AMDGPU: Add baseline test for missing pattern

The optimization to turn an add into a sub isn't triggering when the
pattern to use the zeroed high bits is used.

diff --git a/llvm/test/CodeGen/AMDGPU/shrink-add-sub-constant.ll b/llvm/test/CodeGen/AMDGPU/shrink-add-sub-constant.ll
index 163f919..c5a1913 100644 (file)
--- a/llvm/test/CodeGen/AMDGPU/shrink-add-sub-constant.ll
+++ b/llvm/test/CodeGen/AMDGPU/shrink-add-sub-constant.ll
@@ -2,6 +2,7 @@
 ; RUN: llc < %s -march=amdgcn -verify-machineinstrs | FileCheck %s -check-prefixes=GCN,SI
 ; RUN: llc < %s -march=amdgcn -mcpu=tonga -verify-machineinstrs | FileCheck %s -check-prefixes=GCN,VI
 ; RUN: llc < %s -march=amdgcn -mcpu=gfx900 -verify-machineinstrs | FileCheck %s -check-prefixes=GCN,GFX9
+; RUN: llc < %s -march=amdgcn -mcpu=gfx1010 -verify-machineinstrs | FileCheck %s -check-prefixes=GCN,GFX10
 
 ; Test that add/sub with a constant is swapped to sub/add with negated
 ; constant to minimize code size.
@@ -56,6 +57,22 @@ define amdgpu_kernel void @v_test_i32_x_sub_64(i32 addrspace(1)* %out, i32 addrs
 ; GFX9-NEXT:    v_subrev_u32_e32 v2, 64, v3
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_i32_x_sub_64:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_subrev_nc_u32_e32 v2, 64, v3
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds i32, i32 addrspace(1)* %in, i64 %tid.ext
@@ -128,6 +145,26 @@ define amdgpu_kernel void @v_test_i32_x_sub_64_multi_use(i32 addrspace(1)* %out,
 ; GFX9-NEXT:    global_store_dword v[2:3], v1, off
 ; GFX9-NEXT:    global_store_dword v[2:3], v0, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_i32_x_sub_64_multi_use:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    global_load_dword v4, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(1)
+; GFX10-NEXT:    v_subrev_nc_u32_e32 v2, 64, v3
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_subrev_nc_u32_e32 v3, 64, v4
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    global_store_dword v[0:1], v3, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds i32, i32 addrspace(1)* %in, i64 %tid.ext
@@ -191,6 +228,22 @@ define amdgpu_kernel void @v_test_i32_64_sub_x(i32 addrspace(1)* %out, i32 addrs
 ; GFX9-NEXT:    v_sub_u32_e32 v2, 64, v3
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_i32_64_sub_x:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_sub_nc_u32_e32 v2, 64, v3
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds i32, i32 addrspace(1)* %in, i64 %tid.ext
@@ -251,6 +304,22 @@ define amdgpu_kernel void @v_test_i32_x_sub_65(i32 addrspace(1)* %out, i32 addrs
 ; GFX9-NEXT:    v_add_u32_e32 v2, 0xffffffbf, v3
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_i32_x_sub_65:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_add_nc_u32_e32 v2, 0xffffffbf, v3
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds i32, i32 addrspace(1)* %in, i64 %tid.ext
@@ -311,6 +380,22 @@ define amdgpu_kernel void @v_test_i32_65_sub_x(i32 addrspace(1)* %out, i32 addrs
 ; GFX9-NEXT:    v_sub_u32_e32 v2, 0x41, v3
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_i32_65_sub_x:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_sub_nc_u32_e32 v2, 0x41, v3
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds i32, i32 addrspace(1)* %in, i64 %tid.ext
@@ -371,6 +456,22 @@ define amdgpu_kernel void @v_test_i32_x_sub_neg16(i32 addrspace(1)* %out, i32 ad
 ; GFX9-NEXT:    v_add_u32_e32 v2, 16, v3
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_i32_x_sub_neg16:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_add_nc_u32_e32 v2, 16, v3
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds i32, i32 addrspace(1)* %in, i64 %tid.ext
@@ -431,6 +532,22 @@ define amdgpu_kernel void @v_test_i32_neg16_sub_x(i32 addrspace(1)* %out, i32 ad
 ; GFX9-NEXT:    v_sub_u32_e32 v2, -16, v3
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_i32_neg16_sub_x:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_sub_nc_u32_e32 v2, -16, v3
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds i32, i32 addrspace(1)* %in, i64 %tid.ext
@@ -491,6 +608,22 @@ define amdgpu_kernel void @v_test_i32_x_sub_neg17(i32 addrspace(1)* %out, i32 ad
 ; GFX9-NEXT:    v_add_u32_e32 v2, 17, v3
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_i32_x_sub_neg17:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_add_nc_u32_e32 v2, 17, v3
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds i32, i32 addrspace(1)* %in, i64 %tid.ext
@@ -551,6 +684,22 @@ define amdgpu_kernel void @v_test_i32_neg17_sub_x(i32 addrspace(1)* %out, i32 ad
 ; GFX9-NEXT:    v_sub_u32_e32 v2, 0xffffffef, v3
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_i32_neg17_sub_x:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_sub_nc_u32_e32 v2, 0xffffffef, v3
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds i32, i32 addrspace(1)* %in, i64 %tid.ext
@@ -591,6 +740,17 @@ define amdgpu_kernel void @s_test_i32_x_sub_64(i32 %x) #0 {
 ; GFX9-NEXT:    ; use s0
 ; GFX9-NEXT:    ;;#ASMEND
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: s_test_i32_x_sub_64:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dword s0, s[0:1], 0x24
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    s_sub_i32 s0, s0, 64
+; GFX10-NEXT:    ;;#ASMSTART
+; GFX10-NEXT:    ; use s0
+; GFX10-NEXT:    ;;#ASMEND
+; GFX10-NEXT:    s_endpgm
   %result = sub i32 %x, 64
   call void asm sideeffect "; use $0", "s"(i32 %result)
   ret void
@@ -646,6 +806,22 @@ define amdgpu_kernel void @v_test_i16_x_sub_64(i16 addrspace(1)* %out, i16 addrs
 ; GFX9-NEXT:    v_subrev_u16_e32 v2, 64, v3
 ; GFX9-NEXT:    global_store_short v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_i16_x_sub_64:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 1, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_ushort v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_sub_nc_u16_e64 v2, v3, 64
+; GFX10-NEXT:    global_store_short v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds i16, i16 addrspace(1)* %in, i64 %tid.ext
@@ -656,6 +832,89 @@ define amdgpu_kernel void @v_test_i16_x_sub_64(i16 addrspace(1)* %out, i16 addrs
   ret void
 }
 
+define amdgpu_kernel void @v_test_i16_x_sub_64_zext_to_i32(i32 addrspace(1)* %out, i16 addrspace(1)* %in) #0 {
+; SI-LABEL: v_test_i16_x_sub_64_zext_to_i32:
+; SI:       ; %bb.0:
+; SI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x9
+; SI-NEXT:    s_mov_b32 s7, 0xf000
+; SI-NEXT:    s_mov_b32 s6, 0
+; SI-NEXT:    v_lshlrev_b32_e32 v1, 1, v0
+; SI-NEXT:    v_mov_b32_e32 v2, 0
+; SI-NEXT:    s_waitcnt lgkmcnt(0)
+; SI-NEXT:    s_mov_b64 s[4:5], s[2:3]
+; SI-NEXT:    buffer_load_ushort v3, v[1:2], s[4:7], 0 addr64
+; SI-NEXT:    s_mov_b64 s[2:3], s[6:7]
+; SI-NEXT:    v_lshlrev_b32_e32 v1, 2, v0
+; SI-NEXT:    s_waitcnt vmcnt(0)
+; SI-NEXT:    v_subrev_i32_e32 v0, vcc, 64, v3
+; SI-NEXT:    v_and_b32_e32 v0, 0xffff, v0
+; SI-NEXT:    buffer_store_dword v0, v[1:2], s[0:3], 0 addr64
+; SI-NEXT:    s_endpgm
+;
+; VI-LABEL: v_test_i16_x_sub_64_zext_to_i32:
+; VI:       ; %bb.0:
+; VI-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; VI-NEXT:    v_lshlrev_b32_e32 v1, 1, v0
+; VI-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; VI-NEXT:    s_waitcnt lgkmcnt(0)
+; VI-NEXT:    v_mov_b32_e32 v3, s3
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s2, v1
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v3, vcc
+; VI-NEXT:    flat_load_ushort v3, v[0:1]
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v2
+; VI-NEXT:    v_mov_b32_e32 v1, s1
+; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
+; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
+; VI-NEXT:    v_add_u16_e32 v2, 0xffffffc0, v3
+; VI-NEXT:    flat_store_dword v[0:1], v2
+; VI-NEXT:    s_endpgm
+;
+; GFX9-LABEL: v_test_i16_x_sub_64_zext_to_i32:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX9-NEXT:    v_lshlrev_b32_e32 v1, 1, v0
+; GFX9-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX9-NEXT:    v_mov_b32_e32 v3, s3
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s2, v1
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v3, vcc
+; GFX9-NEXT:    global_load_ushort v3, v[0:1], off
+; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s0, v2
+; GFX9-NEXT:    v_mov_b32_e32 v1, s1
+; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    v_add_u16_e32 v2, 0xffffffc0, v3
+; GFX9-NEXT:    global_store_dword v[0:1], v2, off
+; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_i16_x_sub_64_zext_to_i32:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v1, 1, v0
+; GFX10-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v1, s2, s2, v1
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v0
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v2, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_ushort v1, v[1:2], off
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_sub_nc_u16_e64 v2, v1, 64
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    v_and_b32_e32 v2, 0xffff, v2
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
+  %tid = call i32 @llvm.amdgcn.workitem.id.x()
+  %tid.ext = sext i32 %tid to i64
+  %gep = getelementptr inbounds i16, i16 addrspace(1)* %in, i64 %tid.ext
+  %gep.out = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 %tid.ext
+  %x = load i16, i16 addrspace(1)* %gep
+  %result = sub i16 %x, 64
+  %zext = zext i16 %result to i32
+  store i32 %zext, i32 addrspace(1)* %gep.out
+  ret void
+}
+
 define amdgpu_kernel void @v_test_i16_x_sub_64_multi_use(i16 addrspace(1)* %out, i16 addrspace(1)* %in) #0 {
 ; SI-LABEL: v_test_i16_x_sub_64_multi_use:
 ; SI:       ; %bb.0:
@@ -718,6 +977,26 @@ define amdgpu_kernel void @v_test_i16_x_sub_64_multi_use(i16 addrspace(1)* %out,
 ; GFX9-NEXT:    global_store_short v[2:3], v1, off
 ; GFX9-NEXT:    global_store_short v[2:3], v0, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_i16_x_sub_64_multi_use:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 1, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_ushort v3, v[0:1], off
+; GFX10-NEXT:    global_load_ushort v4, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(1)
+; GFX10-NEXT:    v_sub_nc_u16_e64 v2, v3, 64
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_sub_nc_u16_e64 v3, v4, 64
+; GFX10-NEXT:    global_store_short v[0:1], v2, off
+; GFX10-NEXT:    global_store_short v[0:1], v3, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds i16, i16 addrspace(1)* %in, i64 %tid.ext
@@ -788,6 +1067,22 @@ define amdgpu_kernel void @v_test_v2i16_x_sub_64_64(<2 x i16> addrspace(1)* %out
 ; GFX9-NEXT:    v_pk_sub_i16 v2, v3, 64 op_sel_hi:[1,0]
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_sub_64_64:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_i16 v2, v3, 64 op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -856,6 +1151,22 @@ define amdgpu_kernel void @v_test_v2i16_x_sub_7_64(<2 x i16> addrspace(1)* %out,
 ; GFX9-NEXT:    v_pk_sub_i16 v2, v3, s4
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_sub_7_64:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_i16 v2, v3, 7 op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -924,6 +1235,22 @@ define amdgpu_kernel void @v_test_v2i16_x_sub_64_123(<2 x i16> addrspace(1)* %ou
 ; GFX9-NEXT:    v_pk_sub_i16 v2, v3, s4
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_sub_64_123:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_i16 v2, v3, 64 op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -990,6 +1317,22 @@ define amdgpu_kernel void @v_test_v2i16_x_sub_7_0(<2 x i16> addrspace(1)* %out,
 ; GFX9-NEXT:    v_pk_sub_i16 v2, v3, 7
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_sub_7_0:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_i16 v2, v3, 7
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1053,6 +1396,22 @@ define amdgpu_kernel void @v_test_v2i16_x_sub_0_16(<2 x i16> addrspace(1)* %out,
 ; GFX9-NEXT:    v_pk_sub_i16 v2, v3, 16 op_sel:[0,1] op_sel_hi:[1,0]
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_sub_0_16:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_i16 v2, v3, 16 op_sel:[0,1] op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1115,6 +1474,22 @@ define amdgpu_kernel void @v_test_v2i16_x_sub_0_1_0(<2 x i16> addrspace(1)* %out
 ; GFX9-NEXT:    v_pk_sub_i16 v2, v3, -4.0 op_sel:[0,1] op_sel_hi:[1,0]
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_sub_0_1_0:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_i16 v2, v3, -4.0 op_sel:[0,1] op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1177,6 +1552,22 @@ define amdgpu_kernel void @v_test_v2i16_x_sub_0_neg1_0(<2 x i16> addrspace(1)* %
 ; GFX9-NEXT:    v_pk_sub_i16 v2, v3, 4.0 op_sel:[0,1] op_sel_hi:[1,0]
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_sub_0_neg1_0:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_i16 v2, v3, 4.0 op_sel:[0,1] op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1245,6 +1636,22 @@ define amdgpu_kernel void @v_test_v2i16_x_add_neg32_neg32(<2 x i16> addrspace(1)
 ; GFX9-NEXT:    v_pk_sub_u16 v2, v3, 32 op_sel_hi:[1,0]
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_add_neg32_neg32:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_u16 v2, v3, 32 op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1307,6 +1714,22 @@ define amdgpu_kernel void @v_test_v2i16_x_add_0_neg32(<2 x i16> addrspace(1)* %o
 ; GFX9-NEXT:    v_pk_sub_u16 v2, v3, 32 op_sel:[0,1] op_sel_hi:[1,0]
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_add_0_neg32:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_u16 v2, v3, 32 op_sel:[0,1] op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1372,6 +1795,22 @@ define amdgpu_kernel void @v_test_v2i16_x_add_neg32_0(<2 x i16> addrspace(1)* %o
 ; GFX9-NEXT:    v_pk_sub_u16 v2, v3, 32
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_add_neg32_0:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_u16 v2, v3, 32
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1440,6 +1879,22 @@ define amdgpu_kernel void @v_test_v2i16_x_add_neg16_neg16(<2 x i16> addrspace(1)
 ; GFX9-NEXT:    v_pk_sub_u16 v2, v3, 16 op_sel_hi:[1,0]
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_add_neg16_neg16:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_u16 v2, v3, 16 op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1502,6 +1957,22 @@ define amdgpu_kernel void @v_test_v2i16_x_add_0_neg16(<2 x i16> addrspace(1)* %o
 ; GFX9-NEXT:    v_pk_sub_u16 v2, v3, 16 op_sel:[0,1] op_sel_hi:[1,0]
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_add_0_neg16:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_u16 v2, v3, 16 op_sel:[0,1] op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1567,6 +2038,22 @@ define amdgpu_kernel void @v_test_v2i16_x_add_neg16_0(<2 x i16> addrspace(1)* %o
 ; GFX9-NEXT:    v_pk_sub_u16 v2, v3, 16
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_add_neg16_0:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_u16 v2, v3, 16
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1634,6 +2121,22 @@ define amdgpu_kernel void @v_test_v2i16_x_add_neg_fpone(<2 x i16> addrspace(1)*
 ; GFX9-NEXT:    v_pk_sub_u16 v2, v3, 1.0 op_sel_hi:[1,0]
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_add_neg_fpone:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_u16 v2, v3, 1.0 op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1701,6 +2204,22 @@ define amdgpu_kernel void @v_test_v2i16_x_add_neg_negfpone(<2 x i16> addrspace(1
 ; GFX9-NEXT:    v_pk_sub_u16 v2, v3, -1.0 op_sel_hi:[1,0]
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_add_neg_negfpone:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_u16 v2, v3, -1.0 op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1768,6 +2287,22 @@ define amdgpu_kernel void @v_test_v2i16_x_add_neg_fptwo(<2 x i16> addrspace(1)*
 ; GFX9-NEXT:    v_pk_sub_u16 v2, v3, -2.0 op_sel_hi:[1,0]
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_add_neg_fptwo:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_u16 v2, v3, -2.0 op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1835,6 +2370,22 @@ define amdgpu_kernel void @v_test_v2i16_x_add_neg_negfptwo(<2 x i16> addrspace(1
 ; GFX9-NEXT:    v_pk_sub_u16 v2, v3, 2.0 op_sel_hi:[1,0]
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_add_neg_negfptwo:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_u16 v2, v3, 2.0 op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1897,6 +2448,22 @@ define amdgpu_kernel void @v_test_v2i16_x_add_undef_neg32(<2 x i16> addrspace(1)
 ; GFX9-NEXT:    v_pk_sub_u16 v2, v3, 32 op_sel:[0,1] op_sel_hi:[1,0]
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_add_undef_neg32:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_u16 v2, v3, 32 op_sel:[0,1] op_sel_hi:[1,0]
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext
@@ -1958,6 +2525,22 @@ define amdgpu_kernel void @v_test_v2i16_x_add_neg32_undef(<2 x i16> addrspace(1)
 ; GFX9-NEXT:    v_pk_sub_u16 v2, v3, 32
 ; GFX9-NEXT:    global_store_dword v[0:1], v2, off
 ; GFX9-NEXT:    s_endpgm
+;
+; GFX10-LABEL: v_test_v2i16_x_add_neg32_undef:
+; GFX10:       ; %bb.0:
+; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[0:1], 0x24
+; GFX10-NEXT:    v_lshlrev_b32_e32 v2, 2, v0
+; GFX10-NEXT:    ; implicit-def: $vcc_hi
+; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s2, s2, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s2, s3, 0, s2
+; GFX10-NEXT:    global_load_dword v3, v[0:1], off
+; GFX10-NEXT:    v_add_co_u32_e64 v0, s0, s0, v2
+; GFX10-NEXT:    v_add_co_ci_u32_e64 v1, s0, s1, 0, s0
+; GFX10-NEXT:    s_waitcnt vmcnt(0)
+; GFX10-NEXT:    v_pk_sub_u16 v2, v3, 32
+; GFX10-NEXT:    global_store_dword v[0:1], v2, off
+; GFX10-NEXT:    s_endpgm
   %tid = call i32 @llvm.amdgcn.workitem.id.x()
   %tid.ext = sext i32 %tid to i64
   %gep = getelementptr inbounds <2 x i16>, <2 x i16> addrspace(1)* %in, i64 %tid.ext