-
Notifications
You must be signed in to change notification settings - Fork 15.4k
[AMDGPU] Autogen checks for mfma-no-register-aliasing.ll #132117
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merged
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Change-Id: I72ffc63d877ef47cd22dcadd374e039d6dae78f7
Member
|
@llvm/pr-subscribers-backend-amdgpu Author: Jeffrey Byrnes (jrbyrnes) ChangesFor an upcoming RegisterCoalescer PR Patch is 52.70 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/132117.diff 1 Files Affected:
diff --git a/llvm/test/CodeGen/AMDGPU/mfma-no-register-aliasing.ll b/llvm/test/CodeGen/AMDGPU/mfma-no-register-aliasing.ll
index f76580b94e13c..21af2dde2c4bf 100644
--- a/llvm/test/CodeGen/AMDGPU/mfma-no-register-aliasing.ll
+++ b/llvm/test/CodeGen/AMDGPU/mfma-no-register-aliasing.ll
@@ -1,9 +1,10 @@
-; RUN: llc -mtriple=amdgcn -mcpu=gfx908 -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=GCN,GREEDY,GREEDY908 %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx90a -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=GCN,GREEDY,GREEDY90A %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx90a -early-live-intervals -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=GCN,GREEDY,GREEDY90A %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx942 -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=GCN,GREEDY,GREEDY90A %s
-; RUN: llc -global-isel -mtriple=amdgcn -mcpu=gfx90a -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=GCN,GREEDY,GREEDY90A-GISEL %s
-; RUN: llc -mtriple=amdgcn -mcpu=gfx90a -sgpr-regalloc=fast -vgpr-regalloc=fast -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=GCN,FAST %s
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -mtriple=amdgcn -mcpu=gfx908 -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=GREEDY908 %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx90a -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=GREEDY90A %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx90a -early-live-intervals -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=GREEDY90A %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx942 -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=GREEDY942 %s
+; RUN: llc -global-isel -mtriple=amdgcn -mcpu=gfx90a -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=GREEDY90A-GISEL %s
+; RUN: llc -mtriple=amdgcn -mcpu=gfx90a -sgpr-regalloc=fast -vgpr-regalloc=fast -verify-machineinstrs < %s | FileCheck -enable-var-scope --check-prefixes=FAST90A %s
; This is better with 90a
@@ -14,13 +15,612 @@ declare <32 x float> @llvm.amdgcn.mfma.f32.32x32x1f32(float, float, <32 x float>
declare <16 x float> @llvm.amdgcn.mfma.f32.16x16x1f32(float, float, <16 x float>, i32, i32, i32)
declare <4 x float> @llvm.amdgcn.mfma.f32.4x4x1f32(float, float, <4 x float>, i32, i32, i32)
-; GCN-LABEL: {{^}}test_mfma_f32_32x32x1f32:
-; GREEDY: v_mfma_f32_32x32x1{{.*}} a[0:31], v{{[0-9]+}}, v{{[0-9]+}}, a[0:31]
-; GREEDY: v_mfma_f32_32x32x1{{.*}} a[32:63], v{{[0-9]+}}, v{{[0-9]+}}, a[0:31]
-; FAST: v_mfma_f32_32x32x1{{.*}} a[32:63], v{{[0-9]+}}, v{{[0-9]+}}, a[32:63]
-; FAST: v_mfma_f32_32x32x1{{.*}} a[0:31], v{{[0-9]+}}, v{{[0-9]+}}, a[32:63]
-; GCN: v_mfma_f32_32x32x1{{.*}} a[0:31], v{{[0-9]+}}, v{{[0-9]+}}, a[0:31]
define amdgpu_kernel void @test_mfma_f32_32x32x1f32(ptr addrspace(1) %arg) #0 {
+; GREEDY908-LABEL: test_mfma_f32_32x32x1f32:
+; GREEDY908: ; %bb.0: ; %bb
+; GREEDY908-NEXT: s_load_dwordx2 s[34:35], s[4:5], 0x24
+; GREEDY908-NEXT: s_waitcnt lgkmcnt(0)
+; GREEDY908-NEXT: s_load_dwordx16 s[16:31], s[34:35], 0x0
+; GREEDY908-NEXT: s_load_dwordx16 s[0:15], s[34:35], 0x40
+; GREEDY908-NEXT: s_waitcnt lgkmcnt(0)
+; GREEDY908-NEXT: v_mov_b32_e32 v0, s16
+; GREEDY908-NEXT: v_mov_b32_e32 v1, s17
+; GREEDY908-NEXT: v_mov_b32_e32 v2, s18
+; GREEDY908-NEXT: v_accvgpr_write_b32 a0, v0
+; GREEDY908-NEXT: v_accvgpr_write_b32 a1, v1
+; GREEDY908-NEXT: v_mov_b32_e32 v0, s22
+; GREEDY908-NEXT: v_mov_b32_e32 v1, s23
+; GREEDY908-NEXT: v_accvgpr_write_b32 a2, v2
+; GREEDY908-NEXT: v_accvgpr_write_b32 a6, v0
+; GREEDY908-NEXT: v_accvgpr_write_b32 a7, v1
+; GREEDY908-NEXT: v_mov_b32_e32 v0, s24
+; GREEDY908-NEXT: v_mov_b32_e32 v1, s25
+; GREEDY908-NEXT: v_mov_b32_e32 v2, s26
+; GREEDY908-NEXT: v_accvgpr_write_b32 a8, v0
+; GREEDY908-NEXT: v_accvgpr_write_b32 a9, v1
+; GREEDY908-NEXT: v_accvgpr_write_b32 a10, v2
+; GREEDY908-NEXT: v_mov_b32_e32 v0, s27
+; GREEDY908-NEXT: v_mov_b32_e32 v1, s28
+; GREEDY908-NEXT: v_mov_b32_e32 v2, s29
+; GREEDY908-NEXT: v_accvgpr_write_b32 a11, v0
+; GREEDY908-NEXT: v_accvgpr_write_b32 a12, v1
+; GREEDY908-NEXT: v_accvgpr_write_b32 a13, v2
+; GREEDY908-NEXT: v_mov_b32_e32 v0, s30
+; GREEDY908-NEXT: v_mov_b32_e32 v1, s31
+; GREEDY908-NEXT: v_mov_b32_e32 v2, s0
+; GREEDY908-NEXT: v_accvgpr_write_b32 a14, v0
+; GREEDY908-NEXT: v_accvgpr_write_b32 a15, v1
+; GREEDY908-NEXT: v_accvgpr_write_b32 a16, v2
+; GREEDY908-NEXT: v_mov_b32_e32 v0, s1
+; GREEDY908-NEXT: v_mov_b32_e32 v1, s2
+; GREEDY908-NEXT: v_mov_b32_e32 v2, s3
+; GREEDY908-NEXT: v_accvgpr_write_b32 a17, v0
+; GREEDY908-NEXT: v_accvgpr_write_b32 a18, v1
+; GREEDY908-NEXT: v_accvgpr_write_b32 a19, v2
+; GREEDY908-NEXT: v_mov_b32_e32 v0, s4
+; GREEDY908-NEXT: v_mov_b32_e32 v1, s5
+; GREEDY908-NEXT: v_mov_b32_e32 v2, s6
+; GREEDY908-NEXT: v_accvgpr_write_b32 a20, v0
+; GREEDY908-NEXT: v_accvgpr_write_b32 a21, v1
+; GREEDY908-NEXT: v_accvgpr_write_b32 a22, v2
+; GREEDY908-NEXT: v_mov_b32_e32 v0, s7
+; GREEDY908-NEXT: v_mov_b32_e32 v1, s8
+; GREEDY908-NEXT: v_mov_b32_e32 v2, s9
+; GREEDY908-NEXT: v_mov_b32_e32 v3, s19
+; GREEDY908-NEXT: v_accvgpr_write_b32 a23, v0
+; GREEDY908-NEXT: v_accvgpr_write_b32 a24, v1
+; GREEDY908-NEXT: v_accvgpr_write_b32 a25, v2
+; GREEDY908-NEXT: v_mov_b32_e32 v0, s10
+; GREEDY908-NEXT: v_mov_b32_e32 v1, s11
+; GREEDY908-NEXT: v_mov_b32_e32 v2, s12
+; GREEDY908-NEXT: v_mov_b32_e32 v4, s20
+; GREEDY908-NEXT: v_mov_b32_e32 v5, s21
+; GREEDY908-NEXT: v_accvgpr_write_b32 a3, v3
+; GREEDY908-NEXT: v_accvgpr_write_b32 a26, v0
+; GREEDY908-NEXT: v_accvgpr_write_b32 a27, v1
+; GREEDY908-NEXT: v_accvgpr_write_b32 a28, v2
+; GREEDY908-NEXT: v_mov_b32_e32 v0, s13
+; GREEDY908-NEXT: v_mov_b32_e32 v1, s14
+; GREEDY908-NEXT: v_mov_b32_e32 v2, s15
+; GREEDY908-NEXT: v_mov_b32_e32 v3, 1.0
+; GREEDY908-NEXT: v_accvgpr_write_b32 a4, v4
+; GREEDY908-NEXT: v_accvgpr_write_b32 a5, v5
+; GREEDY908-NEXT: v_accvgpr_write_b32 a29, v0
+; GREEDY908-NEXT: v_accvgpr_write_b32 a30, v1
+; GREEDY908-NEXT: v_accvgpr_write_b32 a31, v2
+; GREEDY908-NEXT: v_mov_b32_e32 v0, 2.0
+; GREEDY908-NEXT: v_mov_b32_e32 v4, 0
+; GREEDY908-NEXT: s_nop 0
+; GREEDY908-NEXT: v_mfma_f32_32x32x1f32 a[0:31], v3, v0, a[0:31]
+; GREEDY908-NEXT: v_mfma_f32_32x32x1f32 a[32:63], v3, v0, a[0:31]
+; GREEDY908-NEXT: s_nop 7
+; GREEDY908-NEXT: s_nop 7
+; GREEDY908-NEXT: s_nop 1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a32
+; GREEDY908-NEXT: v_accvgpr_read_b32 v5, a61
+; GREEDY908-NEXT: v_accvgpr_read_b32 v6, a60
+; GREEDY908-NEXT: v_accvgpr_write_b32 a2, v1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a33
+; GREEDY908-NEXT: v_accvgpr_read_b32 v7, a59
+; GREEDY908-NEXT: v_accvgpr_read_b32 v8, a58
+; GREEDY908-NEXT: v_accvgpr_write_b32 a3, v1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a34
+; GREEDY908-NEXT: v_accvgpr_read_b32 v9, a57
+; GREEDY908-NEXT: v_accvgpr_read_b32 v10, a56
+; GREEDY908-NEXT: v_accvgpr_write_b32 a4, v1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a35
+; GREEDY908-NEXT: v_accvgpr_read_b32 v11, a55
+; GREEDY908-NEXT: v_accvgpr_read_b32 v12, a54
+; GREEDY908-NEXT: v_accvgpr_write_b32 a5, v1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a36
+; GREEDY908-NEXT: v_accvgpr_read_b32 v13, a53
+; GREEDY908-NEXT: v_accvgpr_read_b32 v14, a52
+; GREEDY908-NEXT: v_accvgpr_write_b32 a6, v1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a37
+; GREEDY908-NEXT: v_accvgpr_read_b32 v15, a51
+; GREEDY908-NEXT: v_accvgpr_read_b32 v16, a50
+; GREEDY908-NEXT: v_accvgpr_write_b32 a7, v1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a38
+; GREEDY908-NEXT: v_accvgpr_read_b32 v17, a49
+; GREEDY908-NEXT: v_accvgpr_read_b32 v18, a48
+; GREEDY908-NEXT: v_accvgpr_write_b32 a8, v1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a39
+; GREEDY908-NEXT: v_accvgpr_read_b32 v19, a47
+; GREEDY908-NEXT: v_accvgpr_read_b32 v2, a46
+; GREEDY908-NEXT: v_accvgpr_write_b32 a9, v1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a40
+; GREEDY908-NEXT: v_accvgpr_write_b32 a16, v2
+; GREEDY908-NEXT: v_accvgpr_write_b32 a17, v19
+; GREEDY908-NEXT: v_accvgpr_write_b32 a10, v1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a41
+; GREEDY908-NEXT: v_accvgpr_write_b32 a18, v18
+; GREEDY908-NEXT: v_accvgpr_write_b32 a19, v17
+; GREEDY908-NEXT: v_accvgpr_write_b32 a11, v1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a42
+; GREEDY908-NEXT: v_accvgpr_write_b32 a20, v16
+; GREEDY908-NEXT: v_accvgpr_write_b32 a21, v15
+; GREEDY908-NEXT: v_accvgpr_write_b32 a12, v1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a43
+; GREEDY908-NEXT: v_accvgpr_write_b32 a22, v14
+; GREEDY908-NEXT: v_accvgpr_write_b32 a23, v13
+; GREEDY908-NEXT: v_accvgpr_write_b32 a13, v1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a44
+; GREEDY908-NEXT: v_accvgpr_write_b32 a24, v12
+; GREEDY908-NEXT: v_accvgpr_write_b32 a25, v11
+; GREEDY908-NEXT: v_accvgpr_write_b32 a14, v1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a45
+; GREEDY908-NEXT: v_accvgpr_write_b32 a26, v10
+; GREEDY908-NEXT: v_accvgpr_write_b32 a27, v9
+; GREEDY908-NEXT: v_accvgpr_write_b32 a15, v1
+; GREEDY908-NEXT: v_accvgpr_write_b32 a28, v8
+; GREEDY908-NEXT: v_accvgpr_write_b32 a29, v7
+; GREEDY908-NEXT: v_accvgpr_write_b32 a30, v6
+; GREEDY908-NEXT: v_accvgpr_write_b32 a31, v5
+; GREEDY908-NEXT: s_nop 0
+; GREEDY908-NEXT: v_mfma_f32_32x32x1f32 a[0:31], v3, v0, a[0:31]
+; GREEDY908-NEXT: s_nop 7
+; GREEDY908-NEXT: s_nop 7
+; GREEDY908-NEXT: s_nop 1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v3, a27
+; GREEDY908-NEXT: v_accvgpr_read_b32 v2, a26
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a25
+; GREEDY908-NEXT: v_accvgpr_read_b32 v0, a24
+; GREEDY908-NEXT: s_nop 1
+; GREEDY908-NEXT: global_store_dwordx4 v4, v[0:3], s[34:35] offset:96
+; GREEDY908-NEXT: s_nop 0
+; GREEDY908-NEXT: v_accvgpr_read_b32 v3, a31
+; GREEDY908-NEXT: v_accvgpr_read_b32 v2, a30
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a29
+; GREEDY908-NEXT: v_accvgpr_read_b32 v0, a28
+; GREEDY908-NEXT: s_nop 1
+; GREEDY908-NEXT: global_store_dwordx4 v4, v[0:3], s[34:35] offset:112
+; GREEDY908-NEXT: s_nop 0
+; GREEDY908-NEXT: v_accvgpr_read_b32 v3, a19
+; GREEDY908-NEXT: v_accvgpr_read_b32 v2, a18
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a17
+; GREEDY908-NEXT: v_accvgpr_read_b32 v0, a16
+; GREEDY908-NEXT: s_nop 1
+; GREEDY908-NEXT: global_store_dwordx4 v4, v[0:3], s[34:35] offset:64
+; GREEDY908-NEXT: s_nop 0
+; GREEDY908-NEXT: v_accvgpr_read_b32 v3, a23
+; GREEDY908-NEXT: v_accvgpr_read_b32 v2, a22
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a21
+; GREEDY908-NEXT: v_accvgpr_read_b32 v0, a20
+; GREEDY908-NEXT: s_nop 1
+; GREEDY908-NEXT: global_store_dwordx4 v4, v[0:3], s[34:35] offset:80
+; GREEDY908-NEXT: s_nop 0
+; GREEDY908-NEXT: v_accvgpr_read_b32 v3, a11
+; GREEDY908-NEXT: v_accvgpr_read_b32 v2, a10
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a9
+; GREEDY908-NEXT: v_accvgpr_read_b32 v0, a8
+; GREEDY908-NEXT: s_nop 1
+; GREEDY908-NEXT: global_store_dwordx4 v4, v[0:3], s[34:35] offset:32
+; GREEDY908-NEXT: s_nop 0
+; GREEDY908-NEXT: v_accvgpr_read_b32 v3, a15
+; GREEDY908-NEXT: v_accvgpr_read_b32 v2, a14
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a13
+; GREEDY908-NEXT: v_accvgpr_read_b32 v0, a12
+; GREEDY908-NEXT: s_nop 1
+; GREEDY908-NEXT: global_store_dwordx4 v4, v[0:3], s[34:35] offset:48
+; GREEDY908-NEXT: s_nop 0
+; GREEDY908-NEXT: v_accvgpr_read_b32 v3, a3
+; GREEDY908-NEXT: v_accvgpr_read_b32 v2, a2
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a1
+; GREEDY908-NEXT: v_accvgpr_read_b32 v0, a0
+; GREEDY908-NEXT: s_nop 1
+; GREEDY908-NEXT: global_store_dwordx4 v4, v[0:3], s[34:35]
+; GREEDY908-NEXT: s_nop 0
+; GREEDY908-NEXT: v_accvgpr_read_b32 v3, a7
+; GREEDY908-NEXT: v_accvgpr_read_b32 v2, a6
+; GREEDY908-NEXT: v_accvgpr_read_b32 v1, a5
+; GREEDY908-NEXT: v_accvgpr_read_b32 v0, a4
+; GREEDY908-NEXT: s_nop 1
+; GREEDY908-NEXT: global_store_dwordx4 v4, v[0:3], s[34:35] offset:16
+; GREEDY908-NEXT: s_endpgm
+;
+; GREEDY90A-LABEL: test_mfma_f32_32x32x1f32:
+; GREEDY90A: ; %bb.0: ; %bb
+; GREEDY90A-NEXT: s_load_dwordx2 s[34:35], s[4:5], 0x24
+; GREEDY90A-NEXT: v_mov_b32_e32 v0, 1.0
+; GREEDY90A-NEXT: v_mov_b32_e32 v1, 2.0
+; GREEDY90A-NEXT: v_mov_b32_e32 v2, 0
+; GREEDY90A-NEXT: s_waitcnt lgkmcnt(0)
+; GREEDY90A-NEXT: s_load_dwordx16 s[16:31], s[34:35], 0x0
+; GREEDY90A-NEXT: s_load_dwordx16 s[0:15], s[34:35], 0x40
+; GREEDY90A-NEXT: s_waitcnt lgkmcnt(0)
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a0, s16
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a1, s17
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a2, s18
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a3, s19
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a4, s20
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a5, s21
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a6, s22
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a7, s23
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a8, s24
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a9, s25
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a10, s26
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a11, s27
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a12, s28
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a13, s29
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a14, s30
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a15, s31
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a16, s0
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a17, s1
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a18, s2
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a19, s3
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a20, s4
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a21, s5
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a22, s6
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a23, s7
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a24, s8
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a25, s9
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a26, s10
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a27, s11
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a28, s12
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a29, s13
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a30, s14
+; GREEDY90A-NEXT: v_accvgpr_write_b32 a31, s15
+; GREEDY90A-NEXT: s_nop 1
+; GREEDY90A-NEXT: v_mfma_f32_32x32x1f32 a[0:31], v0, v1, a[0:31]
+; GREEDY90A-NEXT: v_mfma_f32_32x32x1f32 a[32:63], v0, v1, a[0:31]
+; GREEDY90A-NEXT: s_nop 7
+; GREEDY90A-NEXT: s_nop 7
+; GREEDY90A-NEXT: s_nop 2
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a2, a32
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a3, a33
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a4, a34
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a5, a35
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a6, a36
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a7, a37
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a8, a38
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a9, a39
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a10, a40
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a11, a41
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a12, a42
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a13, a43
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a14, a44
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a15, a45
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a16, a46
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a17, a47
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a18, a48
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a19, a49
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a20, a50
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a21, a51
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a22, a52
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a23, a53
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a24, a54
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a25, a55
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a26, a56
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a27, a57
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a28, a58
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a29, a59
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a30, a60
+; GREEDY90A-NEXT: v_accvgpr_mov_b32 a31, a61
+; GREEDY90A-NEXT: s_nop 1
+; GREEDY90A-NEXT: v_mfma_f32_32x32x1f32 a[0:31], v0, v1, a[0:31]
+; GREEDY90A-NEXT: s_nop 7
+; GREEDY90A-NEXT: s_nop 7
+; GREEDY90A-NEXT: s_nop 2
+; GREEDY90A-NEXT: global_store_dwordx4 v2, a[24:27], s[34:35] offset:96
+; GREEDY90A-NEXT: global_store_dwordx4 v2, a[28:31], s[34:35] offset:112
+; GREEDY90A-NEXT: global_store_dwordx4 v2, a[16:19], s[34:35] offset:64
+; GREEDY90A-NEXT: global_store_dwordx4 v2, a[20:23], s[34:35] offset:80
+; GREEDY90A-NEXT: global_store_dwordx4 v2, a[8:11], s[34:35] offset:32
+; GREEDY90A-NEXT: global_store_dwordx4 v2, a[12:15], s[34:35] offset:48
+; GREEDY90A-NEXT: global_store_dwordx4 v2, a[0:3], s[34:35]
+; GREEDY90A-NEXT: global_store_dwordx4 v2, a[4:7], s[34:35] offset:16
+; GREEDY90A-NEXT: s_endpgm
+;
+; GREEDY942-LABEL: test_mfma_f32_32x32x1f32:
+; GREEDY942: ; %bb.0: ; %bb
+; GREEDY942-NEXT: s_load_dwordx2 s[34:35], s[4:5], 0x24
+; GREEDY942-NEXT: v_mov_b32_e32 v0, 1.0
+; GREEDY942-NEXT: v_mov_b32_e32 v1, 2.0
+; GREEDY942-NEXT: v_mov_b32_e32 v2, 0
+; GREEDY942-NEXT: s_waitcnt lgkmcnt(0)
+; GREEDY942-NEXT: s_load_dwordx16 s[16:31], s[34:35], 0x0
+; GREEDY942-NEXT: s_load_dwordx16 s[0:15], s[34:35], 0x40
+; GREEDY942-NEXT: s_waitcnt lgkmcnt(0)
+; GREEDY942-NEXT: v_accvgpr_write_b32 a0, s16
+; GREEDY942-NEXT: v_accvgpr_write_b32 a1, s17
+; GREEDY942-NEXT: v_accvgpr_write_b32 a2, s18
+; GREEDY942-NEXT: v_accvgpr_write_b32 a3, s19
+; GREEDY942-NEXT: v_accvgpr_write_b32 a4, s20
+; GREEDY942-NEXT: v_accvgpr_write_b32 a5, s21
+; GREEDY942-NEXT: v_accvgpr_write_b32 a6, s22
+; GREEDY942-NEXT: v_accvgpr_write_b32 a7, s23
+; GREEDY942-NEXT: v_accvgpr_write_b32 a8, s24
+; GREEDY942-NEXT: v_accvgpr_write_b32 a9, s25
+; GREEDY942-NEXT: v_accvgpr_write_b32 a10, s26
+; GREEDY942-NEXT: v_accvgpr_write_b32 a11, s27
+; GREEDY942-NEXT: v_accvgpr_write_b32 a12, s28
+; GREEDY942-NEXT: v_accvgpr_write_b32 a13, s29
+; GREEDY942-NEXT: v_accvgpr_write_b32 a14, s30
+; GREEDY942-NEXT: v_accvgpr_write_b32 a15, s31
+; GREEDY942-NEXT: v_accvgpr_write_b32 a16, s0
+; GREEDY942-NEXT: v_accvgpr_write_b32 a17, s1
+; GREEDY942-NEXT: v_accvgpr_write_b32 a18, s2
+; GREEDY942-NEXT: v_accvgpr_write_b32 a19, s3
+; GREEDY942-NEXT: v_accvgpr_write_b32 a20, s4
+; GREEDY942-NEXT: v_accvgpr_write_b32 a21, s5
+; GREEDY942-NEXT: v_accvgpr_write_b32 a22, s6
+; GREEDY942-NEXT: v_accvgpr_write_b32 a23, s7
+; GREEDY942-NEXT: v_accvgpr_write_b32 a24, s8
+; GREEDY942-NEXT: v_accvgpr_write_b32 a25, s9
+; GREEDY942-NEXT: v_accvgpr_write_b32 a26, s10
+; GREEDY942-NEXT: v_accvgpr_write_b32 a27, s11
+; GREEDY942-NEXT: v_accvgpr_write_b32 a28, s12
+; GREEDY942-NEXT: v_accvgpr_write_b32 a29, s13
+; GREEDY942-NEXT: v_accvgpr_write_b32 a30, s14
+; GREEDY942-NEXT: v_accvgpr_write_b32 a31, s15
+; GREEDY942-NEXT: s_nop 1
+; GREEDY942-NEXT: v_mfma_f32_32x32x1_2b_f32 a[0:31], v0, v1, a[0:31]
+; GREEDY942-NEXT: v_mfma_f32_32x32x1_2b_f32 a[32:63], v0, v1, a[0:31]
+; GREEDY942-NEXT: s_nop 7
+; GREEDY942-NEXT: s_nop 7
+; GREEDY942-NEXT: s_nop 1
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a2, a32
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a3, a33
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a4, a34
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a5, a35
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a6, a36
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a7, a37
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a8, a38
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a9, a39
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a10, a40
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a11, a41
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a12, a42
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a13, a43
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a14, a44
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a15, a45
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a16, a46
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a17, a47
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a18, a48
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a19, a49
+; GREEDY942-NEXT: v_accvgpr_mov_b32 a20, a50
+; GREEDY942-NEXT: v_accvgpr_mov_b...
[truncated]
|
shiltian
approved these changes
Mar 19, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
For an upcoming RegisterCoalescer PR