NEXT:    v_min_u32_e32 v1, v1, v2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: v_test_umin_ule_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_load_b32 v1, v0, s[2:3] scale_offset
+; GFX1250-NEXT:    global_load_b32 v2, v0, s[4:5] scale_offset
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    v_min_u32_e32 v1, v1, v2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1] scale_offset
+; GFX1250-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %a.gep = getelementptr inbounds i32, ptr addrspace(1) %a.ptr, i32 %tid
   %b.gep = getelementptr inbounds i32, ptr addrspace(1) %b.ptr, i32 %tid
@@ -1893,6 +2117,25 @@ define amdgpu_kernel void @v_test_umin_ule_v3i32(ptr addrspace(1) %out, ptr addr
 ; GFX11-NEXT:    v_min_u32_e32 v0, v0, v3
 ; GFX11-NEXT:    global_store_b96 v6, v[0:2], s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: v_test_umin_ule_v3i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1250-NEXT:    v_lshlrev_b32_e32 v3, 4, v0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_load_b96 v[0:2], v3, s[2:3]
+; GFX1250-NEXT:    global_load_b96 v[4:6], v3, s[4:5]
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    v_min_u32_e32 v2, v2, v6
+; GFX1250-NEXT:    v_min_u32_e32 v1, v1, v5
+; GFX1250-NEXT:    v_min_u32_e32 v0, v0, v4
+; GFX1250-NEXT:    global_store_b96 v3, v[0:2], s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %a.gep = getelementptr inbounds <3 x i32>, ptr addrspace(1) %a.ptr, i32 %tid
   %b.gep = getelementptr inbounds <3 x i32>, ptr addrspace(1) %b.ptr, i32 %tid
@@ -2068,6 +2311,26 @@ define amdgpu_kernel void @v_test_umin_ule_v3i16(ptr addrspace(1) %out, ptr addr
 ; GFX11-NEXT:    global_store_b16 v4, v1, s[0:1] offset:4
 ; GFX11-NEXT:    global_store_b32 v4, v0, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: v_test_umin_ule_v3i16:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_and_b32_e32 v4, 0x3ff, v0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_load_b64 v[0:1], v4, s[2:3] scale_offset
+; GFX1250-NEXT:    global_load_b64 v[2:3], v4, s[4:5] scale_offset
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    v_lshlrev_b32_e32 v4, 3, v4
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    v_pk_min_u16 v1, v1, v3
+; GFX1250-NEXT:    v_pk_min_u16 v0, v0, v2
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_store_b16 v4, v1, s[0:1] offset:4
+; GFX1250-NEXT:    global_store_b32 v4, v0, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %a.gep = getelementptr inbounds <3 x i16>, ptr addrspace(1) %a.ptr, i32 %tid
   %b.gep = getelementptr inbounds <3 x i16>, ptr addrspace(1) %b.ptr, i32 %tid
@@ -2151,6 +2414,17 @@ define amdgpu_kernel void @s_test_umin_ule_i32(ptr addrspace(1) %out, i32 %a, i3
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_umin_ule_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_min_u32 s2, s2, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp ule i32 %a, %b
   %val = select i1 %cmp, i32 %a, i32 %b
   store i32 %val, ptr addrspace(1) %out, align 4
@@ -2274,6 +2548,21 @@ define amdgpu_kernel void @v_test_umin_ult_i32(ptr addrspace(1) %out, ptr addrsp
 ; GFX11-NEXT:    v_min_u32_e32 v1, v1, v2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: v_test_umin_ult_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_load_b32 v1, v0, s[2:3] scale_offset
+; GFX1250-NEXT:    global_load_b32 v2, v0, s[4:5] scale_offset
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    v_min_u32_e32 v1, v1, v2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1] scale_offset
+; GFX1250-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %a.gep = getelementptr inbounds i32, ptr addrspace(1) %a.ptr, i32 %tid
   %b.gep = getelementptr inbounds i32, ptr addrspace(1) %b.ptr, i32 %tid
@@ -2419,6 +2708,21 @@ define amdgpu_kernel void @v_test_umin_ult_i8(ptr addrspace(1) %out, ptr addrspa
 ; GFX11-FAKE16-NEXT:    v_min_u16 v1, v1, v2
 ; GFX11-FAKE16-NEXT:    global_store_b8 v0, v1, s[0:1]
 ; GFX11-FAKE16-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: v_test_umin_ult_i8:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_load_u8 v1, v0, s[2:3]
+; GFX1250-NEXT:    global_load_u8 v2, v0, s[4:5]
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    v_min_u16 v1, v1, v2
+; GFX1250-NEXT:    global_store_b8 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %a.gep = getelementptr inbounds i8, ptr addrspace(1) %a.ptr, i32 %tid
   %b.gep = getelementptr inbounds i8, ptr addrspace(1) %b.ptr, i32 %tid
@@ -2502,6 +2806,17 @@ define amdgpu_kernel void @s_test_umin_ult_i32(ptr addrspace(1) %out, i32 %a, i3
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_umin_ult_i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_min_u32 s2, s2, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp ult i32 %a, %b
   %val = select i1 %cmp, i32 %a, i32 %b
   store i32 %val, ptr addrspace(1) %out, align 4
@@ -2645,6 +2960,27 @@ define amdgpu_kernel void @v_test_umin_ult_i32_multi_use(ptr addrspace(1) %out0,
 ; GFX11-NEXT:    global_store_b32 v1, v2, s[0:1]
 ; GFX11-NEXT:    global_store_b8 v1, v0, s[2:3]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: v_test_umin_ult_i32_multi_use:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b256 s[0:7], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v1, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_load_b32 s4, s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b32 s5, s[6:7], 0x0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_cmp_lt_u32 s4, s5
+; GFX1250-NEXT:    s_cselect_b32 s6, -1, 0
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(SKIP_2) | instid1(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_cndmask_b32_e64 v0, 0, 1, s6
+; GFX1250-NEXT:    s_and_b32 s6, s6, exec_lo
+; GFX1250-NEXT:    s_cselect_b32 s4, s4, s5
+; GFX1250-NEXT:    v_mov_b32_e32 v2, s4
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_store_b32 v1, v2, s[0:1]
+; GFX1250-NEXT:    global_store_b8 v1, v0, s[2:3]
+; GFX1250-NEXT:    s_endpgm
   %a = load i32, ptr addrspace(1) %aptr, align 4
   %b = load i32, ptr addrspace(1) %bptr, align 4
   %cmp = icmp ult i32 %a, %b
@@ -2821,6 +3157,27 @@ define amdgpu_kernel void @v_test_umin_ult_i16_multi_use(ptr addrspace(1) %out0,
 ; GFX11-FAKE16-NEXT:    global_store_b16 v0, v1, s[0:1]
 ; GFX11-FAKE16-NEXT:    global_store_b8 v0, v2, s[2:3]
 ; GFX11-FAKE16-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: v_test_umin_ult_i16_multi_use:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b256 s[0:7], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_load_u16 v1, v0, s[6:7]
+; GFX1250-NEXT:    global_load_u16 v2, v0, s[4:5]
+; GFX1250-NEXT:    s_wait_loadcnt 0x1
+; GFX1250-NEXT:    v_and_b32_e32 v3, 0xffff, v1
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    v_and_b32_e32 v4, 0xffff, v2
+; GFX1250-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX1250-NEXT:    v_cmp_lt_u32_e32 vcc_lo, v4, v3
+; GFX1250-NEXT:    v_cndmask_b32_e32 v1, v1, v2, vcc_lo
+; GFX1250-NEXT:    v_cndmask_b32_e64 v2, 0, 1, vcc_lo
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_store_b16 v0, v1, s[0:1]
+; GFX1250-NEXT:    global_store_b8 v0, v2, s[2:3]
+; GFX1250-NEXT:    s_endpgm
   %a = load i16, ptr addrspace(1) %aptr, align 2
   %b = load i16, ptr addrspace(1) %bptr, align 2
   %cmp = icmp ult i16 %a, %b
@@ -2900,6 +3257,17 @@ define amdgpu_kernel void @s_test_umin_ult_v1i32(ptr addrspace(1) %out, <1 x i32
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_umin_ult_v1i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_min_u32 s2, s2, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp ult <1 x i32> %a, %b
   %val = select <1 x i1> %cmp, <1 x i32> %a, <1 x i32> %b
   store <1 x i32> %val, ptr addrspace(1) %out
@@ -3078,6 +3446,34 @@ define amdgpu_kernel void @s_test_umin_ult_v8i32(ptr addrspace(1) %out, <8 x i32
 ; GFX11-NEXT:    global_store_b128 v8, v[0:3], s[0:1] offset:16
 ; GFX11-NEXT:    global_store_b128 v8, v[4:7], s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_umin_ult_v8i32:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    s_load_b512 s[8:23], s[4:5], 0x20
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v8, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_min_u32 s4, s9, s17
+; GFX1250-NEXT:    s_min_u32 s5, s8, s16
+; GFX1250-NEXT:    s_min_u32 s6, s15, s23
+; GFX1250-NEXT:    s_min_u32 s7, s14, s22
+; GFX1250-NEXT:    s_min_u32 s8, s12, s20
+; GFX1250-NEXT:    s_min_u32 s9, s13, s21
+; GFX1250-NEXT:    s_min_u32 s2, s11, s19
+; GFX1250-NEXT:    s_min_u32 s3, s10, s18
+; GFX1250-NEXT:    v_mov_b32_e32 v0, s8
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s9
+; GFX1250-NEXT:    v_mov_b32_e32 v2, s7
+; GFX1250-NEXT:    v_mov_b32_e32 v3, s6
+; GFX1250-NEXT:    v_mov_b32_e32 v4, s5
+; GFX1250-NEXT:    v_mov_b32_e32 v5, s4
+; GFX1250-NEXT:    v_mov_b32_e32 v6, s3
+; GFX1250-NEXT:    v_mov_b32_e32 v7, s2
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_store_b128 v8, v[0:3], s[0:1] offset:16
+; GFX1250-NEXT:    global_store_b128 v8, v[4:7], s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp ult <8 x i32> %a, %b
   %val = select <8 x i1> %cmp, <8 x i32> %a, <8 x i32> %b
   store <8 x i32> %val, ptr addrspace(1) %out
@@ -3270,6 +3666,20 @@ define amdgpu_kernel void @s_test_umin_ult_v8i16(ptr addrspace(1) %out, <8 x i16
 ; GFX11-NEXT:    v_pk_min_u16 v0, s8, s12
 ; GFX11-NEXT:    global_store_b128 v4, v[0:3], s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_umin_ult_v8i16:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    s_load_b256 s[8:15], s[4:5], 0x10
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v4, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_pk_min_u16 v3, s11, s15
+; GFX1250-NEXT:    v_pk_min_u16 v2, s10, s14
+; GFX1250-NEXT:    v_pk_min_u16 v1, s9, s13
+; GFX1250-NEXT:    v_pk_min_u16 v0, s8, s12
+; GFX1250-NEXT:    global_store_b128 v4, v[0:3], s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp ult <8 x i16> %a, %b
   %val = select <8 x i1> %cmp, <8 x i16> %a, <8 x i16> %b
   store <8 x i16> %val, ptr addrspace(1) %out
@@ -3380,6 +3790,22 @@ define amdgpu_kernel void @simplify_demanded_bits_test_umin_ult_i16(ptr addrspac
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: simplify_demanded_bits_test_umin_ult_i16:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_clause 0x2
+; GFX1250-NEXT:    s_load_b32 s2, s[4:5], 0x28
+; GFX1250-NEXT:    s_load_b32 s3, s[4:5], 0x4c
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_and_b32 s2, s2, 0xffff
+; GFX1250-NEXT:    s_and_b32 s3, s3, 0xffff
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-NEXT:    s_min_u32 s2, s2, s3
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %a.ext = zext i16 %a to i32
   %b.ext = zext i16 %b to i32
   %cmp = icmp ult i32 %a.ext, %b.ext
@@ -3493,6 +3919,22 @@ define amdgpu_kernel void @simplify_demanded_bits_test_min_slt_i16(ptr addrspace
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: simplify_demanded_bits_test_min_slt_i16:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_clause 0x2
+; GFX1250-NEXT:    s_load_b32 s2, s[4:5], 0x28
+; GFX1250-NEXT:    s_load_b32 s3, s[4:5], 0x4c
+; GFX1250-NEXT:    s_load_b64 s[0:1], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_sext_i32_i16 s2, s2
+; GFX1250-NEXT:    s_sext_i32_i16 s3, s3
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-NEXT:    s_min_i32 s2, s2, s3
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %a.ext = sext i16 %a to i32
   %b.ext = sext i16 %b to i32
   %cmp = icmp slt i32 %a.ext, %b.ext
@@ -3609,6 +4051,19 @@ define amdgpu_kernel void @s_test_imin_sle_i16(ptr addrspace(1) %out, i16 %a, i1
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s2
 ; GFX11-NEXT:    global_store_b16 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: s_test_imin_sle_i16:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b96 s[0:2], s[4:5], 0x0
+; GFX1250-NEXT:    v_mov_b32_e32 v0, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_ashr_i32 s3, s2, 16
+; GFX1250-NEXT:    s_sext_i32_i16 s2, s2
+; GFX1250-NEXT:    s_delay_alu instid0(SALU_CYCLE_1) | instskip(NEXT) | instid1(SALU_CYCLE_1)
+; GFX1250-NEXT:    s_min_i32 s2, s2, s3
+; GFX1250-NEXT:    v_mov_b32_e32 v1, s2
+; GFX1250-NEXT:    global_store_b16 v0, v1, s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %cmp = icmp sle i16 %a, %b
   %val = select i1 %cmp, i16 %a, i16 %b
   store i16 %val, ptr addrspace(1) %out
@@ -3724,6 +4179,17 @@ define amdgpu_kernel void @test_umin_ult_i64(ptr addrspace(1) %out, i64 %a, i64
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: test_umin_ult_i64:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_min_u64 v[0:1], s[2:3], s[4:5]
+; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %tmp = icmp ult i64 %a, %b
   %val = select i1 %tmp, i64 %a, i64 %b
   store i64 %val, ptr addrspace(1) %out, align 8
@@ -3837,6 +4303,17 @@ define amdgpu_kernel void @test_umin_ule_i64(ptr addrspace(1) %out, i64 %a, i64
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: test_umin_ule_i64:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_min_u64 v[0:1], s[2:3], s[4:5]
+; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %tmp = icmp ule i64 %a, %b
   %val = select i1 %tmp, i64 %a, i64 %b
   store i64 %val, ptr addrspace(1) %out, align 8
@@ -3950,6 +4427,17 @@ define amdgpu_kernel void @test_imin_slt_i64(ptr addrspace(1) %out, i64 %a, i64
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: test_imin_slt_i64:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_min_i64 v[0:1], s[2:3], s[4:5]
+; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %tmp = icmp slt i64 %a, %b
   %val = select i1 %tmp, i64 %a, i64 %b
   store i64 %val, ptr addrspace(1) %out, align 8
@@ -4063,6 +4551,17 @@ define amdgpu_kernel void @test_imin_sle_i64(ptr addrspace(1) %out, i64 %a, i64
 ; GFX11-NEXT:    v_mov_b32_e32 v1, s3
 ; GFX11-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: test_imin_sle_i64:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_mov_b32_e32 v2, 0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    v_min_i64 v[0:1], s[2:3], s[4:5]
+; GFX1250-NEXT:    global_store_b64 v2, v[0:1], s[0:1]
+; GFX1250-NEXT:    s_endpgm
   %tmp = icmp sle i64 %a, %b
   %val = select i1 %tmp, i64 %a, i64 %b
   store i64 %val, ptr addrspace(1) %out, align 8
@@ -4214,6 +4713,21 @@ define amdgpu_kernel void @v_test_imin_sle_v2i16(ptr addrspace(1) %out, ptr addr
 ; GFX11-NEXT:    v_pk_min_i16 v1, v1, v2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: v_test_imin_sle_v2i16:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_load_b32 v1, v0, s[2:3] scale_offset
+; GFX1250-NEXT:    global_load_b32 v2, v0, s[4:5] scale_offset
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    v_pk_min_i16 v1, v1, v2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1] scale_offset
+; GFX1250-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %a.gep = getelementptr inbounds <2 x i16>, ptr addrspace(1) %a.ptr, i32 %tid
   %b.gep = getelementptr inbounds <2 x i16>, ptr addrspace(1) %b.ptr, i32 %tid
@@ -4369,6 +4883,21 @@ define amdgpu_kernel void @v_test_imin_ule_v2i16(ptr addrspace(1) %out, ptr addr
 ; GFX11-NEXT:    v_pk_min_u16 v1, v1, v2
 ; GFX11-NEXT:    global_store_b32 v0, v1, s[0:1]
 ; GFX11-NEXT:    s_endpgm
+;
+; GFX1250-LABEL: v_test_imin_ule_v2i16:
+; GFX1250:       ; %bb.0:
+; GFX1250-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
+; GFX1250-NEXT:    s_wait_xcnt 0x0
+; GFX1250-NEXT:    s_load_b64 s[4:5], s[4:5], 0x10
+; GFX1250-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
+; GFX1250-NEXT:    s_wait_kmcnt 0x0
+; GFX1250-NEXT:    s_clause 0x1
+; GFX1250-NEXT:    global_load_b32 v1, v0, s[2:3] scale_offset
+; GFX1250-NEXT:    global_load_b32 v2, v0, s[4:5] scale_offset
+; GFX1250-NEXT:    s_wait_loadcnt 0x0
+; GFX1250-NEXT:    v_pk_min_u16 v1, v1, v2
+; GFX1250-NEXT:    global_store_b32 v0, v1, s[0:1] scale_offset
+; GFX1250-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %a.gep = getelementptr inbounds <2 x i16>, ptr addrspace(1) %a.ptr, i32 %tid
   %b.gep = getelementptr inbounds <2 x i16>, ptr addrspace(1) %b.ptr, i32 %tid
@@ -4385,3 +4914,5 @@ declare i32 @llvm.amdgcn.workitem.id.x() #1
 
 attributes #0 = { nounwind }
 attributes #1 = { nounwind readnone }
+;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
+; GFX1250-FAKE16: {{.*}}
diff --git a/llvm/test/MC/AMDGPU/gfx1250_asm_vop3-fake16.s b/llvm/test/MC/AMDGPU/gfx1250_asm_vop3-fake16.s
index 8a00f8851d192..04c55eedc18d4 100644
--- a/llvm/test/MC/AMDGPU/gfx1250_asm_vop3-fake16.s
+++ b/llvm/test/MC/AMDGPU/gfx1250_asm_vop3-fake16.s
@@ -133,6 +133,54 @@ v_mad_u32 v2, s4, 4, v2
 v_mad_u32 v2, v4, v7, 12345
 // GFX1250: v_mad_u32 v2, v4, v7, 0x3039            ; encoding: [0x02,0x00,0x35,0xd6,0x04,0x0f,0xfe,0x03,0x39,0x30,0x00,0x00]
 
+v_max_i64 v[2:3], s[4:5], v[6:7]
+// GFX1250: v_max_i64 v[2:3], s[4:5], v[6:7]        ; encoding: [0x02,0x00,0x1b,0xd7,0x04,0x0c,0x02,0x00]
+
+v_max_i64 v[2:3], v[4:5], 1
+// GFX1250: v_max_i64 v[2:3], v[4:5], 1             ; encoding: [0x02,0x00,0x1b,0xd7,0x04,0x03,0x01,0x00]
+
+v_max_i64 v[2:3], v[4:5], s[2:3]
+// GFX1250: v_max_i64 v[2:3], v[4:5], s[2:3]        ; encoding: [0x02,0x00,0x1b,0xd7,0x04,0x05,0x00,0x00]
+
+v_max_i64 v[2:3], v[4:5], 12345
+// GFX1250: v_max_i64 v[2:3], v[4:5], 0x3039        ; encoding: [0x02,0x00,0x1b,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00]
+
+v_max_u64 v[2:3], s[4:5], v[6:7]
+// GFX1250: v_max_u64 v[2:3], s[4:5], v[6:7]        ; encoding: [0x02,0x00,0x19,0xd7,0x04,0x0c,0x02,0x00]
+
+v_max_u64 v[2:3], v[4:5], 1
+// GFX1250: v_max_u64 v[2:3], v[4:5], 1             ; encoding: [0x02,0x00,0x19,0xd7,0x04,0x03,0x01,0x00]
+
+v_max_u64 v[2:3], v[4:5], s[2:3]
+// GFX1250: v_max_u64 v[2:3], v[4:5], s[2:3]        ; encoding: [0x02,0x00,0x19,0xd7,0x04,0x05,0x00,0x00]
+
+v_max_u64 v[2:3], v[4:5], 12345
+// GFX1250: v_max_u64 v[2:3], v[4:5], 0x3039        ; encoding: [0x02,0x00,0x19,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00]
+
+v_min_i64 v[2:3], s[4:5], v[6:7]
+// GFX1250: v_min_i64 v[2:3], s[4:5], v[6:7]        ; encoding: [0x02,0x00,0x1a,0xd7,0x04,0x0c,0x02,0x00]
+
+v_min_i64 v[2:3], v[4:5], 1
+// GFX1250: v_min_i64 v[2:3], v[4:5], 1             ; encoding: [0x02,0x00,0x1a,0xd7,0x04,0x03,0x01,0x00]
+
+v_min_i64 v[2:3], v[4:5], s[2:3]
+// GFX1250: v_min_i64 v[2:3], v[4:5], s[2:3]        ; encoding: [0x02,0x00,0x1a,0xd7,0x04,0x05,0x00,0x00]
+
+v_min_i64 v[2:3], v[4:5], 12345
+// GFX1250: v_min_i64 v[2:3], v[4:5], 0x3039        ; encoding: [0x02,0x00,0x1a,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00]
+
+v_min_u64 v[2:3], s[4:5], v[6:7]
+// GFX1250: v_min_u64 v[2:3], s[4:5], v[6:7]        ; encoding: [0x02,0x00,0x18,0xd7,0x04,0x0c,0x02,0x00]
+
+v_min_u64 v[2:3], v[4:5], 1
+// GFX1250: v_min_u64 v[2:3], v[4:5], 1             ; encoding: [0x02,0x00,0x18,0xd7,0x04,0x03,0x01,0x00]
+
+v_min_u64 v[2:3], v[4:5], s[2:3]
+// GFX1250: v_min_u64 v[2:3], v[4:5], s[2:3]        ; encoding: [0x02,0x00,0x18,0xd7,0x04,0x05,0x00,0x00]
+
+v_min_u64 v[2:3], v[4:5], 12345
+// GFX1250: v_min_u64 v[2:3], v[4:5], 0x3039        ; encoding: [0x02,0x00,0x18,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00]
+
 v_mad_nc_u64_u32 v[2:3], s4, v7, v[8:9]
 // GFX1250: v_mad_nc_u64_u32 v[2:3], s4, v7, v[8:9] ; encoding: [0x02,0x00,0xfa,0xd6,0x04,0x0e,0x22,0x04]
 
diff --git a/llvm/test/MC/AMDGPU/gfx1250_asm_vop3.s b/llvm/test/MC/AMDGPU/gfx1250_asm_vop3.s
index d9561fadb20d0..ebfeb3f74d752 100644
--- a/llvm/test/MC/AMDGPU/gfx1250_asm_vop3.s
+++ b/llvm/test/MC/AMDGPU/gfx1250_asm_vop3.s
@@ -133,6 +133,54 @@ v_mad_u32 v2, s4, 4, v2
 v_mad_u32 v2, v4, v7, 12345
 // GFX1250: v_mad_u32 v2, v4, v7, 0x3039            ; encoding: [0x02,0x00,0x35,0xd6,0x04,0x0f,0xfe,0x03,0x39,0x30,0x00,0x00]
 
+v_max_i64 v[2:3], s[4:5], v[6:7]
+// GFX1250: v_max_i64 v[2:3], s[4:5], v[6:7]        ; encoding: [0x02,0x00,0x1b,0xd7,0x04,0x0c,0x02,0x00]
+
+v_max_i64 v[2:3], v[4:5], 1
+// GFX1250: v_max_i64 v[2:3], v[4:5], 1             ; encoding: [0x02,0x00,0x1b,0xd7,0x04,0x03,0x01,0x00]
+
+v_max_i64 v[2:3], v[4:5], s[2:3]
+// GFX1250: v_max_i64 v[2:3], v[4:5], s[2:3]        ; encoding: [0x02,0x00,0x1b,0xd7,0x04,0x05,0x00,0x00]
+
+v_max_i64 v[2:3], v[4:5], 12345
+// GFX1250: v_max_i64 v[2:3], v[4:5], 0x3039        ; encoding: [0x02,0x00,0x1b,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00]
+
+v_max_u64 v[2:3], s[4:5], v[6:7]
+// GFX1250: v_max_u64 v[2:3], s[4:5], v[6:7]        ; encoding: [0x02,0x00,0x19,0xd7,0x04,0x0c,0x02,0x00]
+
+v_max_u64 v[2:3], v[4:5], 1
+// GFX1250: v_max_u64 v[2:3], v[4:5], 1             ; encoding: [0x02,0x00,0x19,0xd7,0x04,0x03,0x01,0x00]
+
+v_max_u64 v[2:3], v[4:5], s[2:3]
+// GFX1250: v_max_u64 v[2:3], v[4:5], s[2:3]        ; encoding: [0x02,0x00,0x19,0xd7,0x04,0x05,0x00,0x00]
+
+v_max_u64 v[2:3], v[4:5], 12345
+// GFX1250: v_max_u64 v[2:3], v[4:5], 0x3039        ; encoding: [0x02,0x00,0x19,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00]
+
+v_min_i64 v[2:3], s[4:5], v[6:7]
+// GFX1250: v_min_i64 v[2:3], s[4:5], v[6:7]        ; encoding: [0x02,0x00,0x1a,0xd7,0x04,0x0c,0x02,0x00]
+
+v_min_i64 v[2:3], v[4:5], 1
+// GFX1250: v_min_i64 v[2:3], v[4:5], 1             ; encoding: [0x02,0x00,0x1a,0xd7,0x04,0x03,0x01,0x00]
+
+v_min_i64 v[2:3], v[4:5], s[2:3]
+// GFX1250: v_min_i64 v[2:3], v[4:5], s[2:3]        ; encoding: [0x02,0x00,0x1a,0xd7,0x04,0x05,0x00,0x00]
+
+v_min_i64 v[2:3], v[4:5], 12345
+// GFX1250: v_min_i64 v[2:3], v[4:5], 0x3039        ; encoding: [0x02,0x00,0x1a,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00]
+
+v_min_u64 v[2:3], s[4:5], v[6:7]
+// GFX1250: v_min_u64 v[2:3], s[4:5], v[6:7]        ; encoding: [0x02,0x00,0x18,0xd7,0x04,0x0c,0x02,0x00]
+
+v_min_u64 v[2:3], v[4:5], 1
+// GFX1250: v_min_u64 v[2:3], v[4:5], 1             ; encoding: [0x02,0x00,0x18,0xd7,0x04,0x03,0x01,0x00]
+
+v_min_u64 v[2:3], v[4:5], s[2:3]
+// GFX1250: v_min_u64 v[2:3], v[4:5], s[2:3]        ; encoding: [0x02,0x00,0x18,0xd7,0x04,0x05,0x00,0x00]
+
+v_min_u64 v[2:3], v[4:5], 12345
+// GFX1250: v_min_u64 v[2:3], v[4:5], 0x3039        ; encoding: [0x02,0x00,0x18,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00]
+
 v_mad_nc_u64_u32 v[2:3], s4, v7, v[8:9]
 // GFX1250: v_mad_nc_u64_u32 v[2:3], s4, v7, v[8:9] ; encoding: [0x02,0x00,0xfa,0xd6,0x04,0x0e,0x22,0x04]
 
diff --git a/llvm/test/MC/AMDGPU/gfx1250_asm_vop3_err.s b/llvm/test/MC/AMDGPU/gfx1250_asm_vop3_err.s
index b18249d7ee0a8..7e29d04f06dad 100644
--- a/llvm/test/MC/AMDGPU/gfx1250_asm_vop3_err.s
+++ b/llvm/test/MC/AMDGPU/gfx1250_asm_vop3_err.s
@@ -10,6 +10,26 @@ v_mad_u32 v2, v4, v7, v8 dpp8:[7,6,5,4,3,2,1,0]
 // GFX125X-ERR-NEXT:{{^}}v_mad_u32 v2, v4, v7, v8 dpp8:[7,6,5,4,3,2,1,0]
 // GFX125X-ERR-NEXT:{{^}}                         ^
 
+v_max_i64 v[2:3], v[4:5], v[6:7] dpp8:[7,6,5,4,3,2,1,0]
+// GFX125X-ERR: :[[@LINE-1]]:{{[0-9]+}}: error: not a valid operand.
+// GFX125X-ERR-NEXT:{{^}}v_max_i64 v[2:3], v[4:5], v[6:7] dpp8:[7,6,5,4,3,2,1,0]
+// GFX125X-ERR-NEXT:{{^}}                                 ^
+
+v_max_u64 v[2:3], v[4:5], v[6:7] dpp8:[7,6,5,4,3,2,1,0]
+// GFX125X-ERR: :[[@LINE-1]]:{{[0-9]+}}: error: not a valid operand.
+// GFX125X-ERR-NEXT:{{^}}v_max_u64 v[2:3], v[4:5], v[6:7] dpp8:[7,6,5,4,3,2,1,0]
+// GFX125X-ERR-NEXT:{{^}}                                 ^
+
+v_min_i64 v[2:3], v[4:5], v[6:7] dpp8:[7,6,5,4,3,2,1,0]
+// GFX125X-ERR: :[[@LINE-1]]:{{[0-9]+}}: error: not a valid operand.
+// GFX125X-ERR-NEXT:{{^}}v_min_i64 v[2:3], v[4:5], v[6:7] dpp8:[7,6,5,4,3,2,1,0]
+// GFX125X-ERR-NEXT:{{^}}                                 ^
+
+v_min_u64 v[2:3], v[4:5], v[6:7] dpp8:[7,6,5,4,3,2,1,0]
+// GFX125X-ERR: :[[@LINE-1]]:{{[0-9]+}}: error: not a valid operand.
+// GFX125X-ERR-NEXT:{{^}}v_min_u64 v[2:3], v[4:5], v[6:7] dpp8:[7,6,5,4,3,2,1,0]
+// GFX125X-ERR-NEXT:{{^}}                                 ^
+
 v_mad_nc_u64_u32 v[4:5], v2, v5, v[6:7] dpp8:[7,6,5,4,3,2,1,0]
 // GFX125X-ERR: :[[@LINE-1]]:{{[0-9]+}}: error: not a valid operand.
 // GFX125X-ERR-NEXT:{{^}}v_mad_nc_u64_u32 v[4:5], v2, v5, v[6:7] dpp8:[7,6,5,4,3,2,1,0]
@@ -31,6 +51,30 @@ v_mad_u32 v2, v4, v7, v8 quad_perm:[3,2,1,0]
 // GFX125X-ERR-NEXT:{{^}}v_mad_u32 v2, v4, v7, v8 quad_perm:[3,2,1,0]
 // GFX125X-ERR-NEXT:{{^}}                         ^
 
+v_max_i64 v[2:3], v[4:5], v[6:7] quad_perm:[3,2,1,0]
+// GFX1250-ERR: :[[@LINE-1]]:{{[0-9]+}}: error: not a valid operand.
+// GFX1251-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: DP ALU dpp only supports row_share
+// GFX125X-ERR-NEXT:{{^}}v_max_i64 v[2:3], v[4:5], v[6:7] quad_perm:[3,2,1,0]
+// GFX125X-ERR-NEXT:{{^}}                                 ^
+
+v_max_u64 v[2:3], v[4:5], v[6:7] quad_perm:[3,2,1,0]
+// GFX1250-ERR: :[[@LINE-1]]:{{[0-9]+}}: error: not a valid operand.
+// GFX1251-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: DP ALU dpp only supports row_share
+// GFX125X-ERR-NEXT:{{^}}v_max_u64 v[2:3], v[4:5], v[6:7] quad_perm:[3,2,1,0]
+// GFX125X-ERR-NEXT:{{^}}                                 ^
+
+v_min_i64 v[2:3], v[4:5], v[6:7] quad_perm:[3,2,1,0]
+// GFX1250-ERR: :[[@LINE-1]]:{{[0-9]+}}: error: not a valid operand.
+// GFX1251-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: DP ALU dpp only supports row_share
+// GFX125X-ERR-NEXT:{{^}}v_min_i64 v[2:3], v[4:5], v[6:7] quad_perm:[3,2,1,0]
+// GFX125X-ERR-NEXT:{{^}}                                 ^
+
+v_min_u64 v[2:3], v[4:5], v[6:7] quad_perm:[3,2,1,0]
+// GFX1250-ERR: :[[@LINE-1]]:{{[0-9]+}}: error: not a valid operand.
+// GFX1251-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: DP ALU dpp only supports row_share
+// GFX125X-ERR-NEXT:{{^}}v_min_u64 v[2:3], v[4:5], v[6:7] quad_perm:[3,2,1,0]
+// GFX125X-ERR-NEXT:{{^}}                                 ^
+
 v_mad_nc_u64_u32 v[4:5], v2, v5, v[6:7] quad_perm:[3,2,1,0]
 // GFX1250-ERR: :[[@LINE-1]]:{{[0-9]+}}: error: not a valid operand.
 // GFX1251-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: DP ALU dpp only supports row_share
diff --git a/llvm/test/MC/Disassembler/AMDGPU/gfx1250_dasm_vop3.txt b/llvm/test/MC/Disassembler/AMDGPU/gfx1250_dasm_vop3.txt
index 335e2ed34ed6e..9fd7edd3ffa82 100644
--- a/llvm/test/MC/Disassembler/AMDGPU/gfx1250_dasm_vop3.txt
+++ b/llvm/test/MC/Disassembler/AMDGPU/gfx1250_dasm_vop3.txt
@@ -152,6 +152,54 @@
 0x02,0x00,0x35,0xd6,0x04,0x0f,0xfe,0x03,0x39,0x30,0x00,0x00
 # GFX1250: v_mad_u32 v2, v4, v7, 0x3039            ; encoding: [0x02,0x00,0x35,0xd6,0x04,0x0f,0xfe,0x03,0x39,0x30,0x00,0x00]
 
+0x02,0x00,0x1b,0xd7,0x04,0x0c,0x02,0x00
+# GFX1250: v_max_i64 v[2:3], s[4:5], v[6:7]        ; encoding: [0x02,0x00,0x1b,0xd7,0x04,0x0c,0x02,0x00]
+
+0x02,0x00,0x1b,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00
+# GFX1250: v_max_i64 v[2:3], v[4:5], 0x3039        ; encoding: [0x02,0x00,0x1b,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00]
+
+0x02,0x00,0x1b,0xd7,0x04,0x03,0x01,0x00
+# GFX1250: v_max_i64 v[2:3], v[4:5], 1             ; encoding: [0x02,0x00,0x1b,0xd7,0x04,0x03,0x01,0x00]
+
+0x02,0x00,0x1b,0xd7,0x04,0x05,0x00,0x00
+# GFX1250: v_max_i64 v[2:3], v[4:5], s[2:3]        ; encoding: [0x02,0x00,0x1b,0xd7,0x04,0x05,0x00,0x00]
+
+0x02,0x00,0x19,0xd7,0x04,0x0c,0x02,0x00
+# GFX1250: v_max_u64 v[2:3], s[4:5], v[6:7]        ; encoding: [0x02,0x00,0x19,0xd7,0x04,0x0c,0x02,0x00]
+
+0x02,0x00,0x19,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00
+# GFX1250: v_max_u64 v[2:3], v[4:5], 0x3039        ; encoding: [0x02,0x00,0x19,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00]
+
+0x02,0x00,0x19,0xd7,0x04,0x03,0x01,0x00
+# GFX1250: v_max_u64 v[2:3], v[4:5], 1             ; encoding: [0x02,0x00,0x19,0xd7,0x04,0x03,0x01,0x00]
+
+0x02,0x00,0x19,0xd7,0x04,0x05,0x00,0x00
+# GFX1250: v_max_u64 v[2:3], v[4:5], s[2:3]        ; encoding: [0x02,0x00,0x19,0xd7,0x04,0x05,0x00,0x00]
+
+0x02,0x00,0x1a,0xd7,0x04,0x0c,0x02,0x00
+# GFX1250: v_min_i64 v[2:3], s[4:5], v[6:7]        ; encoding: [0x02,0x00,0x1a,0xd7,0x04,0x0c,0x02,0x00]
+
+0x02,0x00,0x1a,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00
+# GFX1250: v_min_i64 v[2:3], v[4:5], 0x3039        ; encoding: [0x02,0x00,0x1a,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00]
+
+0x02,0x00,0x1a,0xd7,0x04,0x03,0x01,0x00
+# GFX1250: v_min_i64 v[2:3], v[4:5], 1             ; encoding: [0x02,0x00,0x1a,0xd7,0x04,0x03,0x01,0x00]
+
+0x02,0x00,0x1a,0xd7,0x04,0x05,0x00,0x00
+# GFX1250: v_min_i64 v[2:3], v[4:5], s[2:3]        ; encoding: [0x02,0x00,0x1a,0xd7,0x04,0x05,0x00,0x00]
+
+0x02,0x00,0x18,0xd7,0x04,0x0c,0x02,0x00
+# GFX1250: v_min_u64 v[2:3], s[4:5], v[6:7]        ; encoding: [0x02,0x00,0x18,0xd7,0x04,0x0c,0x02,0x00]
+
+0x02,0x00,0x18,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00
+# GFX1250: v_min_u64 v[2:3], v[4:5], 0x3039        ; encoding: [0x02,0x00,0x18,0xd7,0x04,0xff,0x01,0x00,0x39,0x30,0x00,0x00]
+
+0x02,0x00,0x18,0xd7,0x04,0x03,0x01,0x00
+# GFX1250: v_min_u64 v[2:3], v[4:5], 1             ; encoding: [0x02,0x00,0x18,0xd7,0x04,0x03,0x01,0x00]
+
+0x02,0x00,0x18,0xd7,0x04,0x05,0x00,0x00
+# GFX1250: v_min_u64 v[2:3], v[4:5], s[2:3]        ; encoding: [0x02,0x00,0x18,0xd7,0x04,0x05,0x00,0x00]
+
 0x02,0x00,0xfa,0xd6,0x04,0x08,0x09,0x04
 # GFX1250: v_mad_nc_u64_u32 v[2:3], s4, 4, v[2:3]  ; encoding: [0x02,0x00,0xfa,0xd6,0x04,0x08,0x09,0x04]