[llvm-branch-commits] [llvm] 03663e4 - [AMDGPU] Add occupancy level tests for GFX10.3. NFC.
Jay Foad via llvm-branch-commits
llvm-branch-commits at lists.llvm.org
Tue Dec 8 06:31:20 PST 2020
Author: Jay Foad
Date: 2020-12-08T14:15:01Z
New Revision: 03663e4130d700c6c8ea28b357fcac4d31b617f7
URL: https://github.com/llvm/llvm-project/commit/03663e4130d700c6c8ea28b357fcac4d31b617f7
DIFF: https://github.com/llvm/llvm-project/commit/03663e4130d700c6c8ea28b357fcac4d31b617f7.diff
LOG: [AMDGPU] Add occupancy level tests for GFX10.3. NFC.
getMaxWavesPerEU and getVGPRAllocGranule both changed in GFX10.3 and
they both affect the occupancy calculation.
Differential Revision: https://reviews.llvm.org/D92839
Added:
Modified:
llvm/test/CodeGen/AMDGPU/occupancy-levels.ll
Removed:
################################################################################
diff --git a/llvm/test/CodeGen/AMDGPU/occupancy-levels.ll b/llvm/test/CodeGen/AMDGPU/occupancy-levels.ll
index db70c3d9387d..25e0376dd7ee 100644
--- a/llvm/test/CodeGen/AMDGPU/occupancy-levels.ll
+++ b/llvm/test/CodeGen/AMDGPU/occupancy-levels.ll
@@ -1,18 +1,21 @@
; RUN: llc -march=amdgcn -mcpu=gfx900 < %s | FileCheck --check-prefixes=GCN,GFX9 %s
-; RUN: llc -march=amdgcn -mcpu=gfx1010 < %s | FileCheck --check-prefixes=GCN,GFX1010,GFX1010W32 %s
-; RUN: llc -march=amdgcn -mcpu=gfx1010 -mattr=+wavefrontsize64 < %s | FileCheck --check-prefixes=GCN,GFX1010,GFX1010W64 %s
+; RUN: llc -march=amdgcn -mcpu=gfx1010 < %s | FileCheck --check-prefixes=GCN,GFX10,GFX10W32,GFX1010,GFX1010W32 %s
+; RUN: llc -march=amdgcn -mcpu=gfx1010 -mattr=+wavefrontsize64 < %s | FileCheck --check-prefixes=GCN,GFX10,GFX10W64,GFX1010,GFX1010W64 %s
+; RUN: llc -march=amdgcn -mcpu=gfx1030 < %s | FileCheck --check-prefixes=GCN,GFX10,GFX10W32,GFX1030,GFX1030W32 %s
+; RUN: llc -march=amdgcn -mcpu=gfx1030 -mattr=+wavefrontsize64 < %s | FileCheck --check-prefixes=GCN,GFX10,GFX10W64,GFX1030,GFX1030W64 %s
; GCN-LABEL: {{^}}max_occupancy:
; GFX9: ; Occupancy: 10
; GFX1010: ; Occupancy: 20
+; GFX1030: ; Occupancy: 16
define amdgpu_kernel void @max_occupancy() {
ret void
}
; GCN-LABEL: {{^}}limited_occupancy_3:
; GFX9: ; Occupancy: 3
-; GFX1010W64: ; Occupancy: 3
-; GFX1010W32: ; Occupancy: 4
+; GFX10W64: ; Occupancy: 3
+; GFX10W32: ; Occupancy: 4
define amdgpu_kernel void @limited_occupancy_3() #0 {
ret void
}
@@ -20,6 +23,7 @@ define amdgpu_kernel void @limited_occupancy_3() #0 {
; GCN-LABEL: {{^}}limited_occupancy_18:
; GFX9: ; Occupancy: 10
; GFX1010: ; Occupancy: 18
+; GFX1030: ; Occupancy: 16
define amdgpu_kernel void @limited_occupancy_18() #1 {
ret void
}
@@ -27,6 +31,7 @@ define amdgpu_kernel void @limited_occupancy_18() #1 {
; GCN-LABEL: {{^}}limited_occupancy_19:
; GFX9: ; Occupancy: 10
; GFX1010: ; Occupancy: 18
+; GFX1030: ; Occupancy: 16
define amdgpu_kernel void @limited_occupancy_19() #2 {
ret void
}
@@ -34,6 +39,7 @@ define amdgpu_kernel void @limited_occupancy_19() #2 {
; GCN-LABEL: {{^}}used_24_vgprs:
; GFX9: ; Occupancy: 10
; GFX1010: ; Occupancy: 20
+; GFX1030: ; Occupancy: 16
define amdgpu_kernel void @used_24_vgprs() {
call void asm sideeffect "", "~{v23}" ()
ret void
@@ -43,6 +49,7 @@ define amdgpu_kernel void @used_24_vgprs() {
; GFX9: ; Occupancy: 9
; GFX1010W64: ; Occupancy: 18
; GFX1010W32: ; Occupancy: 20
+; GFX1030: ; Occupancy: 16
define amdgpu_kernel void @used_28_vgprs() {
call void asm sideeffect "", "~{v27}" ()
ret void
@@ -50,8 +57,9 @@ define amdgpu_kernel void @used_28_vgprs() {
; GCN-LABEL: {{^}}used_32_vgprs:
; GFX9: ; Occupancy: 8
-; GFX1010W64: ; Occupancy: 16
+; GFX10W64: ; Occupancy: 16
; GFX1010W32: ; Occupancy: 20
+; GFX1030W32: ; Occupancy: 16
define amdgpu_kernel void @used_32_vgprs() {
call void asm sideeffect "", "~{v31}" ()
ret void
@@ -61,6 +69,8 @@ define amdgpu_kernel void @used_32_vgprs() {
; GFX9: ; Occupancy: 7
; GFX1010W64: ; Occupancy: 14
; GFX1010W32: ; Occupancy: 20
+; GFX1030W64: ; Occupancy: 12
+; GFX1030W32: ; Occupancy: 16
define amdgpu_kernel void @used_36_vgprs() {
call void asm sideeffect "", "~{v35}" ()
ret void
@@ -68,8 +78,9 @@ define amdgpu_kernel void @used_36_vgprs() {
; GCN-LABEL: {{^}}used_40_vgprs:
; GFX9: ; Occupancy: 6
-; GFX1010W64: ; Occupancy: 12
+; GFX10W64: ; Occupancy: 12
; GFX1010W32: ; Occupancy: 20
+; GFX1030W32: ; Occupancy: 16
define amdgpu_kernel void @used_40_vgprs() {
call void asm sideeffect "", "~{v39}" ()
ret void
@@ -79,6 +90,8 @@ define amdgpu_kernel void @used_40_vgprs() {
; GFX9: ; Occupancy: 5
; GFX1010W64: ; Occupancy: 11
; GFX1010W32: ; Occupancy: 20
+; GFX1030W64: ; Occupancy: 10
+; GFX1030W32: ; Occupancy: 16
define amdgpu_kernel void @used_44_vgprs() {
call void asm sideeffect "", "~{v43}" ()
ret void
@@ -86,8 +99,9 @@ define amdgpu_kernel void @used_44_vgprs() {
; GCN-LABEL: {{^}}used_48_vgprs:
; GFX9: ; Occupancy: 5
-; GFX1010W64: ; Occupancy: 10
+; GFX10W64: ; Occupancy: 10
; GFX1010W32: ; Occupancy: 20
+; GFX1030W32: ; Occupancy: 16
define amdgpu_kernel void @used_48_vgprs() {
call void asm sideeffect "", "~{v47}" ()
ret void
@@ -95,8 +109,9 @@ define amdgpu_kernel void @used_48_vgprs() {
; GCN-LABEL: {{^}}used_56_vgprs:
; GFX9: ; Occupancy: 4
-; GFX1010W64: ; Occupancy: 9
+; GFX10W64: ; Occupancy: 9
; GFX1010W32: ; Occupancy: 18
+; GFX1030W32: ; Occupancy: 16
define amdgpu_kernel void @used_56_vgprs() {
call void asm sideeffect "", "~{v55}" ()
ret void
@@ -104,8 +119,8 @@ define amdgpu_kernel void @used_56_vgprs() {
; GCN-LABEL: {{^}}used_64_vgprs:
; GFX9: ; Occupancy: 4
-; GFX1010W64: ; Occupancy: 8
-; GFX1010W32: ; Occupancy: 16
+; GFX10W64: ; Occupancy: 8
+; GFX10W32: ; Occupancy: 16
define amdgpu_kernel void @used_64_vgprs() {
call void asm sideeffect "", "~{v63}" ()
ret void
@@ -113,8 +128,9 @@ define amdgpu_kernel void @used_64_vgprs() {
; GCN-LABEL: {{^}}used_72_vgprs:
; GFX9: ; Occupancy: 3
-; GFX1010W64: ; Occupancy: 7
+; GFX10W64: ; Occupancy: 7
; GFX1010W32: ; Occupancy: 14
+; GFX1030W32: ; Occupancy: 12
define amdgpu_kernel void @used_72_vgprs() {
call void asm sideeffect "", "~{v71}" ()
ret void
@@ -122,8 +138,8 @@ define amdgpu_kernel void @used_72_vgprs() {
; GCN-LABEL: {{^}}used_80_vgprs:
; GFX9: ; Occupancy: 3
-; GFX1010W64: ; Occupancy: 6
-; GFX1010W32: ; Occupancy: 12
+; GFX10W64: ; Occupancy: 6
+; GFX10W32: ; Occupancy: 12
define amdgpu_kernel void @used_80_vgprs() {
call void asm sideeffect "", "~{v79}" ()
ret void
@@ -133,6 +149,8 @@ define amdgpu_kernel void @used_80_vgprs() {
; GFX9: ; Occupancy: 3
; GFX1010W64: ; Occupancy: 6
; GFX1010W32: ; Occupancy: 11
+; GFX1030W64: ; Occupancy: 5
+; GFX1030W32: ; Occupancy: 10
define amdgpu_kernel void @used_84_vgprs() {
call void asm sideeffect "", "~{v83}" ()
ret void
@@ -140,8 +158,9 @@ define amdgpu_kernel void @used_84_vgprs() {
; GCN-LABEL: {{^}}used_88_vgprs:
; GFX9: ; Occupancy: 2
-; GFX1010W64: ; Occupancy: 5
+; GFX10W64: ; Occupancy: 5
; GFX1010W32: ; Occupancy: 11
+; GFX1030W32: ; Occupancy: 10
define amdgpu_kernel void @used_88_vgprs() {
call void asm sideeffect "", "~{v87}" ()
ret void
@@ -149,8 +168,8 @@ define amdgpu_kernel void @used_88_vgprs() {
; GCN-LABEL: {{^}}used_96_vgprs:
; GFX9: ; Occupancy: 2
-; GFX1010W64: ; Occupancy: 5
-; GFX1010W32: ; Occupancy: 10
+; GFX10W64: ; Occupancy: 5
+; GFX10W32: ; Occupancy: 10
define amdgpu_kernel void @used_96_vgprs() {
call void asm sideeffect "", "~{v95}" ()
ret void
@@ -159,7 +178,8 @@ define amdgpu_kernel void @used_96_vgprs() {
; GCN-LABEL: {{^}}used_100_vgprs:
; GFX9: ; Occupancy: 2
; GFX1010W64: ; Occupancy: 5
-; GFX1010W32: ; Occupancy: 9
+; GFX1030W64: ; Occupancy: 4
+; GFX10W32: ; Occupancy: 9
define amdgpu_kernel void @used_100_vgprs() {
call void asm sideeffect "", "~{v99}" ()
ret void
@@ -167,8 +187,8 @@ define amdgpu_kernel void @used_100_vgprs() {
; GCN-LABEL: {{^}}used_112_vgprs:
; GFX9: ; Occupancy: 2
-; GFX1010W64: ; Occupancy: 4
-; GFX1010W32: ; Occupancy: 9
+; GFX10W64: ; Occupancy: 4
+; GFX10W32: ; Occupancy: 9
define amdgpu_kernel void @used_112_vgprs() {
call void asm sideeffect "", "~{v111}" ()
ret void
@@ -176,8 +196,8 @@ define amdgpu_kernel void @used_112_vgprs() {
; GCN-LABEL: {{^}}used_128_vgprs:
; GFX9: ; Occupancy: 2
-; GFX1010W64: ; Occupancy: 4
-; GFX1010W32: ; Occupancy: 8
+; GFX10W64: ; Occupancy: 4
+; GFX10W32: ; Occupancy: 8
define amdgpu_kernel void @used_128_vgprs() {
call void asm sideeffect "", "~{v127}" ()
ret void
@@ -185,8 +205,8 @@ define amdgpu_kernel void @used_128_vgprs() {
; GCN-LABEL: {{^}}used_144_vgprs:
; GFX9: ; Occupancy: 1
-; GFX1010W64: ; Occupancy: 3
-; GFX1010W32: ; Occupancy: 7
+; GFX10W64: ; Occupancy: 3
+; GFX10W32: ; Occupancy: 7
define amdgpu_kernel void @used_144_vgprs() {
call void asm sideeffect "", "~{v143}" ()
ret void
@@ -194,8 +214,9 @@ define amdgpu_kernel void @used_144_vgprs() {
; GCN-LABEL: {{^}}used_168_vgprs:
; GFX9: ; Occupancy: 1
-; GFX1010W64: ; Occupancy: 3
+; GFX10W64: ; Occupancy: 3
; GFX1010W32: ; Occupancy: 6
+; GFX1030W32: ; Occupancy: 5
define amdgpu_kernel void @used_168_vgprs() {
call void asm sideeffect "", "~{v167}" ()
ret void
@@ -203,8 +224,9 @@ define amdgpu_kernel void @used_168_vgprs() {
; GCN-LABEL: {{^}}used_200_vgprs:
; GFX9: ; Occupancy: 1
-; GFX1010W64: ; Occupancy: 2
+; GFX10W64: ; Occupancy: 2
; GFX1010W32: ; Occupancy: 5
+; GFX1030W32: ; Occupancy: 4
define amdgpu_kernel void @used_200_vgprs() {
call void asm sideeffect "", "~{v199}" ()
ret void
@@ -212,8 +234,8 @@ define amdgpu_kernel void @used_200_vgprs() {
; GCN-LABEL: {{^}}used_256_vgprs:
; GFX9: ; Occupancy: 1
-; GFX1010W64: ; Occupancy: 2
-; GFX1010W32: ; Occupancy: 4
+; GFX10W64: ; Occupancy: 2
+; GFX10W32: ; Occupancy: 4
define amdgpu_kernel void @used_256_vgprs() {
call void asm sideeffect "", "~{v255}" ()
ret void
@@ -222,6 +244,7 @@ define amdgpu_kernel void @used_256_vgprs() {
; GCN-LABEL: {{^}}used_80_sgprs:
; GFX9: ; Occupancy: 10
; GFX1010: ; Occupancy: 20
+; GFX1030: ; Occupancy: 16
define amdgpu_kernel void @used_80_sgprs() {
call void asm sideeffect "", "~{s79}" ()
ret void
@@ -230,6 +253,7 @@ define amdgpu_kernel void @used_80_sgprs() {
; GCN-LABEL: {{^}}used_88_sgprs:
; GFX9: ; Occupancy: 9
; GFX1010: ; Occupancy: 20
+; GFX1030: ; Occupancy: 16
define amdgpu_kernel void @used_88_sgprs() {
call void asm sideeffect "", "~{s87}" ()
ret void
@@ -238,6 +262,7 @@ define amdgpu_kernel void @used_88_sgprs() {
; GCN-LABEL: {{^}}used_100_sgprs:
; GFX9: ; Occupancy: 8
; GFX1010: ; Occupancy: 20
+; GFX1030: ; Occupancy: 16
define amdgpu_kernel void @used_100_sgprs() {
call void asm sideeffect "", "~{s99}" ()
ret void
@@ -246,6 +271,7 @@ define amdgpu_kernel void @used_100_sgprs() {
; GCN-LABEL: {{^}}used_101_sgprs:
; GFX9: ; Occupancy: 7
; GFX1010: ; Occupancy: 20
+; GFX1030: ; Occupancy: 16
define amdgpu_kernel void @used_101_sgprs() {
call void asm sideeffect "", "~{s100}" ()
ret void
@@ -254,6 +280,7 @@ define amdgpu_kernel void @used_101_sgprs() {
; GCN-LABEL: {{^}}used_lds_6552:
; GFX9: ; Occupancy: 10
; GFX1010: ; Occupancy: 20
+; GFX1030: ; Occupancy: 16
@lds6552 = internal addrspace(3) global [6552 x i8] undef, align 4
define amdgpu_kernel void @used_lds_6552() {
%p = bitcast [6552 x i8] addrspace(3)* @lds6552 to i8 addrspace(3)*
@@ -263,8 +290,8 @@ define amdgpu_kernel void @used_lds_6552() {
; GCN-LABEL: {{^}}used_lds_6556:
; GFX9: ; Occupancy: 10
-; GFX1010W64: ; Occupancy: 20
-; GFX1010W32: ; Occupancy: 20
+; GFX1010: ; Occupancy: 20
+; GFX1030: ; Occupancy: 16
@lds6556 = internal addrspace(3) global [6556 x i8] undef, align 4
define amdgpu_kernel void @used_lds_6556() {
%p = bitcast [6556 x i8] addrspace(3)* @lds6556 to i8 addrspace(3)*
@@ -274,8 +301,8 @@ define amdgpu_kernel void @used_lds_6556() {
; GCN-LABEL: {{^}}used_lds_13112:
; GFX9: ; Occupancy: 10
-; GFX1010W64: ; Occupancy: 20
-; GFX1010W32: ; Occupancy: 20
+; GFX1010: ; Occupancy: 20
+; GFX1030: ; Occupancy: 16
@lds13112 = internal addrspace(3) global [13112 x i8] undef, align 4
define amdgpu_kernel void @used_lds_13112() {
%p = bitcast [13112 x i8] addrspace(3)* @lds13112 to i8 addrspace(3)*
@@ -285,8 +312,8 @@ define amdgpu_kernel void @used_lds_13112() {
; GCN-LABEL: {{^}}used_lds_8252_max_group_size_64:
; GFX9: ; Occupancy: 7{{$}}
-; GFX101064: ; Occupancy: 7{{$}}
-; GFX1010W32: ; Occupancy: 14{{$}}
+; GFX10W64: ; Occupancy: 7{{$}}
+; GFX10W32: ; Occupancy: 14{{$}}
@lds8252 = internal addrspace(3) global [8252 x i8] undef, align 4
define amdgpu_kernel void @used_lds_8252_max_group_size_64() #3 {
%p = bitcast [8252 x i8] addrspace(3)* @lds8252 to i8 addrspace(3)*
@@ -296,8 +323,9 @@ define amdgpu_kernel void @used_lds_8252_max_group_size_64() #3 {
; GCN-LABEL: {{^}}used_lds_8252_max_group_size_96:
; GFX9: ; Occupancy: 10{{$}}
-; GFX1010W64: ; Occupancy: 14{{$}}
-; GFX1010W32: ; Occupancy: 20{{$}}
+; GFX10W64: ; Occupancy: 14{{$}}
+; GFX1010W32: ; Occupancy: 20{{$}}
+; GFX1030W32: ; Occupancy: 16{{$}}
define amdgpu_kernel void @used_lds_8252_max_group_size_96() #4 {
%p = bitcast [8252 x i8] addrspace(3)* @lds8252 to i8 addrspace(3)*
store volatile i8 1, i8 addrspace(3)* %p
@@ -306,8 +334,9 @@ define amdgpu_kernel void @used_lds_8252_max_group_size_96() #4 {
; GCN-LABEL: {{^}}used_lds_8252_max_group_size_128:
; GFX9: ; Occupancy: 10{{$}}
-; GFX1010W64: ; Occupancy: 14{{$}}
-; GFX1010W32: ; Occupancy: 20{{$}}
+; GFX10W64: ; Occupancy: 14{{$}}
+; GFX1010W32: ; Occupancy: 20{{$}}
+; GFX1030W32: ; Occupancy: 16{{$}}
define amdgpu_kernel void @used_lds_8252_max_group_size_128() #5 {
%p = bitcast [8252 x i8] addrspace(3)* @lds8252 to i8 addrspace(3)*
store volatile i8 1, i8 addrspace(3)* %p
@@ -316,8 +345,8 @@ define amdgpu_kernel void @used_lds_8252_max_group_size_128() #5 {
; GCN-LABEL: {{^}}used_lds_8252_max_group_size_192:
; GFX9: ; Occupancy: 10{{$}}
-; GFX1010W64: ; Occupancy: 20{{$}}
-; GFX1010W32: ; Occupancy: 20{{$}}
+; GFX1010: ; Occupancy: 20{{$}}
+; GFX1030: ; Occupancy: 16{{$}}
define amdgpu_kernel void @used_lds_8252_max_group_size_192() #6 {
%p = bitcast [8252 x i8] addrspace(3)* @lds8252 to i8 addrspace(3)*
store volatile i8 1, i8 addrspace(3)* %p
@@ -326,8 +355,8 @@ define amdgpu_kernel void @used_lds_8252_max_group_size_192() #6 {
; GCN-LABEL: {{^}}used_lds_8252_max_group_size_256:
; GFX9: ; Occupancy: 10{{$}}
-; GFX1010W64: ; Occupancy: 20{{$}}
-; GFX1010W32: ; Occupancy: 20{{$}}
+; GFX1010: ; Occupancy: 20{{$}}
+; GFX1030: ; Occupancy: 16{{$}}
define amdgpu_kernel void @used_lds_8252_max_group_size_256() #7 {
%p = bitcast [8252 x i8] addrspace(3)* @lds8252 to i8 addrspace(3)*
store volatile i8 1, i8 addrspace(3)* %p
@@ -336,8 +365,8 @@ define amdgpu_kernel void @used_lds_8252_max_group_size_256() #7 {
; GCN-LABEL: {{^}}used_lds_8252_max_group_size_512:
; GFX9: ; Occupancy: 10{{$}}
-; GFX1010W64: ; Occupancy: 20{{$}}
-; GFX1010W32: ; Occupancy: 20{{$}}
+; GFX1010: ; Occupancy: 20{{$}}
+; GFX1030: ; Occupancy: 16{{$}}
define amdgpu_kernel void @used_lds_8252_max_group_size_512() #8 {
%p = bitcast [8252 x i8] addrspace(3)* @lds8252 to i8 addrspace(3)*
store volatile i8 1, i8 addrspace(3)* %p
@@ -346,8 +375,8 @@ define amdgpu_kernel void @used_lds_8252_max_group_size_512() #8 {
; GCN-LABEL: {{^}}used_lds_8252_max_group_size_1024:
; GFX9: ; Occupancy: 10{{$}}
-; GFX1010W64: ; Occupancy: 20{{$}}
-; GFX1010W32: ; Occupancy: 20{{$}}
+; GFX1010: ; Occupancy: 20{{$}}
+; GFX1030: ; Occupancy: 16{{$}}
define amdgpu_kernel void @used_lds_8252_max_group_size_1024() #9 {
%p = bitcast [8252 x i8] addrspace(3)* @lds8252 to i8 addrspace(3)*
store volatile i8 1, i8 addrspace(3)* %p
@@ -356,8 +385,7 @@ define amdgpu_kernel void @used_lds_8252_max_group_size_1024() #9 {
; GCN-LABEL: {{^}}used_lds_8252_max_group_size_32:
; GFX9: ; Occupancy: 7{{$}}
-; GFX1010W64: ; Occupancy: 7{{$}}
-; GFX1010W32: ; Occupancy: 7{{$}}
+; GFX10: ; Occupancy: 7{{$}}
define amdgpu_kernel void @used_lds_8252_max_group_size_32() #10 {
%p = bitcast [8252 x i8] addrspace(3)* @lds8252 to i8 addrspace(3)*
store volatile i8 1, i8 addrspace(3)* %p
More information about the llvm-branch-commits
mailing list