[llvm] 28f67ae - [AMDGPU] Fix some confusing check prefixes. NFC.
Jay Foad via llvm-commits
llvm-commits at lists.llvm.org
Wed Mar 9 09:05:57 PST 2022
Author: Jay Foad
Date: 2022-03-09T17:05:49Z
New Revision: 28f67aed9d7f42529e34b2d4252b089975be3698
URL: https://github.com/llvm/llvm-project/commit/28f67aed9d7f42529e34b2d4252b089975be3698
DIFF: https://github.com/llvm/llvm-project/commit/28f67aed9d7f42529e34b2d4252b089975be3698.diff
LOG: [AMDGPU] Fix some confusing check prefixes. NFC.
Tahiti is SI/GFX6.
Kaveri and Hawaii are CI/GFX7.
Fiji is VI/GFX8.
Added:
Modified:
llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-sextload-flat.mir
llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-zextload-flat.mir
llvm/test/CodeGen/AMDGPU/half.ll
llvm/test/CodeGen/AMDGPU/min.ll
llvm/test/CodeGen/AMDGPU/sint_to_fp.f64.ll
Removed:
################################################################################
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-sextload-flat.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-sextload-flat.mir
index 683ca816fcd6f..94601eb99bb8b 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-sextload-flat.mir
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-sextload-flat.mir
@@ -1,6 +1,6 @@
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
-# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -run-pass=legalizer -o - %s | FileCheck %s -check-prefix=SI
-# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=tahiti -run-pass=legalizer -o - %s | FileCheck %s -check-prefix=VI
+# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=tahiti -run-pass=legalizer -o - %s | FileCheck %s -check-prefix=SI
+# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -run-pass=legalizer -o - %s | FileCheck %s -check-prefix=VI
---
name: test_sextload_flat_i32_i8
body: |
diff --git a/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-zextload-flat.mir b/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-zextload-flat.mir
index 044d5fc4405bd..45c8b128379ce 100644
--- a/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-zextload-flat.mir
+++ b/llvm/test/CodeGen/AMDGPU/GlobalISel/legalize-zextload-flat.mir
@@ -1,6 +1,6 @@
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
-# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -run-pass=legalizer -o - %s | FileCheck %s --check-prefix=SI
-# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=tahiti -run-pass=legalizer -o - %s | FileCheck %s --check-prefix=VI
+# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=tahiti -run-pass=legalizer -o - %s | FileCheck %s --check-prefix=SI
+# RUN: llc -mtriple=amdgcn-mesa-mesa3d -mcpu=fiji -run-pass=legalizer -o - %s | FileCheck %s --check-prefix=VI
---
name: test_zextload_flat_i32_i8
body: |
diff --git a/llvm/test/CodeGen/AMDGPU/half.ll b/llvm/test/CodeGen/AMDGPU/half.ll
index 9669f01ea3e27..1fa7854991096 100644
--- a/llvm/test/CodeGen/AMDGPU/half.ll
+++ b/llvm/test/CodeGen/AMDGPU/half.ll
@@ -1,20 +1,20 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn-amd-amdhsa -mcpu=kaveri -verify-machineinstrs < %s | FileCheck -allow-deprecated-dag-overlap -enable-var-scope -check-prefixes=GCN,SI %s
+; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn-amd-amdhsa -mcpu=kaveri -verify-machineinstrs < %s | FileCheck -allow-deprecated-dag-overlap -enable-var-scope -check-prefixes=GCN,CI %s
; RUN: llc -amdgpu-scalarize-global-loads=false -mtriple=amdgcn-amd-amdhsa -mcpu=tonga -verify-machineinstrs < %s | FileCheck -allow-deprecated-dag-overlap -enable-var-scope -check-prefixes=GCN,VI %s
-; half args should be promoted to float for SI and lower.
+; half args should be promoted to float for CI and lower.
define amdgpu_kernel void @load_f16_arg(half addrspace(1)* %out, half %arg) #0 {
-; SI-LABEL: load_f16_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
-; SI-NEXT: s_load_dword s2, s[4:5], 0x2
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s0
-; SI-NEXT: v_mov_b32_e32 v1, s1
-; SI-NEXT: v_mov_b32_e32 v2, s2
-; SI-NEXT: flat_store_short v[0:1], v2
-; SI-NEXT: s_endpgm
+; CI-LABEL: load_f16_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; CI-NEXT: s_load_dword s2, s[4:5], 0x2
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s0
+; CI-NEXT: v_mov_b32_e32 v1, s1
+; CI-NEXT: v_mov_b32_e32 v2, s2
+; CI-NEXT: flat_store_short v[0:1], v2
+; CI-NEXT: s_endpgm
;
; VI-LABEL: load_f16_arg:
; VI: ; %bb.0:
@@ -31,16 +31,16 @@ define amdgpu_kernel void @load_f16_arg(half addrspace(1)* %out, half %arg) #0 {
}
define amdgpu_kernel void @load_v2f16_arg(<2 x half> addrspace(1)* %out, <2 x half> %arg) #0 {
-; SI-LABEL: load_v2f16_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
-; SI-NEXT: s_load_dword s2, s[4:5], 0x2
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s0
-; SI-NEXT: v_mov_b32_e32 v1, s1
-; SI-NEXT: v_mov_b32_e32 v2, s2
-; SI-NEXT: flat_store_dword v[0:1], v2
-; SI-NEXT: s_endpgm
+; CI-LABEL: load_v2f16_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; CI-NEXT: s_load_dword s2, s[4:5], 0x2
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s0
+; CI-NEXT: v_mov_b32_e32 v1, s1
+; CI-NEXT: v_mov_b32_e32 v2, s2
+; CI-NEXT: flat_store_dword v[0:1], v2
+; CI-NEXT: s_endpgm
;
; VI-LABEL: load_v2f16_arg:
; VI: ; %bb.0:
@@ -57,22 +57,22 @@ define amdgpu_kernel void @load_v2f16_arg(<2 x half> addrspace(1)* %out, <2 x ha
}
define amdgpu_kernel void @load_v3f16_arg(<3 x half> addrspace(1)* %out, <3 x half> %arg) #0 {
-; SI-LABEL: load_v3f16_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
-; SI-NEXT: s_load_dwordx2 s[2:3], s[4:5], 0x2
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: s_add_u32 s4, s0, 4
-; SI-NEXT: s_addc_u32 s5, s1, 0
-; SI-NEXT: v_mov_b32_e32 v2, s4
-; SI-NEXT: v_mov_b32_e32 v4, s3
-; SI-NEXT: v_mov_b32_e32 v0, s0
-; SI-NEXT: v_mov_b32_e32 v3, s5
-; SI-NEXT: v_mov_b32_e32 v1, s1
-; SI-NEXT: v_mov_b32_e32 v5, s2
-; SI-NEXT: flat_store_short v[2:3], v4
-; SI-NEXT: flat_store_dword v[0:1], v5
-; SI-NEXT: s_endpgm
+; CI-LABEL: load_v3f16_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; CI-NEXT: s_load_dwordx2 s[2:3], s[4:5], 0x2
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: s_add_u32 s4, s0, 4
+; CI-NEXT: s_addc_u32 s5, s1, 0
+; CI-NEXT: v_mov_b32_e32 v2, s4
+; CI-NEXT: v_mov_b32_e32 v4, s3
+; CI-NEXT: v_mov_b32_e32 v0, s0
+; CI-NEXT: v_mov_b32_e32 v3, s5
+; CI-NEXT: v_mov_b32_e32 v1, s1
+; CI-NEXT: v_mov_b32_e32 v5, s2
+; CI-NEXT: flat_store_short v[2:3], v4
+; CI-NEXT: flat_store_dword v[0:1], v5
+; CI-NEXT: s_endpgm
;
; VI-LABEL: load_v3f16_arg:
; VI: ; %bb.0:
@@ -97,17 +97,17 @@ define amdgpu_kernel void @load_v3f16_arg(<3 x half> addrspace(1)* %out, <3 x ha
; FIXME: Why not one load?
define amdgpu_kernel void @load_v4f16_arg(<4 x half> addrspace(1)* %out, <4 x half> %arg) #0 {
-; SI-LABEL: load_v4f16_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
-; SI-NEXT: s_load_dwordx2 s[2:3], s[4:5], 0x2
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s0
-; SI-NEXT: v_mov_b32_e32 v2, s2
-; SI-NEXT: v_mov_b32_e32 v1, s1
-; SI-NEXT: v_mov_b32_e32 v3, s3
-; SI-NEXT: flat_store_dwordx2 v[0:1], v[2:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: load_v4f16_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; CI-NEXT: s_load_dwordx2 s[2:3], s[4:5], 0x2
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s0
+; CI-NEXT: v_mov_b32_e32 v2, s2
+; CI-NEXT: v_mov_b32_e32 v1, s1
+; CI-NEXT: v_mov_b32_e32 v3, s3
+; CI-NEXT: flat_store_dwordx2 v[0:1], v[2:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: load_v4f16_arg:
; VI: ; %bb.0:
@@ -125,19 +125,19 @@ define amdgpu_kernel void @load_v4f16_arg(<4 x half> addrspace(1)* %out, <4 x ha
}
define amdgpu_kernel void @load_v8f16_arg(<8 x half> addrspace(1)* %out, <8 x half> %arg) #0 {
-; SI-LABEL: load_v8f16_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x0
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x4
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v4, s6
-; SI-NEXT: v_mov_b32_e32 v0, s0
-; SI-NEXT: v_mov_b32_e32 v5, s7
-; SI-NEXT: v_mov_b32_e32 v1, s1
-; SI-NEXT: v_mov_b32_e32 v2, s2
-; SI-NEXT: v_mov_b32_e32 v3, s3
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: load_v8f16_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx2 s[6:7], s[4:5], 0x0
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x4
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v4, s6
+; CI-NEXT: v_mov_b32_e32 v0, s0
+; CI-NEXT: v_mov_b32_e32 v5, s7
+; CI-NEXT: v_mov_b32_e32 v1, s1
+; CI-NEXT: v_mov_b32_e32 v2, s2
+; CI-NEXT: v_mov_b32_e32 v3, s3
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: load_v8f16_arg:
; VI: ; %bb.0:
@@ -157,18 +157,18 @@ define amdgpu_kernel void @load_v8f16_arg(<8 x half> addrspace(1)* %out, <8 x ha
}
define amdgpu_kernel void @extload_v2f16_arg(<2 x float> addrspace(1)* %out, <2 x half> %in) #0 {
-; SI-LABEL: extload_v2f16_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dword s2, s[4:5], 0x2
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: s_lshr_b32 s3, s2, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v1, s3
-; SI-NEXT: v_cvt_f32_f16_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v3, s1
-; SI-NEXT: v_mov_b32_e32 v2, s0
-; SI-NEXT: flat_store_dwordx2 v[2:3], v[0:1]
-; SI-NEXT: s_endpgm
+; CI-LABEL: extload_v2f16_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dword s2, s[4:5], 0x2
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: s_lshr_b32 s3, s2, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v1, s3
+; CI-NEXT: v_cvt_f32_f16_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v3, s1
+; CI-NEXT: v_mov_b32_e32 v2, s0
+; CI-NEXT: flat_store_dwordx2 v[2:3], v[0:1]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: extload_v2f16_arg:
; VI: ; %bb.0:
@@ -188,16 +188,16 @@ define amdgpu_kernel void @extload_v2f16_arg(<2 x float> addrspace(1)* %out, <2
}
define amdgpu_kernel void @extload_f16_to_f32_arg(float addrspace(1)* %out, half %arg) #0 {
-; SI-LABEL: extload_f16_to_f32_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dword s2, s[4:5], 0x2
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_cvt_f32_f16_e32 v2, s2
-; SI-NEXT: v_mov_b32_e32 v0, s0
-; SI-NEXT: v_mov_b32_e32 v1, s1
-; SI-NEXT: flat_store_dword v[0:1], v2
-; SI-NEXT: s_endpgm
+; CI-LABEL: extload_f16_to_f32_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dword s2, s[4:5], 0x2
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_cvt_f32_f16_e32 v2, s2
+; CI-NEXT: v_mov_b32_e32 v0, s0
+; CI-NEXT: v_mov_b32_e32 v1, s1
+; CI-NEXT: flat_store_dword v[0:1], v2
+; CI-NEXT: s_endpgm
;
; VI-LABEL: extload_f16_to_f32_arg:
; VI: ; %bb.0:
@@ -215,18 +215,18 @@ define amdgpu_kernel void @extload_f16_to_f32_arg(float addrspace(1)* %out, half
}
define amdgpu_kernel void @extload_v2f16_to_v2f32_arg(<2 x float> addrspace(1)* %out, <2 x half> %arg) #0 {
-; SI-LABEL: extload_v2f16_to_v2f32_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dword s2, s[4:5], 0x2
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: s_lshr_b32 s3, s2, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v1, s3
-; SI-NEXT: v_cvt_f32_f16_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v3, s1
-; SI-NEXT: v_mov_b32_e32 v2, s0
-; SI-NEXT: flat_store_dwordx2 v[2:3], v[0:1]
-; SI-NEXT: s_endpgm
+; CI-LABEL: extload_v2f16_to_v2f32_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dword s2, s[4:5], 0x2
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: s_lshr_b32 s3, s2, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v1, s3
+; CI-NEXT: v_cvt_f32_f16_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v3, s1
+; CI-NEXT: v_mov_b32_e32 v2, s0
+; CI-NEXT: flat_store_dwordx2 v[2:3], v[0:1]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: extload_v2f16_to_v2f32_arg:
; VI: ; %bb.0:
@@ -246,19 +246,19 @@ define amdgpu_kernel void @extload_v2f16_to_v2f32_arg(<2 x float> addrspace(1)*
}
define amdgpu_kernel void @extload_v3f16_to_v3f32_arg(<3 x float> addrspace(1)* %out, <3 x half> %arg) #0 {
-; SI-LABEL: extload_v3f16_to_v3f32_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x2
-; SI-NEXT: s_load_dwordx2 s[2:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: s_lshr_b32 s4, s0, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v2, s1
-; SI-NEXT: v_cvt_f32_f16_e32 v1, s4
-; SI-NEXT: v_cvt_f32_f16_e32 v0, s0
-; SI-NEXT: v_mov_b32_e32 v4, s3
-; SI-NEXT: v_mov_b32_e32 v3, s2
-; SI-NEXT: flat_store_dwordx3 v[3:4], v[0:2]
-; SI-NEXT: s_endpgm
+; CI-LABEL: extload_v3f16_to_v3f32_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x2
+; CI-NEXT: s_load_dwordx2 s[2:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: s_lshr_b32 s4, s0, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v2, s1
+; CI-NEXT: v_cvt_f32_f16_e32 v1, s4
+; CI-NEXT: v_cvt_f32_f16_e32 v0, s0
+; CI-NEXT: v_mov_b32_e32 v4, s3
+; CI-NEXT: v_mov_b32_e32 v3, s2
+; CI-NEXT: flat_store_dwordx3 v[3:4], v[0:2]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: extload_v3f16_to_v3f32_arg:
; VI: ; %bb.0:
@@ -279,21 +279,21 @@ define amdgpu_kernel void @extload_v3f16_to_v3f32_arg(<3 x float> addrspace(1)*
}
define amdgpu_kernel void @extload_v4f16_to_v4f32_arg(<4 x float> addrspace(1)* %out, <4 x half> %arg) #0 {
-; SI-LABEL: extload_v4f16_to_v4f32_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x2
-; SI-NEXT: s_load_dwordx2 s[2:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: s_lshr_b32 s4, s1, 16
-; SI-NEXT: s_lshr_b32 s5, s0, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v2, s1
-; SI-NEXT: v_cvt_f32_f16_e32 v3, s4
-; SI-NEXT: v_cvt_f32_f16_e32 v1, s5
-; SI-NEXT: v_cvt_f32_f16_e32 v0, s0
-; SI-NEXT: v_mov_b32_e32 v5, s3
-; SI-NEXT: v_mov_b32_e32 v4, s2
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: extload_v4f16_to_v4f32_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x2
+; CI-NEXT: s_load_dwordx2 s[2:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: s_lshr_b32 s4, s1, 16
+; CI-NEXT: s_lshr_b32 s5, s0, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v2, s1
+; CI-NEXT: v_cvt_f32_f16_e32 v3, s4
+; CI-NEXT: v_cvt_f32_f16_e32 v1, s5
+; CI-NEXT: v_cvt_f32_f16_e32 v0, s0
+; CI-NEXT: v_mov_b32_e32 v5, s3
+; CI-NEXT: v_mov_b32_e32 v4, s2
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: extload_v4f16_to_v4f32_arg:
; VI: ; %bb.0:
@@ -316,33 +316,33 @@ define amdgpu_kernel void @extload_v4f16_to_v4f32_arg(<4 x float> addrspace(1)*
}
define amdgpu_kernel void @extload_v8f16_to_v8f32_arg(<8 x float> addrspace(1)* %out, <8 x half> %arg) #0 {
-; SI-LABEL: extload_v8f16_to_v8f32_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x4
-; SI-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: s_lshr_b32 s6, s1, 16
-; SI-NEXT: s_lshr_b32 s7, s0, 16
-; SI-NEXT: s_lshr_b32 s8, s3, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v3, s6
-; SI-NEXT: s_lshr_b32 s6, s2, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v7, s8
-; SI-NEXT: v_cvt_f32_f16_e32 v5, s6
-; SI-NEXT: v_cvt_f32_f16_e32 v0, s0
-; SI-NEXT: v_cvt_f32_f16_e32 v6, s3
-; SI-NEXT: v_cvt_f32_f16_e32 v4, s2
-; SI-NEXT: s_add_u32 s0, s4, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v2, s1
-; SI-NEXT: s_addc_u32 s1, s5, 0
-; SI-NEXT: v_cvt_f32_f16_e32 v1, s7
-; SI-NEXT: v_mov_b32_e32 v9, s1
-; SI-NEXT: v_mov_b32_e32 v8, s0
-; SI-NEXT: flat_store_dwordx4 v[8:9], v[4:7]
-; SI-NEXT: s_nop 0
-; SI-NEXT: v_mov_b32_e32 v4, s4
-; SI-NEXT: v_mov_b32_e32 v5, s5
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: extload_v8f16_to_v8f32_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x4
+; CI-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: s_lshr_b32 s6, s1, 16
+; CI-NEXT: s_lshr_b32 s7, s0, 16
+; CI-NEXT: s_lshr_b32 s8, s3, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v3, s6
+; CI-NEXT: s_lshr_b32 s6, s2, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v7, s8
+; CI-NEXT: v_cvt_f32_f16_e32 v5, s6
+; CI-NEXT: v_cvt_f32_f16_e32 v0, s0
+; CI-NEXT: v_cvt_f32_f16_e32 v6, s3
+; CI-NEXT: v_cvt_f32_f16_e32 v4, s2
+; CI-NEXT: s_add_u32 s0, s4, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v2, s1
+; CI-NEXT: s_addc_u32 s1, s5, 0
+; CI-NEXT: v_cvt_f32_f16_e32 v1, s7
+; CI-NEXT: v_mov_b32_e32 v9, s1
+; CI-NEXT: v_mov_b32_e32 v8, s0
+; CI-NEXT: flat_store_dwordx4 v[8:9], v[4:7]
+; CI-NEXT: s_nop 0
+; CI-NEXT: v_mov_b32_e32 v4, s4
+; CI-NEXT: v_mov_b32_e32 v5, s5
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: extload_v8f16_to_v8f32_arg:
; VI: ; %bb.0:
@@ -377,18 +377,18 @@ define amdgpu_kernel void @extload_v8f16_to_v8f32_arg(<8 x float> addrspace(1)*
}
define amdgpu_kernel void @extload_f16_to_f64_arg(double addrspace(1)* %out, half %arg) #0 {
-; SI-LABEL: extload_f16_to_f64_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dword s0, s[4:5], 0x2
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_cvt_f32_f16_e32 v0, s0
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v3, s1
-; SI-NEXT: v_mov_b32_e32 v2, s0
-; SI-NEXT: flat_store_dwordx2 v[2:3], v[0:1]
-; SI-NEXT: s_endpgm
+; CI-LABEL: extload_f16_to_f64_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dword s0, s[4:5], 0x2
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_cvt_f32_f16_e32 v0, s0
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v3, s1
+; CI-NEXT: v_mov_b32_e32 v2, s0
+; CI-NEXT: flat_store_dwordx2 v[2:3], v[0:1]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: extload_f16_to_f64_arg:
; VI: ; %bb.0:
@@ -408,21 +408,21 @@ define amdgpu_kernel void @extload_f16_to_f64_arg(double addrspace(1)* %out, hal
}
define amdgpu_kernel void @extload_v2f16_to_v2f64_arg(<2 x double> addrspace(1)* %out, <2 x half> %arg) #0 {
-; SI-LABEL: extload_v2f16_to_v2f64_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dword s0, s[4:5], 0x2
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: s_lshr_b32 s1, s0, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v0, s1
-; SI-NEXT: v_cvt_f32_f16_e32 v1, s0
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
-; SI-NEXT: v_cvt_f64_f32_e32 v[2:3], v0
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v1
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v5, s1
-; SI-NEXT: v_mov_b32_e32 v4, s0
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: extload_v2f16_to_v2f64_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dword s0, s[4:5], 0x2
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: s_lshr_b32 s1, s0, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v0, s1
+; CI-NEXT: v_cvt_f32_f16_e32 v1, s0
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; CI-NEXT: v_cvt_f64_f32_e32 v[2:3], v0
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v1
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v5, s1
+; CI-NEXT: v_mov_b32_e32 v4, s0
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: extload_v2f16_to_v2f64_arg:
; VI: ; %bb.0:
@@ -445,27 +445,27 @@ define amdgpu_kernel void @extload_v2f16_to_v2f64_arg(<2 x double> addrspace(1)*
}
define amdgpu_kernel void @extload_v3f16_to_v3f64_arg(<3 x double> addrspace(1)* %out, <3 x half> %arg) #0 {
-; SI-LABEL: extload_v3f16_to_v3f64_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x2
-; SI-NEXT: s_load_dwordx2 s[2:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_cvt_f32_f16_e32 v0, s1
-; SI-NEXT: s_lshr_b32 s4, s0, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v1, s0
-; SI-NEXT: v_cvt_f32_f16_e32 v2, s4
-; SI-NEXT: s_add_u32 s0, s2, 16
-; SI-NEXT: v_cvt_f64_f32_e32 v[4:5], v0
-; SI-NEXT: s_addc_u32 s1, s3, 0
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v1
-; SI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
-; SI-NEXT: v_mov_b32_e32 v7, s1
-; SI-NEXT: v_mov_b32_e32 v6, s0
-; SI-NEXT: flat_store_dwordx2 v[6:7], v[4:5]
-; SI-NEXT: v_mov_b32_e32 v5, s3
-; SI-NEXT: v_mov_b32_e32 v4, s2
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: extload_v3f16_to_v3f64_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x2
+; CI-NEXT: s_load_dwordx2 s[2:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_cvt_f32_f16_e32 v0, s1
+; CI-NEXT: s_lshr_b32 s4, s0, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v1, s0
+; CI-NEXT: v_cvt_f32_f16_e32 v2, s4
+; CI-NEXT: s_add_u32 s0, s2, 16
+; CI-NEXT: v_cvt_f64_f32_e32 v[4:5], v0
+; CI-NEXT: s_addc_u32 s1, s3, 0
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v1
+; CI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
+; CI-NEXT: v_mov_b32_e32 v7, s1
+; CI-NEXT: v_mov_b32_e32 v6, s0
+; CI-NEXT: flat_store_dwordx2 v[6:7], v[4:5]
+; CI-NEXT: v_mov_b32_e32 v5, s3
+; CI-NEXT: v_mov_b32_e32 v4, s2
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: extload_v3f16_to_v3f64_arg:
; VI: ; %bb.0:
@@ -494,31 +494,31 @@ define amdgpu_kernel void @extload_v3f16_to_v3f64_arg(<3 x double> addrspace(1)*
}
define amdgpu_kernel void @extload_v4f16_to_v4f64_arg(<4 x double> addrspace(1)* %out, <4 x half> %arg) #0 {
-; SI-LABEL: extload_v4f16_to_v4f64_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x2
-; SI-NEXT: s_load_dwordx2 s[2:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: s_lshr_b32 s4, s1, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v4, s4
-; SI-NEXT: v_cvt_f32_f16_e32 v5, s1
-; SI-NEXT: s_lshr_b32 s5, s0, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v0, s0
-; SI-NEXT: v_cvt_f32_f16_e32 v2, s5
-; SI-NEXT: v_cvt_f64_f32_e32 v[6:7], v4
-; SI-NEXT: v_cvt_f64_f32_e32 v[4:5], v5
-; SI-NEXT: s_add_u32 s0, s2, 16
-; SI-NEXT: s_addc_u32 s1, s3, 0
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v0
-; SI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
-; SI-NEXT: v_mov_b32_e32 v9, s1
-; SI-NEXT: v_mov_b32_e32 v8, s0
-; SI-NEXT: flat_store_dwordx4 v[8:9], v[4:7]
-; SI-NEXT: s_nop 0
-; SI-NEXT: v_mov_b32_e32 v5, s3
-; SI-NEXT: v_mov_b32_e32 v4, s2
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: extload_v4f16_to_v4f64_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x2
+; CI-NEXT: s_load_dwordx2 s[2:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: s_lshr_b32 s4, s1, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v4, s4
+; CI-NEXT: v_cvt_f32_f16_e32 v5, s1
+; CI-NEXT: s_lshr_b32 s5, s0, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v0, s0
+; CI-NEXT: v_cvt_f32_f16_e32 v2, s5
+; CI-NEXT: v_cvt_f64_f32_e32 v[6:7], v4
+; CI-NEXT: v_cvt_f64_f32_e32 v[4:5], v5
+; CI-NEXT: s_add_u32 s0, s2, 16
+; CI-NEXT: s_addc_u32 s1, s3, 0
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v0
+; CI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
+; CI-NEXT: v_mov_b32_e32 v9, s1
+; CI-NEXT: v_mov_b32_e32 v8, s0
+; CI-NEXT: flat_store_dwordx4 v[8:9], v[4:7]
+; CI-NEXT: s_nop 0
+; CI-NEXT: v_mov_b32_e32 v5, s3
+; CI-NEXT: v_mov_b32_e32 v4, s2
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: extload_v4f16_to_v4f64_arg:
; VI: ; %bb.0:
@@ -551,52 +551,52 @@ define amdgpu_kernel void @extload_v4f16_to_v4f64_arg(<4 x double> addrspace(1)*
}
define amdgpu_kernel void @extload_v8f16_to_v8f64_arg(<8 x double> addrspace(1)* %out, <8 x half> %arg) #0 {
-; SI-LABEL: extload_v8f16_to_v8f64_arg:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x4
-; SI-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: s_lshr_b32 s6, s3, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v0, s6
-; SI-NEXT: v_cvt_f32_f16_e32 v12, s3
-; SI-NEXT: s_lshr_b32 s7, s2, 16
-; SI-NEXT: s_lshr_b32 s8, s1, 16
-; SI-NEXT: s_lshr_b32 s6, s0, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v1, s7
-; SI-NEXT: v_cvt_f32_f16_e32 v8, s2
-; SI-NEXT: v_cvt_f32_f16_e32 v9, s0
-; SI-NEXT: s_add_u32 s0, s4, 48
-; SI-NEXT: v_cvt_f32_f16_e32 v5, s1
-; SI-NEXT: v_cvt_f64_f32_e32 v[14:15], v0
-; SI-NEXT: v_cvt_f64_f32_e32 v[12:13], v12
-; SI-NEXT: s_addc_u32 s1, s5, 0
-; SI-NEXT: v_cvt_f32_f16_e32 v4, s8
-; SI-NEXT: v_mov_b32_e32 v17, s1
-; SI-NEXT: v_mov_b32_e32 v16, s0
-; SI-NEXT: s_add_u32 s0, s4, 32
-; SI-NEXT: v_cvt_f32_f16_e32 v2, s6
-; SI-NEXT: v_cvt_f64_f32_e32 v[10:11], v1
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v9
-; SI-NEXT: v_cvt_f64_f32_e32 v[8:9], v8
-; SI-NEXT: s_addc_u32 s1, s5, 0
-; SI-NEXT: flat_store_dwordx4 v[16:17], v[12:15]
-; SI-NEXT: v_cvt_f64_f32_e32 v[6:7], v4
-; SI-NEXT: v_mov_b32_e32 v13, s1
-; SI-NEXT: v_cvt_f64_f32_e32 v[4:5], v5
-; SI-NEXT: v_mov_b32_e32 v12, s0
-; SI-NEXT: s_add_u32 s0, s4, 16
-; SI-NEXT: s_addc_u32 s1, s5, 0
-; SI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
-; SI-NEXT: flat_store_dwordx4 v[12:13], v[8:11]
-; SI-NEXT: s_nop 0
-; SI-NEXT: v_mov_b32_e32 v9, s1
-; SI-NEXT: v_mov_b32_e32 v8, s0
-; SI-NEXT: flat_store_dwordx4 v[8:9], v[4:7]
-; SI-NEXT: s_nop 0
-; SI-NEXT: v_mov_b32_e32 v4, s4
-; SI-NEXT: v_mov_b32_e32 v5, s5
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: extload_v8f16_to_v8f64_arg:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x4
+; CI-NEXT: s_load_dwordx2 s[4:5], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: s_lshr_b32 s6, s3, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v0, s6
+; CI-NEXT: v_cvt_f32_f16_e32 v12, s3
+; CI-NEXT: s_lshr_b32 s7, s2, 16
+; CI-NEXT: s_lshr_b32 s8, s1, 16
+; CI-NEXT: s_lshr_b32 s6, s0, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v1, s7
+; CI-NEXT: v_cvt_f32_f16_e32 v8, s2
+; CI-NEXT: v_cvt_f32_f16_e32 v9, s0
+; CI-NEXT: s_add_u32 s0, s4, 48
+; CI-NEXT: v_cvt_f32_f16_e32 v5, s1
+; CI-NEXT: v_cvt_f64_f32_e32 v[14:15], v0
+; CI-NEXT: v_cvt_f64_f32_e32 v[12:13], v12
+; CI-NEXT: s_addc_u32 s1, s5, 0
+; CI-NEXT: v_cvt_f32_f16_e32 v4, s8
+; CI-NEXT: v_mov_b32_e32 v17, s1
+; CI-NEXT: v_mov_b32_e32 v16, s0
+; CI-NEXT: s_add_u32 s0, s4, 32
+; CI-NEXT: v_cvt_f32_f16_e32 v2, s6
+; CI-NEXT: v_cvt_f64_f32_e32 v[10:11], v1
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v9
+; CI-NEXT: v_cvt_f64_f32_e32 v[8:9], v8
+; CI-NEXT: s_addc_u32 s1, s5, 0
+; CI-NEXT: flat_store_dwordx4 v[16:17], v[12:15]
+; CI-NEXT: v_cvt_f64_f32_e32 v[6:7], v4
+; CI-NEXT: v_mov_b32_e32 v13, s1
+; CI-NEXT: v_cvt_f64_f32_e32 v[4:5], v5
+; CI-NEXT: v_mov_b32_e32 v12, s0
+; CI-NEXT: s_add_u32 s0, s4, 16
+; CI-NEXT: s_addc_u32 s1, s5, 0
+; CI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
+; CI-NEXT: flat_store_dwordx4 v[12:13], v[8:11]
+; CI-NEXT: s_nop 0
+; CI-NEXT: v_mov_b32_e32 v9, s1
+; CI-NEXT: v_mov_b32_e32 v8, s0
+; CI-NEXT: flat_store_dwordx4 v[8:9], v[4:7]
+; CI-NEXT: s_nop 0
+; CI-NEXT: v_mov_b32_e32 v4, s4
+; CI-NEXT: v_mov_b32_e32 v5, s5
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: extload_v8f16_to_v8f64_arg:
; VI: ; %bb.0:
@@ -742,21 +742,21 @@ define amdgpu_kernel void @global_extload_f16_to_f32(float addrspace(1)* %out, h
}
define amdgpu_kernel void @global_extload_v2f16_to_v2f32(<2 x float> addrspace(1)* %out, <2 x half> addrspace(1)* %in) #0 {
-; SI-LABEL: global_extload_v2f16_to_v2f32:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_load_dword v1, v[0:1]
-; SI-NEXT: v_mov_b32_e32 v2, s0
-; SI-NEXT: v_mov_b32_e32 v3, s1
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_cvt_f32_f16_e32 v0, v1
-; SI-NEXT: v_lshrrev_b32_e32 v1, 16, v1
-; SI-NEXT: v_cvt_f32_f16_e32 v1, v1
-; SI-NEXT: flat_store_dwordx2 v[2:3], v[0:1]
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_extload_v2f16_to_v2f32:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_load_dword v1, v[0:1]
+; CI-NEXT: v_mov_b32_e32 v2, s0
+; CI-NEXT: v_mov_b32_e32 v3, s1
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_cvt_f32_f16_e32 v0, v1
+; CI-NEXT: v_lshrrev_b32_e32 v1, 16, v1
+; CI-NEXT: v_cvt_f32_f16_e32 v1, v1
+; CI-NEXT: flat_store_dwordx2 v[2:3], v[0:1]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_extload_v2f16_to_v2f32:
; VI: ; %bb.0:
@@ -779,22 +779,22 @@ define amdgpu_kernel void @global_extload_v2f16_to_v2f32(<2 x float> addrspace(1
}
define amdgpu_kernel void @global_extload_v3f16_to_v3f32(<3 x float> addrspace(1)* %out, <3 x half> addrspace(1)* %in) #0 {
-; SI-LABEL: global_extload_v3f16_to_v3f32:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_load_dwordx2 v[1:2], v[0:1]
-; SI-NEXT: v_mov_b32_e32 v3, s0
-; SI-NEXT: v_mov_b32_e32 v4, s1
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_cvt_f32_f16_e32 v0, v1
-; SI-NEXT: v_lshrrev_b32_e32 v1, 16, v1
-; SI-NEXT: v_cvt_f32_f16_e32 v2, v2
-; SI-NEXT: v_cvt_f32_f16_e32 v1, v1
-; SI-NEXT: flat_store_dwordx3 v[3:4], v[0:2]
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_extload_v3f16_to_v3f32:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_load_dwordx2 v[1:2], v[0:1]
+; CI-NEXT: v_mov_b32_e32 v3, s0
+; CI-NEXT: v_mov_b32_e32 v4, s1
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_cvt_f32_f16_e32 v0, v1
+; CI-NEXT: v_lshrrev_b32_e32 v1, 16, v1
+; CI-NEXT: v_cvt_f32_f16_e32 v2, v2
+; CI-NEXT: v_cvt_f32_f16_e32 v1, v1
+; CI-NEXT: flat_store_dwordx3 v[3:4], v[0:2]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_extload_v3f16_to_v3f32:
; VI: ; %bb.0:
@@ -818,24 +818,24 @@ define amdgpu_kernel void @global_extload_v3f16_to_v3f32(<3 x float> addrspace(1
}
define amdgpu_kernel void @global_extload_v4f16_to_v4f32(<4 x float> addrspace(1)* %out, <4 x half> addrspace(1)* %in) #0 {
-; SI-LABEL: global_extload_v4f16_to_v4f32:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_load_dwordx2 v[3:4], v[0:1]
-; SI-NEXT: v_mov_b32_e32 v5, s1
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_cvt_f32_f16_e32 v2, v4
-; SI-NEXT: v_lshrrev_b32_e32 v1, 16, v4
-; SI-NEXT: v_lshrrev_b32_e32 v4, 16, v3
-; SI-NEXT: v_cvt_f32_f16_e32 v0, v3
-; SI-NEXT: v_cvt_f32_f16_e32 v3, v1
-; SI-NEXT: v_cvt_f32_f16_e32 v1, v4
-; SI-NEXT: v_mov_b32_e32 v4, s0
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_extload_v4f16_to_v4f32:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_load_dwordx2 v[3:4], v[0:1]
+; CI-NEXT: v_mov_b32_e32 v5, s1
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_cvt_f32_f16_e32 v2, v4
+; CI-NEXT: v_lshrrev_b32_e32 v1, 16, v4
+; CI-NEXT: v_lshrrev_b32_e32 v4, 16, v3
+; CI-NEXT: v_cvt_f32_f16_e32 v0, v3
+; CI-NEXT: v_cvt_f32_f16_e32 v3, v1
+; CI-NEXT: v_cvt_f32_f16_e32 v1, v4
+; CI-NEXT: v_mov_b32_e32 v4, s0
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_extload_v4f16_to_v4f32:
; VI: ; %bb.0:
@@ -860,35 +860,35 @@ define amdgpu_kernel void @global_extload_v4f16_to_v4f32(<4 x float> addrspace(1
}
define amdgpu_kernel void @global_extload_v8f16_to_v8f32(<8 x float> addrspace(1)* %out, <8 x half> addrspace(1)* %in) #0 {
-; SI-LABEL: global_extload_v8f16_to_v8f32:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
-; SI-NEXT: s_add_u32 s2, s0, 16
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: v_mov_b32_e32 v13, s1
-; SI-NEXT: v_mov_b32_e32 v12, s0
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_cvt_f32_f16_e32 v10, v3
-; SI-NEXT: v_cvt_f32_f16_e32 v8, v2
-; SI-NEXT: v_lshrrev_b32_e32 v3, 16, v3
-; SI-NEXT: v_lshrrev_b32_e32 v2, 16, v2
-; SI-NEXT: v_cvt_f32_f16_e32 v6, v1
-; SI-NEXT: v_cvt_f32_f16_e32 v4, v0
-; SI-NEXT: v_lshrrev_b32_e32 v1, 16, v1
-; SI-NEXT: v_lshrrev_b32_e32 v0, 16, v0
-; SI-NEXT: v_cvt_f32_f16_e32 v11, v3
-; SI-NEXT: v_cvt_f32_f16_e32 v9, v2
-; SI-NEXT: v_cvt_f32_f16_e32 v7, v1
-; SI-NEXT: v_cvt_f32_f16_e32 v5, v0
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_store_dwordx4 v[0:1], v[8:11]
-; SI-NEXT: flat_store_dwordx4 v[12:13], v[4:7]
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_extload_v8f16_to_v8f32:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
+; CI-NEXT: s_add_u32 s2, s0, 16
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: v_mov_b32_e32 v13, s1
+; CI-NEXT: v_mov_b32_e32 v12, s0
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_cvt_f32_f16_e32 v10, v3
+; CI-NEXT: v_cvt_f32_f16_e32 v8, v2
+; CI-NEXT: v_lshrrev_b32_e32 v3, 16, v3
+; CI-NEXT: v_lshrrev_b32_e32 v2, 16, v2
+; CI-NEXT: v_cvt_f32_f16_e32 v6, v1
+; CI-NEXT: v_cvt_f32_f16_e32 v4, v0
+; CI-NEXT: v_lshrrev_b32_e32 v1, 16, v1
+; CI-NEXT: v_lshrrev_b32_e32 v0, 16, v0
+; CI-NEXT: v_cvt_f32_f16_e32 v11, v3
+; CI-NEXT: v_cvt_f32_f16_e32 v9, v2
+; CI-NEXT: v_cvt_f32_f16_e32 v7, v1
+; CI-NEXT: v_cvt_f32_f16_e32 v5, v0
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_store_dwordx4 v[0:1], v[8:11]
+; CI-NEXT: flat_store_dwordx4 v[12:13], v[4:7]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_extload_v8f16_to_v8f32:
; VI: ; %bb.0:
@@ -922,63 +922,63 @@ define amdgpu_kernel void @global_extload_v8f16_to_v8f32(<8 x float> addrspace(1
}
define amdgpu_kernel void @global_extload_v16f16_to_v16f32(<16 x float> addrspace(1)* %out, <16 x half> addrspace(1)* %in) #0 {
-; SI-LABEL: global_extload_v16f16_to_v16f32:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: s_add_u32 s4, s2, 16
-; SI-NEXT: v_mov_b32_e32 v5, s3
-; SI-NEXT: s_addc_u32 s5, s3, 0
-; SI-NEXT: v_mov_b32_e32 v0, s4
-; SI-NEXT: v_mov_b32_e32 v4, s2
-; SI-NEXT: v_mov_b32_e32 v1, s5
-; SI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
-; SI-NEXT: flat_load_dwordx4 v[4:7], v[4:5]
-; SI-NEXT: s_add_u32 s2, s0, 16
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: v_mov_b32_e32 v14, s3
-; SI-NEXT: v_mov_b32_e32 v13, s2
-; SI-NEXT: s_add_u32 s2, s0, 48
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: s_waitcnt vmcnt(1)
-; SI-NEXT: v_cvt_f32_f16_e32 v8, v1
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_cvt_f32_f16_e32 v11, v7
-; SI-NEXT: v_cvt_f32_f16_e32 v9, v6
-; SI-NEXT: v_lshrrev_b32_e32 v7, 16, v7
-; SI-NEXT: v_lshrrev_b32_e32 v6, 16, v6
-; SI-NEXT: v_cvt_f32_f16_e32 v12, v7
-; SI-NEXT: v_cvt_f32_f16_e32 v10, v6
-; SI-NEXT: v_lshrrev_b32_e32 v1, 16, v1
-; SI-NEXT: v_lshrrev_b32_e32 v16, 16, v5
-; SI-NEXT: v_lshrrev_b32_e32 v17, 16, v4
-; SI-NEXT: flat_store_dwordx4 v[13:14], v[9:12]
-; SI-NEXT: v_cvt_f32_f16_e32 v6, v0
-; SI-NEXT: v_cvt_f32_f16_e32 v12, v3
-; SI-NEXT: v_lshrrev_b32_e32 v3, 16, v3
-; SI-NEXT: v_lshrrev_b32_e32 v7, 16, v0
-; SI-NEXT: v_cvt_f32_f16_e32 v10, v2
-; SI-NEXT: v_lshrrev_b32_e32 v11, 16, v2
-; SI-NEXT: v_cvt_f32_f16_e32 v2, v5
-; SI-NEXT: v_cvt_f32_f16_e32 v0, v4
-; SI-NEXT: v_mov_b32_e32 v5, s1
-; SI-NEXT: v_cvt_f32_f16_e32 v9, v1
-; SI-NEXT: v_cvt_f32_f16_e32 v13, v3
-; SI-NEXT: v_cvt_f32_f16_e32 v3, v16
-; SI-NEXT: v_cvt_f32_f16_e32 v1, v17
-; SI-NEXT: v_mov_b32_e32 v4, s0
-; SI-NEXT: s_add_u32 s0, s0, 32
-; SI-NEXT: v_cvt_f32_f16_e32 v11, v11
-; SI-NEXT: s_addc_u32 s1, s1, 0
-; SI-NEXT: v_cvt_f32_f16_e32 v7, v7
-; SI-NEXT: v_mov_b32_e32 v15, s3
-; SI-NEXT: v_mov_b32_e32 v17, s1
-; SI-NEXT: v_mov_b32_e32 v14, s2
-; SI-NEXT: v_mov_b32_e32 v16, s0
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: flat_store_dwordx4 v[14:15], v[10:13]
-; SI-NEXT: flat_store_dwordx4 v[16:17], v[6:9]
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_extload_v16f16_to_v16f32:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: s_add_u32 s4, s2, 16
+; CI-NEXT: v_mov_b32_e32 v5, s3
+; CI-NEXT: s_addc_u32 s5, s3, 0
+; CI-NEXT: v_mov_b32_e32 v0, s4
+; CI-NEXT: v_mov_b32_e32 v4, s2
+; CI-NEXT: v_mov_b32_e32 v1, s5
+; CI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
+; CI-NEXT: flat_load_dwordx4 v[4:7], v[4:5]
+; CI-NEXT: s_add_u32 s2, s0, 16
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: v_mov_b32_e32 v14, s3
+; CI-NEXT: v_mov_b32_e32 v13, s2
+; CI-NEXT: s_add_u32 s2, s0, 48
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: s_waitcnt vmcnt(1)
+; CI-NEXT: v_cvt_f32_f16_e32 v8, v1
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_cvt_f32_f16_e32 v11, v7
+; CI-NEXT: v_cvt_f32_f16_e32 v9, v6
+; CI-NEXT: v_lshrrev_b32_e32 v7, 16, v7
+; CI-NEXT: v_lshrrev_b32_e32 v6, 16, v6
+; CI-NEXT: v_cvt_f32_f16_e32 v12, v7
+; CI-NEXT: v_cvt_f32_f16_e32 v10, v6
+; CI-NEXT: v_lshrrev_b32_e32 v1, 16, v1
+; CI-NEXT: v_lshrrev_b32_e32 v16, 16, v5
+; CI-NEXT: v_lshrrev_b32_e32 v17, 16, v4
+; CI-NEXT: flat_store_dwordx4 v[13:14], v[9:12]
+; CI-NEXT: v_cvt_f32_f16_e32 v6, v0
+; CI-NEXT: v_cvt_f32_f16_e32 v12, v3
+; CI-NEXT: v_lshrrev_b32_e32 v3, 16, v3
+; CI-NEXT: v_lshrrev_b32_e32 v7, 16, v0
+; CI-NEXT: v_cvt_f32_f16_e32 v10, v2
+; CI-NEXT: v_lshrrev_b32_e32 v11, 16, v2
+; CI-NEXT: v_cvt_f32_f16_e32 v2, v5
+; CI-NEXT: v_cvt_f32_f16_e32 v0, v4
+; CI-NEXT: v_mov_b32_e32 v5, s1
+; CI-NEXT: v_cvt_f32_f16_e32 v9, v1
+; CI-NEXT: v_cvt_f32_f16_e32 v13, v3
+; CI-NEXT: v_cvt_f32_f16_e32 v3, v16
+; CI-NEXT: v_cvt_f32_f16_e32 v1, v17
+; CI-NEXT: v_mov_b32_e32 v4, s0
+; CI-NEXT: s_add_u32 s0, s0, 32
+; CI-NEXT: v_cvt_f32_f16_e32 v11, v11
+; CI-NEXT: s_addc_u32 s1, s1, 0
+; CI-NEXT: v_cvt_f32_f16_e32 v7, v7
+; CI-NEXT: v_mov_b32_e32 v15, s3
+; CI-NEXT: v_mov_b32_e32 v17, s1
+; CI-NEXT: v_mov_b32_e32 v14, s2
+; CI-NEXT: v_mov_b32_e32 v16, s0
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: flat_store_dwordx4 v[14:15], v[10:13]
+; CI-NEXT: flat_store_dwordx4 v[16:17], v[6:9]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_extload_v16f16_to_v16f32:
; VI: ; %bb.0:
@@ -1057,23 +1057,23 @@ define amdgpu_kernel void @global_extload_f16_to_f64(double addrspace(1)* %out,
}
define amdgpu_kernel void @global_extload_v2f16_to_v2f64(<2 x double> addrspace(1)* %out, <2 x half> addrspace(1)* %in) #0 {
-; SI-LABEL: global_extload_v2f16_to_v2f64:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_load_dword v0, v[0:1]
-; SI-NEXT: v_mov_b32_e32 v4, s0
-; SI-NEXT: v_mov_b32_e32 v5, s1
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_lshrrev_b32_e32 v1, 16, v0
-; SI-NEXT: v_cvt_f32_f16_e32 v0, v0
-; SI-NEXT: v_cvt_f32_f16_e32 v2, v1
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v0
-; SI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_extload_v2f16_to_v2f64:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_load_dword v0, v[0:1]
+; CI-NEXT: v_mov_b32_e32 v4, s0
+; CI-NEXT: v_mov_b32_e32 v5, s1
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_lshrrev_b32_e32 v1, 16, v0
+; CI-NEXT: v_cvt_f32_f16_e32 v0, v0
+; CI-NEXT: v_cvt_f32_f16_e32 v2, v1
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v0
+; CI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_extload_v2f16_to_v2f64:
; VI: ; %bb.0:
@@ -1098,30 +1098,30 @@ define amdgpu_kernel void @global_extload_v2f16_to_v2f64(<2 x double> addrspace(
}
define amdgpu_kernel void @global_extload_v3f16_to_v3f64(<3 x double> addrspace(1)* %out, <3 x half> addrspace(1)* %in) #0 {
-; SI-LABEL: global_extload_v3f16_to_v3f64:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_load_dwordx2 v[0:1], v[0:1]
-; SI-NEXT: s_add_u32 s2, s0, 16
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: v_mov_b32_e32 v7, s3
-; SI-NEXT: v_mov_b32_e32 v6, s2
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_cvt_f32_f16_e32 v1, v1
-; SI-NEXT: v_lshrrev_b32_e32 v2, 16, v0
-; SI-NEXT: v_cvt_f32_f16_e32 v0, v0
-; SI-NEXT: v_cvt_f32_f16_e32 v2, v2
-; SI-NEXT: v_cvt_f64_f32_e32 v[4:5], v1
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v0
-; SI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
-; SI-NEXT: flat_store_dwordx2 v[6:7], v[4:5]
-; SI-NEXT: v_mov_b32_e32 v5, s1
-; SI-NEXT: v_mov_b32_e32 v4, s0
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_extload_v3f16_to_v3f64:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_load_dwordx2 v[0:1], v[0:1]
+; CI-NEXT: s_add_u32 s2, s0, 16
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: v_mov_b32_e32 v7, s3
+; CI-NEXT: v_mov_b32_e32 v6, s2
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_cvt_f32_f16_e32 v1, v1
+; CI-NEXT: v_lshrrev_b32_e32 v2, 16, v0
+; CI-NEXT: v_cvt_f32_f16_e32 v0, v0
+; CI-NEXT: v_cvt_f32_f16_e32 v2, v2
+; CI-NEXT: v_cvt_f64_f32_e32 v[4:5], v1
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v0
+; CI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
+; CI-NEXT: flat_store_dwordx2 v[6:7], v[4:5]
+; CI-NEXT: v_mov_b32_e32 v5, s1
+; CI-NEXT: v_mov_b32_e32 v4, s0
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_extload_v3f16_to_v3f64:
; VI: ; %bb.0:
@@ -1153,33 +1153,33 @@ define amdgpu_kernel void @global_extload_v3f16_to_v3f64(<3 x double> addrspace(
}
define amdgpu_kernel void @global_extload_v4f16_to_v4f64(<4 x double> addrspace(1)* %out, <4 x half> addrspace(1)* %in) #0 {
-; SI-LABEL: global_extload_v4f16_to_v4f64:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_load_dwordx2 v[0:1], v[0:1]
-; SI-NEXT: s_add_u32 s2, s0, 16
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: v_mov_b32_e32 v9, s1
-; SI-NEXT: v_mov_b32_e32 v8, s0
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_lshrrev_b32_e32 v2, 16, v1
-; SI-NEXT: v_cvt_f32_f16_e32 v3, v1
-; SI-NEXT: v_cvt_f32_f16_e32 v1, v0
-; SI-NEXT: v_lshrrev_b32_e32 v0, 16, v0
-; SI-NEXT: v_cvt_f32_f16_e32 v2, v2
-; SI-NEXT: v_cvt_f32_f16_e32 v10, v0
-; SI-NEXT: v_cvt_f64_f32_e32 v[4:5], v3
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v1
-; SI-NEXT: v_cvt_f64_f32_e32 v[6:7], v2
-; SI-NEXT: v_cvt_f64_f32_e32 v[2:3], v10
-; SI-NEXT: v_mov_b32_e32 v11, s3
-; SI-NEXT: v_mov_b32_e32 v10, s2
-; SI-NEXT: flat_store_dwordx4 v[10:11], v[4:7]
-; SI-NEXT: flat_store_dwordx4 v[8:9], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_extload_v4f16_to_v4f64:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_load_dwordx2 v[0:1], v[0:1]
+; CI-NEXT: s_add_u32 s2, s0, 16
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: v_mov_b32_e32 v9, s1
+; CI-NEXT: v_mov_b32_e32 v8, s0
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_lshrrev_b32_e32 v2, 16, v1
+; CI-NEXT: v_cvt_f32_f16_e32 v3, v1
+; CI-NEXT: v_cvt_f32_f16_e32 v1, v0
+; CI-NEXT: v_lshrrev_b32_e32 v0, 16, v0
+; CI-NEXT: v_cvt_f32_f16_e32 v2, v2
+; CI-NEXT: v_cvt_f32_f16_e32 v10, v0
+; CI-NEXT: v_cvt_f64_f32_e32 v[4:5], v3
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v1
+; CI-NEXT: v_cvt_f64_f32_e32 v[6:7], v2
+; CI-NEXT: v_cvt_f64_f32_e32 v[2:3], v10
+; CI-NEXT: v_mov_b32_e32 v11, s3
+; CI-NEXT: v_mov_b32_e32 v10, s2
+; CI-NEXT: flat_store_dwordx4 v[10:11], v[4:7]
+; CI-NEXT: flat_store_dwordx4 v[8:9], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_extload_v4f16_to_v4f64:
; VI: ; %bb.0:
@@ -1213,53 +1213,53 @@ define amdgpu_kernel void @global_extload_v4f16_to_v4f64(<4 x double> addrspace(
}
define amdgpu_kernel void @global_extload_v8f16_to_v8f64(<8 x double> addrspace(1)* %out, <8 x half> addrspace(1)* %in) #0 {
-; SI-LABEL: global_extload_v8f16_to_v8f64:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
-; SI-NEXT: s_add_u32 s2, s0, 48
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: v_mov_b32_e32 v7, s3
-; SI-NEXT: v_mov_b32_e32 v6, s2
-; SI-NEXT: s_add_u32 s2, s0, 32
-; SI-NEXT: v_mov_b32_e32 v13, s1
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: v_mov_b32_e32 v12, s0
-; SI-NEXT: s_add_u32 s0, s0, 16
-; SI-NEXT: v_mov_b32_e32 v15, s3
-; SI-NEXT: s_addc_u32 s1, s1, 0
-; SI-NEXT: v_mov_b32_e32 v14, s2
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_lshrrev_b32_e32 v4, 16, v3
-; SI-NEXT: v_cvt_f32_f16_e32 v3, v3
-; SI-NEXT: v_lshrrev_b32_e32 v5, 16, v2
-; SI-NEXT: v_cvt_f32_f16_e32 v8, v2
-; SI-NEXT: v_cvt_f32_f16_e32 v2, v4
-; SI-NEXT: v_lshrrev_b32_e32 v9, 16, v1
-; SI-NEXT: v_cvt_f32_f16_e32 v10, v1
-; SI-NEXT: v_lshrrev_b32_e32 v11, 16, v0
-; SI-NEXT: v_cvt_f32_f16_e32 v4, v0
-; SI-NEXT: v_cvt_f32_f16_e32 v16, v5
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v3
-; SI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
-; SI-NEXT: v_cvt_f32_f16_e32 v17, v9
-; SI-NEXT: v_cvt_f32_f16_e32 v18, v11
-; SI-NEXT: v_cvt_f64_f32_e32 v[8:9], v8
-; SI-NEXT: flat_store_dwordx4 v[6:7], v[0:3]
-; SI-NEXT: v_cvt_f64_f32_e32 v[4:5], v4
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v10
-; SI-NEXT: v_cvt_f64_f32_e32 v[10:11], v16
-; SI-NEXT: v_cvt_f64_f32_e32 v[2:3], v17
-; SI-NEXT: v_cvt_f64_f32_e32 v[6:7], v18
-; SI-NEXT: v_mov_b32_e32 v17, s1
-; SI-NEXT: v_mov_b32_e32 v16, s0
-; SI-NEXT: flat_store_dwordx4 v[14:15], v[8:11]
-; SI-NEXT: flat_store_dwordx4 v[16:17], v[0:3]
-; SI-NEXT: flat_store_dwordx4 v[12:13], v[4:7]
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_extload_v8f16_to_v8f64:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
+; CI-NEXT: s_add_u32 s2, s0, 48
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: v_mov_b32_e32 v7, s3
+; CI-NEXT: v_mov_b32_e32 v6, s2
+; CI-NEXT: s_add_u32 s2, s0, 32
+; CI-NEXT: v_mov_b32_e32 v13, s1
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: v_mov_b32_e32 v12, s0
+; CI-NEXT: s_add_u32 s0, s0, 16
+; CI-NEXT: v_mov_b32_e32 v15, s3
+; CI-NEXT: s_addc_u32 s1, s1, 0
+; CI-NEXT: v_mov_b32_e32 v14, s2
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_lshrrev_b32_e32 v4, 16, v3
+; CI-NEXT: v_cvt_f32_f16_e32 v3, v3
+; CI-NEXT: v_lshrrev_b32_e32 v5, 16, v2
+; CI-NEXT: v_cvt_f32_f16_e32 v8, v2
+; CI-NEXT: v_cvt_f32_f16_e32 v2, v4
+; CI-NEXT: v_lshrrev_b32_e32 v9, 16, v1
+; CI-NEXT: v_cvt_f32_f16_e32 v10, v1
+; CI-NEXT: v_lshrrev_b32_e32 v11, 16, v0
+; CI-NEXT: v_cvt_f32_f16_e32 v4, v0
+; CI-NEXT: v_cvt_f32_f16_e32 v16, v5
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v3
+; CI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
+; CI-NEXT: v_cvt_f32_f16_e32 v17, v9
+; CI-NEXT: v_cvt_f32_f16_e32 v18, v11
+; CI-NEXT: v_cvt_f64_f32_e32 v[8:9], v8
+; CI-NEXT: flat_store_dwordx4 v[6:7], v[0:3]
+; CI-NEXT: v_cvt_f64_f32_e32 v[4:5], v4
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v10
+; CI-NEXT: v_cvt_f64_f32_e32 v[10:11], v16
+; CI-NEXT: v_cvt_f64_f32_e32 v[2:3], v17
+; CI-NEXT: v_cvt_f64_f32_e32 v[6:7], v18
+; CI-NEXT: v_mov_b32_e32 v17, s1
+; CI-NEXT: v_mov_b32_e32 v16, s0
+; CI-NEXT: flat_store_dwordx4 v[14:15], v[8:11]
+; CI-NEXT: flat_store_dwordx4 v[16:17], v[0:3]
+; CI-NEXT: flat_store_dwordx4 v[12:13], v[4:7]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_extload_v8f16_to_v8f64:
; VI: ; %bb.0:
@@ -1311,99 +1311,99 @@ define amdgpu_kernel void @global_extload_v8f16_to_v8f64(<8 x double> addrspace(
}
define amdgpu_kernel void @global_extload_v16f16_to_v16f64(<16 x double> addrspace(1)* %out, <16 x half> addrspace(1)* %in) #0 {
-; SI-LABEL: global_extload_v16f16_to_v16f64:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
-; SI-NEXT: s_add_u32 s2, s2, 16
-; SI-NEXT: s_addc_u32 s3, s3, 0
-; SI-NEXT: v_mov_b32_e32 v5, s3
-; SI-NEXT: v_mov_b32_e32 v4, s2
-; SI-NEXT: flat_load_dwordx4 v[4:7], v[4:5]
-; SI-NEXT: s_add_u32 s2, s0, 48
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: v_mov_b32_e32 v15, s3
-; SI-NEXT: v_mov_b32_e32 v14, s2
-; SI-NEXT: s_add_u32 s2, s0, 32
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: v_mov_b32_e32 v17, s3
-; SI-NEXT: v_mov_b32_e32 v16, s2
-; SI-NEXT: s_add_u32 s2, s0, 16
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: v_mov_b32_e32 v19, s3
-; SI-NEXT: v_mov_b32_e32 v18, s2
-; SI-NEXT: s_add_u32 s2, s0, 0x70
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: v_mov_b32_e32 v13, s1
-; SI-NEXT: v_mov_b32_e32 v12, s0
-; SI-NEXT: s_waitcnt vmcnt(1)
-; SI-NEXT: v_lshrrev_b32_e32 v8, 16, v3
-; SI-NEXT: v_cvt_f32_f16_e32 v3, v3
-; SI-NEXT: v_cvt_f32_f16_e32 v10, v8
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_lshrrev_b32_e32 v20, 16, v5
-; SI-NEXT: v_cvt_f64_f32_e32 v[8:9], v3
-; SI-NEXT: v_lshrrev_b32_e32 v3, 16, v2
-; SI-NEXT: v_cvt_f64_f32_e32 v[10:11], v10
-; SI-NEXT: v_cvt_f32_f16_e32 v2, v2
-; SI-NEXT: v_cvt_f32_f16_e32 v3, v3
-; SI-NEXT: v_cvt_f32_f16_e32 v21, v5
-; SI-NEXT: flat_store_dwordx4 v[14:15], v[8:11]
-; SI-NEXT: v_mov_b32_e32 v15, s3
-; SI-NEXT: v_cvt_f64_f32_e32 v[8:9], v2
-; SI-NEXT: v_cvt_f64_f32_e32 v[10:11], v3
-; SI-NEXT: v_lshrrev_b32_e32 v2, 16, v1
-; SI-NEXT: v_cvt_f32_f16_e32 v1, v1
-; SI-NEXT: v_cvt_f32_f16_e32 v2, v2
-; SI-NEXT: flat_store_dwordx4 v[16:17], v[8:11]
-; SI-NEXT: v_mov_b32_e32 v14, s2
-; SI-NEXT: v_lshrrev_b32_e32 v8, 16, v0
-; SI-NEXT: v_cvt_f32_f16_e32 v9, v0
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v1
-; SI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
-; SI-NEXT: v_cvt_f32_f16_e32 v8, v8
-; SI-NEXT: v_lshrrev_b32_e32 v10, 16, v7
-; SI-NEXT: v_cvt_f32_f16_e32 v7, v7
-; SI-NEXT: flat_store_dwordx4 v[18:19], v[0:3]
-; SI-NEXT: v_lshrrev_b32_e32 v11, 16, v6
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v9
-; SI-NEXT: v_cvt_f64_f32_e32 v[2:3], v8
-; SI-NEXT: v_cvt_f32_f16_e32 v8, v10
-; SI-NEXT: s_add_u32 s2, s0, 0x60
-; SI-NEXT: v_cvt_f32_f16_e32 v6, v6
-; SI-NEXT: v_cvt_f32_f16_e32 v10, v11
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: v_lshrrev_b32_e32 v5, 16, v4
-; SI-NEXT: flat_store_dwordx4 v[12:13], v[0:3]
-; SI-NEXT: v_mov_b32_e32 v17, s3
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v7
-; SI-NEXT: v_cvt_f64_f32_e32 v[2:3], v8
-; SI-NEXT: v_cvt_f32_f16_e32 v7, v20
-; SI-NEXT: v_cvt_f32_f16_e32 v4, v4
-; SI-NEXT: v_cvt_f32_f16_e32 v12, v5
-; SI-NEXT: v_mov_b32_e32 v16, s2
-; SI-NEXT: s_add_u32 s2, s0, 0x50
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: v_cvt_f64_f32_e32 v[8:9], v6
-; SI-NEXT: v_cvt_f64_f32_e32 v[10:11], v10
-; SI-NEXT: s_add_u32 s0, s0, 64
-; SI-NEXT: flat_store_dwordx4 v[14:15], v[0:3]
-; SI-NEXT: s_addc_u32 s1, s1, 0
-; SI-NEXT: v_cvt_f64_f32_e32 v[0:1], v21
-; SI-NEXT: v_cvt_f64_f32_e32 v[2:3], v7
-; SI-NEXT: v_cvt_f64_f32_e32 v[4:5], v4
-; SI-NEXT: v_cvt_f64_f32_e32 v[6:7], v12
-; SI-NEXT: v_mov_b32_e32 v19, s3
-; SI-NEXT: v_mov_b32_e32 v13, s1
-; SI-NEXT: v_mov_b32_e32 v18, s2
-; SI-NEXT: v_mov_b32_e32 v12, s0
-; SI-NEXT: flat_store_dwordx4 v[16:17], v[8:11]
-; SI-NEXT: flat_store_dwordx4 v[18:19], v[0:3]
-; SI-NEXT: flat_store_dwordx4 v[12:13], v[4:7]
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_extload_v16f16_to_v16f64:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
+; CI-NEXT: s_add_u32 s2, s2, 16
+; CI-NEXT: s_addc_u32 s3, s3, 0
+; CI-NEXT: v_mov_b32_e32 v5, s3
+; CI-NEXT: v_mov_b32_e32 v4, s2
+; CI-NEXT: flat_load_dwordx4 v[4:7], v[4:5]
+; CI-NEXT: s_add_u32 s2, s0, 48
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: v_mov_b32_e32 v15, s3
+; CI-NEXT: v_mov_b32_e32 v14, s2
+; CI-NEXT: s_add_u32 s2, s0, 32
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: v_mov_b32_e32 v17, s3
+; CI-NEXT: v_mov_b32_e32 v16, s2
+; CI-NEXT: s_add_u32 s2, s0, 16
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: v_mov_b32_e32 v19, s3
+; CI-NEXT: v_mov_b32_e32 v18, s2
+; CI-NEXT: s_add_u32 s2, s0, 0x70
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: v_mov_b32_e32 v13, s1
+; CI-NEXT: v_mov_b32_e32 v12, s0
+; CI-NEXT: s_waitcnt vmcnt(1)
+; CI-NEXT: v_lshrrev_b32_e32 v8, 16, v3
+; CI-NEXT: v_cvt_f32_f16_e32 v3, v3
+; CI-NEXT: v_cvt_f32_f16_e32 v10, v8
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_lshrrev_b32_e32 v20, 16, v5
+; CI-NEXT: v_cvt_f64_f32_e32 v[8:9], v3
+; CI-NEXT: v_lshrrev_b32_e32 v3, 16, v2
+; CI-NEXT: v_cvt_f64_f32_e32 v[10:11], v10
+; CI-NEXT: v_cvt_f32_f16_e32 v2, v2
+; CI-NEXT: v_cvt_f32_f16_e32 v3, v3
+; CI-NEXT: v_cvt_f32_f16_e32 v21, v5
+; CI-NEXT: flat_store_dwordx4 v[14:15], v[8:11]
+; CI-NEXT: v_mov_b32_e32 v15, s3
+; CI-NEXT: v_cvt_f64_f32_e32 v[8:9], v2
+; CI-NEXT: v_cvt_f64_f32_e32 v[10:11], v3
+; CI-NEXT: v_lshrrev_b32_e32 v2, 16, v1
+; CI-NEXT: v_cvt_f32_f16_e32 v1, v1
+; CI-NEXT: v_cvt_f32_f16_e32 v2, v2
+; CI-NEXT: flat_store_dwordx4 v[16:17], v[8:11]
+; CI-NEXT: v_mov_b32_e32 v14, s2
+; CI-NEXT: v_lshrrev_b32_e32 v8, 16, v0
+; CI-NEXT: v_cvt_f32_f16_e32 v9, v0
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v1
+; CI-NEXT: v_cvt_f64_f32_e32 v[2:3], v2
+; CI-NEXT: v_cvt_f32_f16_e32 v8, v8
+; CI-NEXT: v_lshrrev_b32_e32 v10, 16, v7
+; CI-NEXT: v_cvt_f32_f16_e32 v7, v7
+; CI-NEXT: flat_store_dwordx4 v[18:19], v[0:3]
+; CI-NEXT: v_lshrrev_b32_e32 v11, 16, v6
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v9
+; CI-NEXT: v_cvt_f64_f32_e32 v[2:3], v8
+; CI-NEXT: v_cvt_f32_f16_e32 v8, v10
+; CI-NEXT: s_add_u32 s2, s0, 0x60
+; CI-NEXT: v_cvt_f32_f16_e32 v6, v6
+; CI-NEXT: v_cvt_f32_f16_e32 v10, v11
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: v_lshrrev_b32_e32 v5, 16, v4
+; CI-NEXT: flat_store_dwordx4 v[12:13], v[0:3]
+; CI-NEXT: v_mov_b32_e32 v17, s3
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v7
+; CI-NEXT: v_cvt_f64_f32_e32 v[2:3], v8
+; CI-NEXT: v_cvt_f32_f16_e32 v7, v20
+; CI-NEXT: v_cvt_f32_f16_e32 v4, v4
+; CI-NEXT: v_cvt_f32_f16_e32 v12, v5
+; CI-NEXT: v_mov_b32_e32 v16, s2
+; CI-NEXT: s_add_u32 s2, s0, 0x50
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: v_cvt_f64_f32_e32 v[8:9], v6
+; CI-NEXT: v_cvt_f64_f32_e32 v[10:11], v10
+; CI-NEXT: s_add_u32 s0, s0, 64
+; CI-NEXT: flat_store_dwordx4 v[14:15], v[0:3]
+; CI-NEXT: s_addc_u32 s1, s1, 0
+; CI-NEXT: v_cvt_f64_f32_e32 v[0:1], v21
+; CI-NEXT: v_cvt_f64_f32_e32 v[2:3], v7
+; CI-NEXT: v_cvt_f64_f32_e32 v[4:5], v4
+; CI-NEXT: v_cvt_f64_f32_e32 v[6:7], v12
+; CI-NEXT: v_mov_b32_e32 v19, s3
+; CI-NEXT: v_mov_b32_e32 v13, s1
+; CI-NEXT: v_mov_b32_e32 v18, s2
+; CI-NEXT: v_mov_b32_e32 v12, s0
+; CI-NEXT: flat_store_dwordx4 v[16:17], v[8:11]
+; CI-NEXT: flat_store_dwordx4 v[18:19], v[0:3]
+; CI-NEXT: flat_store_dwordx4 v[12:13], v[4:7]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_extload_v16f16_to_v16f64:
; VI: ; %bb.0:
@@ -1518,22 +1518,22 @@ define amdgpu_kernel void @global_truncstore_f32_to_f16(half addrspace(1)* %out,
}
define amdgpu_kernel void @global_truncstore_v2f32_to_v2f16(<2 x half> addrspace(1)* %out, <2 x float> addrspace(1)* %in) #0 {
-; SI-LABEL: global_truncstore_v2f32_to_v2f16:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_load_dwordx2 v[0:1], v[0:1]
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_cvt_f16_f32_e32 v2, v1
-; SI-NEXT: v_cvt_f16_f32_e32 v3, v0
-; SI-NEXT: v_mov_b32_e32 v0, s0
-; SI-NEXT: v_mov_b32_e32 v1, s1
-; SI-NEXT: v_lshlrev_b32_e32 v2, 16, v2
-; SI-NEXT: v_or_b32_e32 v2, v3, v2
-; SI-NEXT: flat_store_dword v[0:1], v2
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_truncstore_v2f32_to_v2f16:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_load_dwordx2 v[0:1], v[0:1]
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_cvt_f16_f32_e32 v2, v1
+; CI-NEXT: v_cvt_f16_f32_e32 v3, v0
+; CI-NEXT: v_mov_b32_e32 v0, s0
+; CI-NEXT: v_mov_b32_e32 v1, s1
+; CI-NEXT: v_lshlrev_b32_e32 v2, 16, v2
+; CI-NEXT: v_or_b32_e32 v2, v3, v2
+; CI-NEXT: flat_store_dword v[0:1], v2
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_truncstore_v2f32_to_v2f16:
; VI: ; %bb.0:
@@ -1557,28 +1557,28 @@ define amdgpu_kernel void @global_truncstore_v2f32_to_v2f16(<2 x half> addrspace
}
define amdgpu_kernel void @global_truncstore_v3f32_to_v3f16(<3 x half> addrspace(1)* %out, <3 x float> addrspace(1)* %in) #0 {
-; SI-LABEL: global_truncstore_v3f32_to_v3f16:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_load_dwordx3 v[0:2], v[0:1]
-; SI-NEXT: s_add_u32 s2, s0, 4
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_cvt_f16_f32_e32 v3, v1
-; SI-NEXT: v_cvt_f16_f32_e32 v2, v2
-; SI-NEXT: v_cvt_f16_f32_e32 v4, v0
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: v_lshlrev_b32_e32 v3, 16, v3
-; SI-NEXT: flat_store_short v[0:1], v2
-; SI-NEXT: v_mov_b32_e32 v0, s0
-; SI-NEXT: v_or_b32_e32 v2, v4, v3
-; SI-NEXT: v_mov_b32_e32 v1, s1
-; SI-NEXT: flat_store_dword v[0:1], v2
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_truncstore_v3f32_to_v3f16:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_load_dwordx3 v[0:2], v[0:1]
+; CI-NEXT: s_add_u32 s2, s0, 4
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_cvt_f16_f32_e32 v3, v1
+; CI-NEXT: v_cvt_f16_f32_e32 v2, v2
+; CI-NEXT: v_cvt_f16_f32_e32 v4, v0
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: v_lshlrev_b32_e32 v3, 16, v3
+; CI-NEXT: flat_store_short v[0:1], v2
+; CI-NEXT: v_mov_b32_e32 v0, s0
+; CI-NEXT: v_or_b32_e32 v2, v4, v3
+; CI-NEXT: v_mov_b32_e32 v1, s1
+; CI-NEXT: flat_store_dword v[0:1], v2
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_truncstore_v3f32_to_v3f16:
; VI: ; %bb.0:
@@ -1608,26 +1608,26 @@ define amdgpu_kernel void @global_truncstore_v3f32_to_v3f16(<3 x half> addrspace
}
define amdgpu_kernel void @global_truncstore_v4f32_to_v4f16(<4 x half> addrspace(1)* %out, <4 x float> addrspace(1)* %in) #0 {
-; SI-LABEL: global_truncstore_v4f32_to_v4f16:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
-; SI-NEXT: v_mov_b32_e32 v4, s0
-; SI-NEXT: v_mov_b32_e32 v5, s1
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_cvt_f16_f32_e32 v3, v3
-; SI-NEXT: v_cvt_f16_f32_e32 v1, v1
-; SI-NEXT: v_cvt_f16_f32_e32 v2, v2
-; SI-NEXT: v_cvt_f16_f32_e32 v0, v0
-; SI-NEXT: v_lshlrev_b32_e32 v3, 16, v3
-; SI-NEXT: v_lshlrev_b32_e32 v6, 16, v1
-; SI-NEXT: v_or_b32_e32 v1, v2, v3
-; SI-NEXT: v_or_b32_e32 v0, v0, v6
-; SI-NEXT: flat_store_dwordx2 v[4:5], v[0:1]
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_truncstore_v4f32_to_v4f16:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
+; CI-NEXT: v_mov_b32_e32 v4, s0
+; CI-NEXT: v_mov_b32_e32 v5, s1
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_cvt_f16_f32_e32 v3, v3
+; CI-NEXT: v_cvt_f16_f32_e32 v1, v1
+; CI-NEXT: v_cvt_f16_f32_e32 v2, v2
+; CI-NEXT: v_cvt_f16_f32_e32 v0, v0
+; CI-NEXT: v_lshlrev_b32_e32 v3, 16, v3
+; CI-NEXT: v_lshlrev_b32_e32 v6, 16, v1
+; CI-NEXT: v_or_b32_e32 v1, v2, v3
+; CI-NEXT: v_or_b32_e32 v0, v0, v6
+; CI-NEXT: flat_store_dwordx2 v[4:5], v[0:1]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_truncstore_v4f32_to_v4f16:
; VI: ; %bb.0:
@@ -1654,40 +1654,40 @@ define amdgpu_kernel void @global_truncstore_v4f32_to_v4f16(<4 x half> addrspace
}
define amdgpu_kernel void @global_truncstore_v8f32_to_v8f16(<8 x half> addrspace(1)* %out, <8 x float> addrspace(1)* %in) #0 {
-; SI-LABEL: global_truncstore_v8f32_to_v8f16:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: s_add_u32 s2, s2, 16
-; SI-NEXT: s_addc_u32 s3, s3, 0
-; SI-NEXT: v_mov_b32_e32 v5, s3
-; SI-NEXT: v_mov_b32_e32 v4, s2
-; SI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
-; SI-NEXT: flat_load_dwordx4 v[4:7], v[4:5]
-; SI-NEXT: v_mov_b32_e32 v8, s0
-; SI-NEXT: v_mov_b32_e32 v9, s1
-; SI-NEXT: s_waitcnt vmcnt(1)
-; SI-NEXT: v_cvt_f16_f32_e32 v3, v3
-; SI-NEXT: v_cvt_f16_f32_e32 v1, v1
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_cvt_f16_f32_e32 v7, v7
-; SI-NEXT: v_cvt_f16_f32_e32 v5, v5
-; SI-NEXT: v_cvt_f16_f32_e32 v2, v2
-; SI-NEXT: v_cvt_f16_f32_e32 v0, v0
-; SI-NEXT: v_cvt_f16_f32_e32 v6, v6
-; SI-NEXT: v_cvt_f16_f32_e32 v4, v4
-; SI-NEXT: v_lshlrev_b32_e32 v3, 16, v3
-; SI-NEXT: v_lshlrev_b32_e32 v10, 16, v1
-; SI-NEXT: v_lshlrev_b32_e32 v7, 16, v7
-; SI-NEXT: v_lshlrev_b32_e32 v5, 16, v5
-; SI-NEXT: v_or_b32_e32 v1, v2, v3
-; SI-NEXT: v_or_b32_e32 v0, v0, v10
-; SI-NEXT: v_or_b32_e32 v3, v6, v7
-; SI-NEXT: v_or_b32_e32 v2, v4, v5
-; SI-NEXT: flat_store_dwordx4 v[8:9], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_truncstore_v8f32_to_v8f16:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: s_add_u32 s2, s2, 16
+; CI-NEXT: s_addc_u32 s3, s3, 0
+; CI-NEXT: v_mov_b32_e32 v5, s3
+; CI-NEXT: v_mov_b32_e32 v4, s2
+; CI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
+; CI-NEXT: flat_load_dwordx4 v[4:7], v[4:5]
+; CI-NEXT: v_mov_b32_e32 v8, s0
+; CI-NEXT: v_mov_b32_e32 v9, s1
+; CI-NEXT: s_waitcnt vmcnt(1)
+; CI-NEXT: v_cvt_f16_f32_e32 v3, v3
+; CI-NEXT: v_cvt_f16_f32_e32 v1, v1
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_cvt_f16_f32_e32 v7, v7
+; CI-NEXT: v_cvt_f16_f32_e32 v5, v5
+; CI-NEXT: v_cvt_f16_f32_e32 v2, v2
+; CI-NEXT: v_cvt_f16_f32_e32 v0, v0
+; CI-NEXT: v_cvt_f16_f32_e32 v6, v6
+; CI-NEXT: v_cvt_f16_f32_e32 v4, v4
+; CI-NEXT: v_lshlrev_b32_e32 v3, 16, v3
+; CI-NEXT: v_lshlrev_b32_e32 v10, 16, v1
+; CI-NEXT: v_lshlrev_b32_e32 v7, 16, v7
+; CI-NEXT: v_lshlrev_b32_e32 v5, 16, v5
+; CI-NEXT: v_or_b32_e32 v1, v2, v3
+; CI-NEXT: v_or_b32_e32 v0, v0, v10
+; CI-NEXT: v_or_b32_e32 v3, v6, v7
+; CI-NEXT: v_or_b32_e32 v2, v4, v5
+; CI-NEXT: flat_store_dwordx4 v[8:9], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_truncstore_v8f32_to_v8f16:
; VI: ; %bb.0:
@@ -1726,73 +1726,73 @@ define amdgpu_kernel void @global_truncstore_v8f32_to_v8f16(<8 x half> addrspace
}
define amdgpu_kernel void @global_truncstore_v16f32_to_v16f16(<16 x half> addrspace(1)* %out, <16 x float> addrspace(1)* %in) #0 {
-; SI-LABEL: global_truncstore_v16f32_to_v16f16:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: s_add_u32 s4, s2, 32
-; SI-NEXT: s_addc_u32 s5, s3, 0
-; SI-NEXT: v_mov_b32_e32 v0, s4
-; SI-NEXT: v_mov_b32_e32 v1, s5
-; SI-NEXT: s_add_u32 s4, s2, 48
-; SI-NEXT: s_addc_u32 s5, s3, 0
-; SI-NEXT: v_mov_b32_e32 v9, s3
-; SI-NEXT: v_mov_b32_e32 v4, s4
-; SI-NEXT: v_mov_b32_e32 v8, s2
-; SI-NEXT: s_add_u32 s2, s2, 16
-; SI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
-; SI-NEXT: v_mov_b32_e32 v5, s5
-; SI-NEXT: flat_load_dwordx4 v[4:7], v[4:5]
-; SI-NEXT: s_addc_u32 s3, s3, 0
-; SI-NEXT: v_mov_b32_e32 v13, s3
-; SI-NEXT: v_mov_b32_e32 v12, s2
-; SI-NEXT: flat_load_dwordx4 v[8:11], v[8:9]
-; SI-NEXT: flat_load_dwordx4 v[12:15], v[12:13]
-; SI-NEXT: s_add_u32 s2, s0, 16
-; SI-NEXT: s_addc_u32 s3, s1, 0
-; SI-NEXT: s_waitcnt vmcnt(3)
-; SI-NEXT: v_cvt_f16_f32_e32 v3, v3
-; SI-NEXT: v_cvt_f16_f32_e32 v2, v2
-; SI-NEXT: v_cvt_f16_f32_e32 v1, v1
-; SI-NEXT: s_waitcnt vmcnt(2)
-; SI-NEXT: v_cvt_f16_f32_e32 v7, v7
-; SI-NEXT: v_cvt_f16_f32_e32 v16, v5
-; SI-NEXT: v_cvt_f16_f32_e32 v0, v0
-; SI-NEXT: v_cvt_f16_f32_e32 v6, v6
-; SI-NEXT: v_cvt_f16_f32_e32 v17, v4
-; SI-NEXT: s_waitcnt vmcnt(1)
-; SI-NEXT: v_cvt_f16_f32_e32 v11, v11
-; SI-NEXT: v_cvt_f16_f32_e32 v9, v9
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_cvt_f16_f32_e32 v15, v15
-; SI-NEXT: v_cvt_f16_f32_e32 v13, v13
-; SI-NEXT: v_cvt_f16_f32_e32 v10, v10
-; SI-NEXT: v_cvt_f16_f32_e32 v8, v8
-; SI-NEXT: v_cvt_f16_f32_e32 v14, v14
-; SI-NEXT: v_cvt_f16_f32_e32 v12, v12
-; SI-NEXT: v_lshlrev_b32_e32 v3, 16, v3
-; SI-NEXT: v_mov_b32_e32 v5, s3
-; SI-NEXT: v_lshlrev_b32_e32 v18, 16, v1
-; SI-NEXT: v_or_b32_e32 v1, v2, v3
-; SI-NEXT: v_lshlrev_b32_e32 v2, 16, v7
-; SI-NEXT: v_lshlrev_b32_e32 v7, 16, v16
-; SI-NEXT: v_mov_b32_e32 v4, s2
-; SI-NEXT: v_or_b32_e32 v0, v0, v18
-; SI-NEXT: v_or_b32_e32 v3, v6, v2
-; SI-NEXT: v_or_b32_e32 v2, v17, v7
-; SI-NEXT: v_lshlrev_b32_e32 v6, 16, v11
-; SI-NEXT: v_lshlrev_b32_e32 v7, 16, v9
-; SI-NEXT: v_lshlrev_b32_e32 v9, 16, v15
-; SI-NEXT: v_lshlrev_b32_e32 v11, 16, v13
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: v_mov_b32_e32 v5, s1
-; SI-NEXT: v_or_b32_e32 v1, v10, v6
-; SI-NEXT: v_or_b32_e32 v0, v8, v7
-; SI-NEXT: v_or_b32_e32 v3, v14, v9
-; SI-NEXT: v_or_b32_e32 v2, v12, v11
-; SI-NEXT: v_mov_b32_e32 v4, s0
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: global_truncstore_v16f32_to_v16f16:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: s_add_u32 s4, s2, 32
+; CI-NEXT: s_addc_u32 s5, s3, 0
+; CI-NEXT: v_mov_b32_e32 v0, s4
+; CI-NEXT: v_mov_b32_e32 v1, s5
+; CI-NEXT: s_add_u32 s4, s2, 48
+; CI-NEXT: s_addc_u32 s5, s3, 0
+; CI-NEXT: v_mov_b32_e32 v9, s3
+; CI-NEXT: v_mov_b32_e32 v4, s4
+; CI-NEXT: v_mov_b32_e32 v8, s2
+; CI-NEXT: s_add_u32 s2, s2, 16
+; CI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
+; CI-NEXT: v_mov_b32_e32 v5, s5
+; CI-NEXT: flat_load_dwordx4 v[4:7], v[4:5]
+; CI-NEXT: s_addc_u32 s3, s3, 0
+; CI-NEXT: v_mov_b32_e32 v13, s3
+; CI-NEXT: v_mov_b32_e32 v12, s2
+; CI-NEXT: flat_load_dwordx4 v[8:11], v[8:9]
+; CI-NEXT: flat_load_dwordx4 v[12:15], v[12:13]
+; CI-NEXT: s_add_u32 s2, s0, 16
+; CI-NEXT: s_addc_u32 s3, s1, 0
+; CI-NEXT: s_waitcnt vmcnt(3)
+; CI-NEXT: v_cvt_f16_f32_e32 v3, v3
+; CI-NEXT: v_cvt_f16_f32_e32 v2, v2
+; CI-NEXT: v_cvt_f16_f32_e32 v1, v1
+; CI-NEXT: s_waitcnt vmcnt(2)
+; CI-NEXT: v_cvt_f16_f32_e32 v7, v7
+; CI-NEXT: v_cvt_f16_f32_e32 v16, v5
+; CI-NEXT: v_cvt_f16_f32_e32 v0, v0
+; CI-NEXT: v_cvt_f16_f32_e32 v6, v6
+; CI-NEXT: v_cvt_f16_f32_e32 v17, v4
+; CI-NEXT: s_waitcnt vmcnt(1)
+; CI-NEXT: v_cvt_f16_f32_e32 v11, v11
+; CI-NEXT: v_cvt_f16_f32_e32 v9, v9
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_cvt_f16_f32_e32 v15, v15
+; CI-NEXT: v_cvt_f16_f32_e32 v13, v13
+; CI-NEXT: v_cvt_f16_f32_e32 v10, v10
+; CI-NEXT: v_cvt_f16_f32_e32 v8, v8
+; CI-NEXT: v_cvt_f16_f32_e32 v14, v14
+; CI-NEXT: v_cvt_f16_f32_e32 v12, v12
+; CI-NEXT: v_lshlrev_b32_e32 v3, 16, v3
+; CI-NEXT: v_mov_b32_e32 v5, s3
+; CI-NEXT: v_lshlrev_b32_e32 v18, 16, v1
+; CI-NEXT: v_or_b32_e32 v1, v2, v3
+; CI-NEXT: v_lshlrev_b32_e32 v2, 16, v7
+; CI-NEXT: v_lshlrev_b32_e32 v7, 16, v16
+; CI-NEXT: v_mov_b32_e32 v4, s2
+; CI-NEXT: v_or_b32_e32 v0, v0, v18
+; CI-NEXT: v_or_b32_e32 v3, v6, v2
+; CI-NEXT: v_or_b32_e32 v2, v17, v7
+; CI-NEXT: v_lshlrev_b32_e32 v6, 16, v11
+; CI-NEXT: v_lshlrev_b32_e32 v7, 16, v9
+; CI-NEXT: v_lshlrev_b32_e32 v9, 16, v15
+; CI-NEXT: v_lshlrev_b32_e32 v11, 16, v13
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: v_mov_b32_e32 v5, s1
+; CI-NEXT: v_or_b32_e32 v1, v10, v6
+; CI-NEXT: v_or_b32_e32 v0, v8, v7
+; CI-NEXT: v_or_b32_e32 v3, v14, v9
+; CI-NEXT: v_or_b32_e32 v2, v12, v11
+; CI-NEXT: v_mov_b32_e32 v4, s0
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: global_truncstore_v16f32_to_v16f16:
; VI: ; %bb.0:
@@ -1861,21 +1861,21 @@ define amdgpu_kernel void @global_truncstore_v16f32_to_v16f16(<16 x half> addrsp
; FIXME: Unsafe math should fold conversions away
define amdgpu_kernel void @fadd_f16(half addrspace(1)* %out, half %a, half %b) #0 {
-; SI-LABEL: fadd_f16:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dword s0, s[4:5], 0x2
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_cvt_f32_f16_e32 v0, s0
-; SI-NEXT: s_lshr_b32 s0, s0, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v1, s0
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
-; SI-NEXT: v_add_f32_e32 v0, v0, v1
-; SI-NEXT: v_cvt_f16_f32_e32 v2, v0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s0
-; SI-NEXT: v_mov_b32_e32 v1, s1
-; SI-NEXT: flat_store_short v[0:1], v2
-; SI-NEXT: s_endpgm
+; CI-LABEL: fadd_f16:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dword s0, s[4:5], 0x2
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_cvt_f32_f16_e32 v0, s0
+; CI-NEXT: s_lshr_b32 s0, s0, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v1, s0
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; CI-NEXT: v_add_f32_e32 v0, v0, v1
+; CI-NEXT: v_cvt_f16_f32_e32 v2, v0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s0
+; CI-NEXT: v_mov_b32_e32 v1, s1
+; CI-NEXT: flat_store_short v[0:1], v2
+; CI-NEXT: s_endpgm
;
; VI-LABEL: fadd_f16:
; VI: ; %bb.0:
@@ -1895,28 +1895,28 @@ define amdgpu_kernel void @fadd_f16(half addrspace(1)* %out, half %a, half %b) #
}
define amdgpu_kernel void @fadd_v2f16(<2 x half> addrspace(1)* %out, <2 x half> %a, <2 x half> %b) #0 {
-; SI-LABEL: fadd_v2f16:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x2
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: s_lshr_b32 s2, s0, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v0, s0
-; SI-NEXT: s_lshr_b32 s0, s1, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v1, s1
-; SI-NEXT: v_cvt_f32_f16_e32 v2, s2
-; SI-NEXT: v_cvt_f32_f16_e32 v3, s0
-; SI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
-; SI-NEXT: v_add_f32_e32 v0, v0, v1
-; SI-NEXT: v_cvt_f16_f32_e32 v0, v0
-; SI-NEXT: v_add_f32_e32 v1, v2, v3
-; SI-NEXT: v_cvt_f16_f32_e32 v1, v1
-; SI-NEXT: v_lshlrev_b32_e32 v1, 16, v1
-; SI-NEXT: v_or_b32_e32 v2, v0, v1
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s0
-; SI-NEXT: v_mov_b32_e32 v1, s1
-; SI-NEXT: flat_store_dword v[0:1], v2
-; SI-NEXT: s_endpgm
+; CI-LABEL: fadd_v2f16:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x2
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: s_lshr_b32 s2, s0, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v0, s0
+; CI-NEXT: s_lshr_b32 s0, s1, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v1, s1
+; CI-NEXT: v_cvt_f32_f16_e32 v2, s2
+; CI-NEXT: v_cvt_f32_f16_e32 v3, s0
+; CI-NEXT: s_load_dwordx2 s[0:1], s[4:5], 0x0
+; CI-NEXT: v_add_f32_e32 v0, v0, v1
+; CI-NEXT: v_cvt_f16_f32_e32 v0, v0
+; CI-NEXT: v_add_f32_e32 v1, v2, v3
+; CI-NEXT: v_cvt_f16_f32_e32 v1, v1
+; CI-NEXT: v_lshlrev_b32_e32 v1, 16, v1
+; CI-NEXT: v_or_b32_e32 v2, v0, v1
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s0
+; CI-NEXT: v_mov_b32_e32 v1, s1
+; CI-NEXT: flat_store_dword v[0:1], v2
+; CI-NEXT: s_endpgm
;
; VI-LABEL: fadd_v2f16:
; VI: ; %bb.0:
@@ -1941,42 +1941,42 @@ define amdgpu_kernel void @fadd_v2f16(<2 x half> addrspace(1)* %out, <2 x half>
}
define amdgpu_kernel void @fadd_v4f16(<4 x half> addrspace(1)* %out, <4 x half> addrspace(1)* %in) #0 {
-; SI-LABEL: fadd_v4f16:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: v_mov_b32_e32 v0, s2
-; SI-NEXT: v_mov_b32_e32 v1, s3
-; SI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
-; SI-NEXT: v_mov_b32_e32 v4, s0
-; SI-NEXT: v_mov_b32_e32 v5, s1
-; SI-NEXT: s_waitcnt vmcnt(0)
-; SI-NEXT: v_cvt_f32_f16_e32 v6, v0
-; SI-NEXT: v_lshrrev_b32_e32 v0, 16, v0
-; SI-NEXT: v_cvt_f32_f16_e32 v7, v1
-; SI-NEXT: v_lshrrev_b32_e32 v1, 16, v1
-; SI-NEXT: v_cvt_f32_f16_e32 v8, v2
-; SI-NEXT: v_lshrrev_b32_e32 v2, 16, v2
-; SI-NEXT: v_cvt_f32_f16_e32 v9, v3
-; SI-NEXT: v_lshrrev_b32_e32 v3, 16, v3
-; SI-NEXT: v_cvt_f32_f16_e32 v0, v0
-; SI-NEXT: v_cvt_f32_f16_e32 v1, v1
-; SI-NEXT: v_cvt_f32_f16_e32 v3, v3
-; SI-NEXT: v_cvt_f32_f16_e32 v2, v2
-; SI-NEXT: v_add_f32_e32 v7, v7, v9
-; SI-NEXT: v_add_f32_e32 v6, v6, v8
-; SI-NEXT: v_add_f32_e32 v1, v1, v3
-; SI-NEXT: v_add_f32_e32 v0, v0, v2
-; SI-NEXT: v_cvt_f16_f32_e32 v1, v1
-; SI-NEXT: v_cvt_f16_f32_e32 v0, v0
-; SI-NEXT: v_cvt_f16_f32_e32 v2, v7
-; SI-NEXT: v_cvt_f16_f32_e32 v3, v6
-; SI-NEXT: v_lshlrev_b32_e32 v1, 16, v1
-; SI-NEXT: v_lshlrev_b32_e32 v0, 16, v0
-; SI-NEXT: v_or_b32_e32 v1, v2, v1
-; SI-NEXT: v_or_b32_e32 v0, v3, v0
-; SI-NEXT: flat_store_dwordx2 v[4:5], v[0:1]
-; SI-NEXT: s_endpgm
+; CI-LABEL: fadd_v4f16:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x0
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: v_mov_b32_e32 v0, s2
+; CI-NEXT: v_mov_b32_e32 v1, s3
+; CI-NEXT: flat_load_dwordx4 v[0:3], v[0:1]
+; CI-NEXT: v_mov_b32_e32 v4, s0
+; CI-NEXT: v_mov_b32_e32 v5, s1
+; CI-NEXT: s_waitcnt vmcnt(0)
+; CI-NEXT: v_cvt_f32_f16_e32 v6, v0
+; CI-NEXT: v_lshrrev_b32_e32 v0, 16, v0
+; CI-NEXT: v_cvt_f32_f16_e32 v7, v1
+; CI-NEXT: v_lshrrev_b32_e32 v1, 16, v1
+; CI-NEXT: v_cvt_f32_f16_e32 v8, v2
+; CI-NEXT: v_lshrrev_b32_e32 v2, 16, v2
+; CI-NEXT: v_cvt_f32_f16_e32 v9, v3
+; CI-NEXT: v_lshrrev_b32_e32 v3, 16, v3
+; CI-NEXT: v_cvt_f32_f16_e32 v0, v0
+; CI-NEXT: v_cvt_f32_f16_e32 v1, v1
+; CI-NEXT: v_cvt_f32_f16_e32 v3, v3
+; CI-NEXT: v_cvt_f32_f16_e32 v2, v2
+; CI-NEXT: v_add_f32_e32 v7, v7, v9
+; CI-NEXT: v_add_f32_e32 v6, v6, v8
+; CI-NEXT: v_add_f32_e32 v1, v1, v3
+; CI-NEXT: v_add_f32_e32 v0, v0, v2
+; CI-NEXT: v_cvt_f16_f32_e32 v1, v1
+; CI-NEXT: v_cvt_f16_f32_e32 v0, v0
+; CI-NEXT: v_cvt_f16_f32_e32 v2, v7
+; CI-NEXT: v_cvt_f16_f32_e32 v3, v6
+; CI-NEXT: v_lshlrev_b32_e32 v1, 16, v1
+; CI-NEXT: v_lshlrev_b32_e32 v0, 16, v0
+; CI-NEXT: v_or_b32_e32 v1, v2, v1
+; CI-NEXT: v_or_b32_e32 v0, v3, v0
+; CI-NEXT: flat_store_dwordx2 v[4:5], v[0:1]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: fadd_v4f16:
; VI: ; %bb.0:
@@ -2005,64 +2005,64 @@ define amdgpu_kernel void @fadd_v4f16(<4 x half> addrspace(1)* %out, <4 x half>
}
define amdgpu_kernel void @fadd_v8f16(<8 x half> addrspace(1)* %out, <8 x half> %a, <8 x half> %b) #0 {
-; SI-LABEL: fadd_v8f16:
-; SI: ; %bb.0:
-; SI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x4
-; SI-NEXT: s_load_dwordx2 s[8:9], s[4:5], 0x0
-; SI-NEXT: s_load_dwordx4 s[4:7], s[4:5], 0x8
-; SI-NEXT: s_waitcnt lgkmcnt(0)
-; SI-NEXT: s_lshr_b32 s10, s0, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v4, s0
-; SI-NEXT: s_lshr_b32 s0, s4, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v8, s0
-; SI-NEXT: s_lshr_b32 s0, s5, 16
-; SI-NEXT: s_lshr_b32 s11, s1, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v0, s10
-; SI-NEXT: s_lshr_b32 s10, s2, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v9, s0
-; SI-NEXT: s_lshr_b32 s0, s6, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v1, s11
-; SI-NEXT: v_cvt_f32_f16_e32 v2, s10
-; SI-NEXT: s_lshr_b32 s10, s3, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v10, s0
-; SI-NEXT: s_lshr_b32 s0, s7, 16
-; SI-NEXT: v_cvt_f32_f16_e32 v3, s10
-; SI-NEXT: v_cvt_f32_f16_e32 v5, s1
-; SI-NEXT: v_cvt_f32_f16_e32 v11, s0
-; SI-NEXT: v_cvt_f32_f16_e32 v12, s4
-; SI-NEXT: v_cvt_f32_f16_e32 v13, s5
-; SI-NEXT: v_cvt_f32_f16_e32 v6, s2
-; SI-NEXT: v_cvt_f32_f16_e32 v7, s3
-; SI-NEXT: v_cvt_f32_f16_e32 v14, s7
-; SI-NEXT: v_cvt_f32_f16_e32 v15, s6
-; SI-NEXT: v_add_f32_e32 v1, v1, v9
-; SI-NEXT: v_add_f32_e32 v0, v0, v8
-; SI-NEXT: v_add_f32_e32 v3, v3, v11
-; SI-NEXT: v_add_f32_e32 v2, v2, v10
-; SI-NEXT: v_cvt_f16_f32_e32 v1, v1
-; SI-NEXT: v_add_f32_e32 v5, v5, v13
-; SI-NEXT: v_cvt_f16_f32_e32 v0, v0
-; SI-NEXT: v_add_f32_e32 v4, v4, v12
-; SI-NEXT: v_cvt_f16_f32_e32 v3, v3
-; SI-NEXT: v_add_f32_e32 v7, v7, v14
-; SI-NEXT: v_cvt_f16_f32_e32 v2, v2
-; SI-NEXT: v_add_f32_e32 v6, v6, v15
-; SI-NEXT: v_cvt_f16_f32_e32 v5, v5
-; SI-NEXT: v_cvt_f16_f32_e32 v4, v4
-; SI-NEXT: v_cvt_f16_f32_e32 v7, v7
-; SI-NEXT: v_cvt_f16_f32_e32 v6, v6
-; SI-NEXT: v_lshlrev_b32_e32 v1, 16, v1
-; SI-NEXT: v_lshlrev_b32_e32 v0, 16, v0
-; SI-NEXT: v_lshlrev_b32_e32 v3, 16, v3
-; SI-NEXT: v_lshlrev_b32_e32 v2, 16, v2
-; SI-NEXT: v_or_b32_e32 v1, v5, v1
-; SI-NEXT: v_or_b32_e32 v0, v4, v0
-; SI-NEXT: v_mov_b32_e32 v4, s8
-; SI-NEXT: v_or_b32_e32 v3, v7, v3
-; SI-NEXT: v_or_b32_e32 v2, v6, v2
-; SI-NEXT: v_mov_b32_e32 v5, s9
-; SI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
-; SI-NEXT: s_endpgm
+; CI-LABEL: fadd_v8f16:
+; CI: ; %bb.0:
+; CI-NEXT: s_load_dwordx4 s[0:3], s[4:5], 0x4
+; CI-NEXT: s_load_dwordx2 s[8:9], s[4:5], 0x0
+; CI-NEXT: s_load_dwordx4 s[4:7], s[4:5], 0x8
+; CI-NEXT: s_waitcnt lgkmcnt(0)
+; CI-NEXT: s_lshr_b32 s10, s0, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v4, s0
+; CI-NEXT: s_lshr_b32 s0, s4, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v8, s0
+; CI-NEXT: s_lshr_b32 s0, s5, 16
+; CI-NEXT: s_lshr_b32 s11, s1, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v0, s10
+; CI-NEXT: s_lshr_b32 s10, s2, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v9, s0
+; CI-NEXT: s_lshr_b32 s0, s6, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v1, s11
+; CI-NEXT: v_cvt_f32_f16_e32 v2, s10
+; CI-NEXT: s_lshr_b32 s10, s3, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v10, s0
+; CI-NEXT: s_lshr_b32 s0, s7, 16
+; CI-NEXT: v_cvt_f32_f16_e32 v3, s10
+; CI-NEXT: v_cvt_f32_f16_e32 v5, s1
+; CI-NEXT: v_cvt_f32_f16_e32 v11, s0
+; CI-NEXT: v_cvt_f32_f16_e32 v12, s4
+; CI-NEXT: v_cvt_f32_f16_e32 v13, s5
+; CI-NEXT: v_cvt_f32_f16_e32 v6, s2
+; CI-NEXT: v_cvt_f32_f16_e32 v7, s3
+; CI-NEXT: v_cvt_f32_f16_e32 v14, s7
+; CI-NEXT: v_cvt_f32_f16_e32 v15, s6
+; CI-NEXT: v_add_f32_e32 v1, v1, v9
+; CI-NEXT: v_add_f32_e32 v0, v0, v8
+; CI-NEXT: v_add_f32_e32 v3, v3, v11
+; CI-NEXT: v_add_f32_e32 v2, v2, v10
+; CI-NEXT: v_cvt_f16_f32_e32 v1, v1
+; CI-NEXT: v_add_f32_e32 v5, v5, v13
+; CI-NEXT: v_cvt_f16_f32_e32 v0, v0
+; CI-NEXT: v_add_f32_e32 v4, v4, v12
+; CI-NEXT: v_cvt_f16_f32_e32 v3, v3
+; CI-NEXT: v_add_f32_e32 v7, v7, v14
+; CI-NEXT: v_cvt_f16_f32_e32 v2, v2
+; CI-NEXT: v_add_f32_e32 v6, v6, v15
+; CI-NEXT: v_cvt_f16_f32_e32 v5, v5
+; CI-NEXT: v_cvt_f16_f32_e32 v4, v4
+; CI-NEXT: v_cvt_f16_f32_e32 v7, v7
+; CI-NEXT: v_cvt_f16_f32_e32 v6, v6
+; CI-NEXT: v_lshlrev_b32_e32 v1, 16, v1
+; CI-NEXT: v_lshlrev_b32_e32 v0, 16, v0
+; CI-NEXT: v_lshlrev_b32_e32 v3, 16, v3
+; CI-NEXT: v_lshlrev_b32_e32 v2, 16, v2
+; CI-NEXT: v_or_b32_e32 v1, v5, v1
+; CI-NEXT: v_or_b32_e32 v0, v4, v0
+; CI-NEXT: v_mov_b32_e32 v4, s8
+; CI-NEXT: v_or_b32_e32 v3, v7, v3
+; CI-NEXT: v_or_b32_e32 v2, v6, v2
+; CI-NEXT: v_mov_b32_e32 v5, s9
+; CI-NEXT: flat_store_dwordx4 v[4:5], v[0:3]
+; CI-NEXT: s_endpgm
;
; VI-LABEL: fadd_v8f16:
; VI: ; %bb.0:
diff --git a/llvm/test/CodeGen/AMDGPU/min.ll b/llvm/test/CodeGen/AMDGPU/min.ll
index c5e83127166bb..e97fece0d1842 100644
--- a/llvm/test/CodeGen/AMDGPU/min.ll
+++ b/llvm/test/CodeGen/AMDGPU/min.ll
@@ -1,4 +1,4 @@
-; RUN: llc -march=amdgcn -mtriple=amdgcn-amd-amdhsa -mcpu=kaveri -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,SI,FUNC %s
+; RUN: llc -march=amdgcn -mtriple=amdgcn-amd-amdhsa -mcpu=kaveri -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,CI,FUNC %s
; RUN: llc -march=amdgcn -mtriple=amdgcn-amd-amdhsa -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,VI,GFX8_9_10,FUNC %s
; RUN: llc -march=amdgcn -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,GFX9_10,GFX8_9_10,FUNC %s
; RUN: llc -march=amdgcn -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1010 -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck --check-prefixes=GCN,GFX10,GFX9_10,GFX8_9_10,FUNC %s
@@ -80,10 +80,10 @@ define amdgpu_kernel void @s_test_imin_sle_i8(i8 addrspace(1)* %out, [8 x i32],
; GCN-DAG: s_load_dword s
; GCN-NOT: _load_
-; SI: s_min_i32
-; SI: s_min_i32
-; SI: s_min_i32
-; SI: s_min_i32
+; CI: s_min_i32
+; CI: s_min_i32
+; CI: s_min_i32
+; CI: s_min_i32
; VI-DAG: s_min_i32
; VI-DAG: s_min_i32
@@ -110,12 +110,12 @@ define amdgpu_kernel void @s_test_imin_sle_v4i8(<4 x i8> addrspace(1)* %out, [8
; GCN: s_load_dwordx2 s
; GCN: s_load_dwordx2 s
-; SI: s_ashr_i32
-; SI: s_sext_i32_i16
-; SI: s_ashr_i32
-; SI: s_sext_i32_i16
-; SI: s_min_i32
-; SI: s_min_i32
+; CI: s_ashr_i32
+; CI: s_sext_i32_i16
+; CI: s_ashr_i32
+; CI: s_sext_i32_i16
+; CI: s_min_i32
+; CI: s_min_i32
; VI: s_sext_i32_i16
; VI: s_sext_i32_i16
@@ -134,11 +134,11 @@ define amdgpu_kernel void @s_test_imin_sle_v2i16(<2 x i16> addrspace(1)* %out, <
}
; FUNC-LABEL: {{^}}s_test_imin_sle_v4i16:
-; SI-NOT: buffer_load
-; SI: s_min_i32
-; SI: s_min_i32
-; SI: s_min_i32
-; SI: s_min_i32
+; CI-NOT: buffer_load
+; CI: s_min_i32
+; CI: s_min_i32
+; CI: s_min_i32
+; CI: s_min_i32
; VI: s_min_i32
; VI: s_min_i32
@@ -177,7 +177,7 @@ define amdgpu_kernel void @v_test_imin_slt_i32(i32 addrspace(1)* %out, i32 addrs
}
; FUNC-LABEL: @v_test_imin_slt_i16
-; SI: v_min_i32_e32
+; CI: v_min_i32_e32
; GFX8_9: v_min_i16_e32
; GFX10: v_min_i16
@@ -286,10 +286,10 @@ define amdgpu_kernel void @v_test_umin_ule_v3i32(<3 x i32> addrspace(1)* %out, <
; FIXME: Reduce unused packed component to scalar
; FUNC-LABEL: @v_test_umin_ule_v3i16{{$}}
-; SI: v_min_u32_e32
-; SI: v_min_u32_e32
-; SI: v_min_u32_e32
-; SI-NOT: v_min_u32_e32
+; CI: v_min_u32_e32
+; CI: v_min_u32_e32
+; CI: v_min_u32_e32
+; CI-NOT: v_min_u32_e32
; VI: v_min_u16_e32
; VI: v_min_u16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
@@ -347,9 +347,9 @@ define amdgpu_kernel void @v_test_umin_ult_i32(i32 addrspace(1)* %out, i32 addrs
}
; FUNC-LABEL: {{^}}v_test_umin_ult_i8:
-; SI: {{buffer|flat|global}}_load_ubyte
-; SI: {{buffer|flat|global}}_load_ubyte
-; SI: v_min_u32_e32
+; CI: {{buffer|flat|global}}_load_ubyte
+; CI: {{buffer|flat|global}}_load_ubyte
+; CI: v_min_u32_e32
; GFX8_9_10: {{flat|global}}_load_ubyte
; GFX8_9_10: {{flat|global}}_load_ubyte
@@ -383,11 +383,11 @@ define amdgpu_kernel void @s_test_umin_ult_i32(i32 addrspace(1)* %out, i32 %a, i
}
; FUNC-LABEL: @v_test_umin_ult_i32_multi_use
-; SI-NOT: v_min
+; CI-NOT: v_min
; GCN: s_cmp_lt_u32
-; SI-NOT: v_min
-; SI: v_cndmask_b32
-; SI-NOT: v_min
+; CI-NOT: v_min
+; CI: v_cndmask_b32
+; CI-NOT: v_min
; GCN: s_endpgm
; EG-NOT: MIN_UINT
@@ -458,14 +458,14 @@ define amdgpu_kernel void @s_test_umin_ult_v8i32(<8 x i32> addrspace(1)* %out, <
; FUNC-LABEL: {{^}}s_test_umin_ult_v8i16:
; GCN-NOT: {{buffer|flat|global}}_load
-; SI: s_min_u32
-; SI: s_min_u32
-; SI: s_min_u32
-; SI: s_min_u32
-; SI: s_min_u32
-; SI: s_min_u32
-; SI: s_min_u32
-; SI: s_min_u32
+; CI: s_min_u32
+; CI: s_min_u32
+; CI: s_min_u32
+; CI: s_min_u32
+; CI: s_min_u32
+; CI: s_min_u32
+; CI: s_min_u32
+; CI: s_min_u32
; VI: s_min_u32
; VI: s_min_u32
@@ -595,8 +595,8 @@ define amdgpu_kernel void @test_imin_sle_i64(i64 addrspace(1)* %out, i64 %a, i64
}
; FUNC-LABEL: {{^}}v_test_imin_sle_v2i16:
-; SI: v_min_i32
-; SI: v_min_i32
+; CI: v_min_i32
+; CI: v_min_i32
; VI: v_min_i16
; VI: v_min_i16
@@ -620,8 +620,8 @@ define amdgpu_kernel void @v_test_imin_sle_v2i16(<2 x i16> addrspace(1)* %out, <
; FIXME: i16 min
; FUNC-LABEL: {{^}}v_test_imin_ule_v2i16:
-; SI: v_min_u32
-; SI: v_min_u32
+; CI: v_min_u32
+; CI: v_min_u32
; VI: v_min_u16
; VI: v_min_u16
diff --git a/llvm/test/CodeGen/AMDGPU/sint_to_fp.f64.ll b/llvm/test/CodeGen/AMDGPU/sint_to_fp.f64.ll
index fb30f0716afbf..c5ceb23337577 100644
--- a/llvm/test/CodeGen/AMDGPU/sint_to_fp.f64.ll
+++ b/llvm/test/CodeGen/AMDGPU/sint_to_fp.f64.ll
@@ -1,4 +1,4 @@
-; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=hawaii -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,SI %s
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=hawaii -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,CI %s
; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=fiji -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,VI %s
declare i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
@@ -22,12 +22,12 @@ define amdgpu_kernel void @sint_to_fp_i32_to_f64(double addrspace(1)* %out, i32
; VI: flat_store_dwordx2 v{{\[[0-9]+:[0-9]+\]}}, v[[[ZERO]]:[[SEL]]]
; VI: s_endpgm
-; SI-DAG: s_cmp_eq_u32
-; SI-DAG: s_cselect_b64 vcc, -1, 0
-; SI-DAG: v_cndmask_b32_e32 v[[SEL:[0-9]+]], 0, v{{[0-9]+}}, vcc
-; SI-DAG: v_mov_b32_e32 v[[ZERO:[0-9]+]], 0{{$}}
-; SI: flat_store_dwordx2 v{{\[[0-9]+:[0-9]+\]}}, v[[[ZERO]]:[[SEL]]]
-; SI: s_endpgm
+; CI-DAG: s_cmp_eq_u32
+; CI-DAG: s_cselect_b64 vcc, -1, 0
+; CI-DAG: v_cndmask_b32_e32 v[[SEL:[0-9]+]], 0, v{{[0-9]+}}, vcc
+; CI-DAG: v_mov_b32_e32 v[[ZERO:[0-9]+]], 0{{$}}
+; CI: flat_store_dwordx2 v{{\[[0-9]+:[0-9]+\]}}, v[[[ZERO]]:[[SEL]]]
+; CI: s_endpgm
define amdgpu_kernel void @sint_to_fp_i1_f64(double addrspace(1)* %out, i32 %in) {
%cmp = icmp eq i32 %in, 0
%fp = sitofp i1 %cmp to double
@@ -72,8 +72,8 @@ define amdgpu_kernel void @v_sint_to_fp_i64_to_f64(double addrspace(1)* %out, i6
; FIXME: bfe and sext on VI+
; GCN-LABEL: {{^}}s_sint_to_fp_i8_to_f64:
; GCN: s_load_dword [[VAL:s[0-9]+]]
-; SI-NOT: bfe
-; SI: s_sext_i32_i8 [[SEXT:s[0-9]+]], [[VAL]]
+; CI-NOT: bfe
+; CI: s_sext_i32_i8 [[SEXT:s[0-9]+]], [[VAL]]
; VI: s_bfe_i32 [[BFE:s[0-9]+]], [[VAL]], 0x80000
; VI: s_sext_i32_i16 [[SEXT:s[0-9]+]], [[BFE]]
More information about the llvm-commits
mailing list