Skip to content

Commit 7639242

Browse files
authored
[AMDGPU] Create new directive .amdhsa_inst_pref_size (#126622)
The field INST_PREF_SIZE is available since gfx11.
1 parent 2ad9d5f commit 7639242

File tree

10 files changed

+60
-13
lines changed

10 files changed

+60
-13
lines changed

llvm/docs/AMDGPUUsage.rst

+3
Original file line numberDiff line numberDiff line change
@@ -18213,6 +18213,9 @@ terminated by an ``.end_amdhsa_kernel`` directive.
1821318213
:ref:`amdgpu-amdhsa-compute_pgm_rsrc1-gfx6-gfx12-table`.
1821418214
``.amdhsa_shared_vgpr_count`` 0 GFX10-GFX11 Controls SHARED_VGPR_COUNT in
1821518215
:ref:`amdgpu-amdhsa-compute_pgm_rsrc3-gfx10-gfx11-table`.
18216+
``.amdhsa_inst_pref_size`` 0 GFX11-GFX12 Controls INST_PREF_SIZE in
18217+
:ref:`amdgpu-amdhsa-compute_pgm_rsrc3-gfx10-gfx11-table` or
18218+
:ref:`amdgpu-amdhsa-compute_pgm_rsrc3-gfx12-table`
1821618219
``.amdhsa_exception_fp_ieee_invalid_op`` 0 GFX6-GFX12 Controls ENABLE_EXCEPTION_IEEE_754_FP_INVALID_OPERATION in
1821718220
:ref:`amdgpu-amdhsa-compute_pgm_rsrc2-gfx6-gfx12-table`.
1821818221
``.amdhsa_exception_fp_denorm_src`` 0 GFX6-GFX12 Controls ENABLE_EXCEPTION_FP_DENORMAL_SOURCE in

llvm/lib/Target/AMDGPU/AsmParser/AMDGPUAsmParser.cpp

+12
Original file line numberDiff line numberDiff line change
@@ -5876,6 +5876,18 @@ bool AMDGPUAsmParser::ParseDirectiveAMDHSAKernel() {
58765876
PARSE_BITS_ENTRY(KD.compute_pgm_rsrc3,
58775877
COMPUTE_PGM_RSRC3_GFX10_GFX11_SHARED_VGPR_COUNT, ExprVal,
58785878
ValRange);
5879+
} else if (ID == ".amdhsa_inst_pref_size") {
5880+
if (IVersion.Major < 11)
5881+
return Error(IDRange.Start, "directive requires gfx11+", IDRange);
5882+
if (IVersion.Major == 11) {
5883+
PARSE_BITS_ENTRY(KD.compute_pgm_rsrc3,
5884+
COMPUTE_PGM_RSRC3_GFX11_INST_PREF_SIZE, ExprVal,
5885+
ValRange);
5886+
} else {
5887+
PARSE_BITS_ENTRY(KD.compute_pgm_rsrc3,
5888+
COMPUTE_PGM_RSRC3_GFX12_PLUS_INST_PREF_SIZE, ExprVal,
5889+
ValRange);
5890+
}
58795891
} else if (ID == ".amdhsa_exception_fp_ieee_invalid_op") {
58805892
PARSE_BITS_ENTRY(
58815893
KD.compute_pgm_rsrc2,

llvm/lib/Target/AMDGPU/Disassembler/AMDGPUDisassembler.cpp

+4-4
Original file line numberDiff line numberDiff line change
@@ -2233,15 +2233,15 @@ Expected<bool> AMDGPUDisassembler::decodeCOMPUTE_PGM_RSRC3(
22332233

22342234
// Bits [4-11].
22352235
if (isGFX11()) {
2236-
PRINT_PSEUDO_DIRECTIVE_COMMENT("INST_PREF_SIZE",
2237-
COMPUTE_PGM_RSRC3_GFX11_INST_PREF_SIZE);
2236+
PRINT_DIRECTIVE(".amdhsa_inst_pref_size",
2237+
COMPUTE_PGM_RSRC3_GFX11_INST_PREF_SIZE);
22382238
PRINT_PSEUDO_DIRECTIVE_COMMENT("TRAP_ON_START",
22392239
COMPUTE_PGM_RSRC3_GFX11_TRAP_ON_START);
22402240
PRINT_PSEUDO_DIRECTIVE_COMMENT("TRAP_ON_END",
22412241
COMPUTE_PGM_RSRC3_GFX11_TRAP_ON_END);
22422242
} else if (isGFX12Plus()) {
2243-
PRINT_PSEUDO_DIRECTIVE_COMMENT(
2244-
"INST_PREF_SIZE", COMPUTE_PGM_RSRC3_GFX12_PLUS_INST_PREF_SIZE);
2243+
PRINT_DIRECTIVE(".amdhsa_inst_pref_size",
2244+
COMPUTE_PGM_RSRC3_GFX12_PLUS_INST_PREF_SIZE);
22452245
} else {
22462246
CHECK_RESERVED_BITS_DESC_MSG(COMPUTE_PGM_RSRC3_GFX10_RESERVED1,
22472247
"COMPUTE_PGM_RSRC3",

llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.cpp

+10
Original file line numberDiff line numberDiff line change
@@ -579,7 +579,17 @@ void AMDGPUTargetAsmStreamer::EmitAmdhsaKernelDescriptor(
579579
amdhsa::COMPUTE_PGM_RSRC3_GFX10_GFX11_SHARED_VGPR_COUNT,
580580
".amdhsa_shared_vgpr_count");
581581
}
582+
if (IVersion.Major == 11) {
583+
PrintField(KD.compute_pgm_rsrc3,
584+
amdhsa::COMPUTE_PGM_RSRC3_GFX11_INST_PREF_SIZE_SHIFT,
585+
amdhsa::COMPUTE_PGM_RSRC3_GFX11_INST_PREF_SIZE,
586+
".amdhsa_inst_pref_size");
587+
}
582588
if (IVersion.Major >= 12) {
589+
PrintField(KD.compute_pgm_rsrc3,
590+
amdhsa::COMPUTE_PGM_RSRC3_GFX12_PLUS_INST_PREF_SIZE_SHIFT,
591+
amdhsa::COMPUTE_PGM_RSRC3_GFX12_PLUS_INST_PREF_SIZE,
592+
".amdhsa_inst_pref_size");
583593
PrintField(KD.compute_pgm_rsrc1,
584594
amdhsa::COMPUTE_PGM_RSRC1_GFX12_PLUS_ENABLE_WG_RR_EN_SHIFT,
585595
amdhsa::COMPUTE_PGM_RSRC1_GFX12_PLUS_ENABLE_WG_RR_EN,

llvm/test/MC/AMDGPU/hsa-diag-v4.s

+10
Original file line numberDiff line numberDiff line change
@@ -280,6 +280,16 @@
280280
.amdhsa_shared_vgpr_count 15
281281
.end_amdhsa_kernel
282282

283+
// GCN-LABEL: warning: test_amdhsa_inst_pref_size_invalid
284+
// PREGFX10: error: directive requires gfx11+
285+
// NONAMDHSA: error: unknown directive
286+
.warning "test_amdhsa_inst_pref_size_invalid"
287+
.amdhsa_kernel test_amdhsa_inst_pref_size_invalid
288+
.amdhsa_next_free_vgpr 273
289+
.amdhsa_next_free_sgpr 0
290+
.amdhsa_inst_pref_size 15
291+
.end_amdhsa_kernel
292+
283293
// GCN-LABEL: warning: test_next_free_vgpr_invalid
284294
// AMDHSA: error: .amdgcn.next_free_{v,s}gpr symbols must be absolute expressions
285295
// NONAMDHSA-NOT: error:

llvm/test/MC/AMDGPU/hsa-gfx12-v4.s

+3-1
Original file line numberDiff line numberDiff line change
@@ -33,7 +33,7 @@
3333
// complete
3434
// OBJDUMP-NEXT: 0040 01000000 01000000 08000000 00000000
3535
// OBJDUMP-NEXT: 0050 00000000 00000000 00000000 00000000
36-
// OBJDUMP-NEXT: 0060 00000000 00000000 00000000 00000000
36+
// OBJDUMP-NEXT: 0060 00000000 00000000 00000000 f00f0000
3737
// OBJDUMP-NEXT: 0070 015021e4 1f0f007f 5e040000 00000000
3838
// special_sgpr
3939
// OBJDUMP-NEXT: 0080 00000000 00000000 00000000 00000000
@@ -120,6 +120,7 @@ disabled_user_sgpr:
120120
.amdhsa_workgroup_processor_mode 1
121121
.amdhsa_memory_ordered 1
122122
.amdhsa_forward_progress 1
123+
.amdhsa_inst_pref_size 255
123124
.amdhsa_round_robin_scheduling 1
124125
.amdhsa_exception_fp_ieee_invalid_op 1
125126
.amdhsa_exception_fp_denorm_src 1
@@ -158,6 +159,7 @@ disabled_user_sgpr:
158159
// ASM-NEXT: .amdhsa_workgroup_processor_mode 1
159160
// ASM-NEXT: .amdhsa_memory_ordered 1
160161
// ASM-NEXT: .amdhsa_forward_progress 1
162+
// ASM-NEXT: .amdhsa_inst_pref_size 255
161163
// ASM-NEXT: .amdhsa_round_robin_scheduling 1
162164
// ASM-NEXT: .amdhsa_exception_fp_ieee_invalid_op 1
163165
// ASM-NEXT: .amdhsa_exception_fp_denorm_src 1

llvm/test/MC/AMDGPU/hsa-sym-exprs-gfx11.s

+2
Original file line numberDiff line numberDiff line change
@@ -133,6 +133,7 @@ expr_defined:
133133
// ASM-NEXT: .amdhsa_memory_ordered (((((((((((((((((((1621884928|(defined_2_bits<<12))&(~49152))|(defined_2_bits<<14))&(~196608))|(defined_2_bits<<16))&(~786432))|(defined_2_bits<<18))&(~67108864))|(defined_boolean<<26))&(~536870912))|(defined_boolean<<29))&(~1073741824))|(defined_boolean<<30))&(~2147483648))|(defined_boolean<<31))&(~63))|(((alignto(max(defined_value+4, 1), 8))/8)-1))&(~960))&1073741824)>>30
134134
// ASM-NEXT: .amdhsa_forward_progress (((((((((((((((((((1621884928|(defined_2_bits<<12))&(~49152))|(defined_2_bits<<14))&(~196608))|(defined_2_bits<<16))&(~786432))|(defined_2_bits<<18))&(~67108864))|(defined_boolean<<26))&(~536870912))|(defined_boolean<<29))&(~1073741824))|(defined_boolean<<30))&(~2147483648))|(defined_boolean<<31))&(~63))|(((alignto(max(defined_value+4, 1), 8))/8)-1))&(~960))&2147483648)>>31
135135
// ASM-NEXT: .amdhsa_shared_vgpr_count 0
136+
// ASM-NEXT: .amdhsa_inst_pref_size 0
136137
// ASM-NEXT: .amdhsa_exception_fp_ieee_invalid_op (((((((((((((((((((((((((((128|(defined_2_bits<<11))&(~128))|(defined_boolean<<7))&(~256))|(defined_boolean<<8))&(~512))|(defined_boolean<<9))&(~1024))|(defined_boolean<<10))&(~16777216))|(defined_boolean<<24))&(~33554432))|(defined_boolean<<25))&(~67108864))|(defined_boolean<<26))&(~134217728))|(defined_boolean<<27))&(~268435456))|(defined_boolean<<28))&(~536870912))|(defined_boolean<<29))&(~1073741824))|(defined_boolean<<30))&(~1))|defined_boolean)&(~62))&16777216)>>24
137138
// ASM-NEXT: .amdhsa_exception_fp_denorm_src (((((((((((((((((((((((((((128|(defined_2_bits<<11))&(~128))|(defined_boolean<<7))&(~256))|(defined_boolean<<8))&(~512))|(defined_boolean<<9))&(~1024))|(defined_boolean<<10))&(~16777216))|(defined_boolean<<24))&(~33554432))|(defined_boolean<<25))&(~67108864))|(defined_boolean<<26))&(~134217728))|(defined_boolean<<27))&(~268435456))|(defined_boolean<<28))&(~536870912))|(defined_boolean<<29))&(~1073741824))|(defined_boolean<<30))&(~1))|defined_boolean)&(~62))&33554432)>>25
138139
// ASM-NEXT: .amdhsa_exception_fp_ieee_div_zero (((((((((((((((((((((((((((128|(defined_2_bits<<11))&(~128))|(defined_boolean<<7))&(~256))|(defined_boolean<<8))&(~512))|(defined_boolean<<9))&(~1024))|(defined_boolean<<10))&(~16777216))|(defined_boolean<<24))&(~33554432))|(defined_boolean<<25))&(~67108864))|(defined_boolean<<26))&(~134217728))|(defined_boolean<<27))&(~268435456))|(defined_boolean<<28))&(~536870912))|(defined_boolean<<29))&(~1073741824))|(defined_boolean<<30))&(~1))|defined_boolean)&(~62))&67108864)>>26
@@ -180,6 +181,7 @@ expr_defined:
180181
// ASM-NEXT: .amdhsa_memory_ordered 1
181182
// ASM-NEXT: .amdhsa_forward_progress 1
182183
// ASM-NEXT: .amdhsa_shared_vgpr_count 0
184+
// ASM-NEXT: .amdhsa_inst_pref_size 0
183185
// ASM-NEXT: .amdhsa_exception_fp_ieee_invalid_op 1
184186
// ASM-NEXT: .amdhsa_exception_fp_denorm_src 1
185187
// ASM-NEXT: .amdhsa_exception_fp_ieee_div_zero 1

llvm/test/MC/AMDGPU/hsa-sym-exprs-gfx12.s

+6-2
Original file line numberDiff line numberDiff line change
@@ -9,12 +9,12 @@
99
// expr_defined_later
1010
// OBJDUMP-NEXT: 0000 2b000000 2c000000 00000000 00000000
1111
// OBJDUMP-NEXT: 0010 00000000 00000000 00000000 00000000
12-
// OBJDUMP-NEXT: 0020 00000000 00000000 00000000 00000000
12+
// OBJDUMP-NEXT: 0020 00000000 00000000 00000000 f0020000
1313
// OBJDUMP-NEXT: 0030 05f02fe4 811f007f 000c0000 00000000
1414
// expr_defined
1515
// OBJDUMP-NEXT: 0040 2a000000 2b000000 00000000 00000000
1616
// OBJDUMP-NEXT: 0050 00000000 00000000 00000000 00000000
17-
// OBJDUMP-NEXT: 0060 00000000 00000000 00000000 00000000
17+
// OBJDUMP-NEXT: 0060 00000000 00000000 00000000 f0020000
1818
// OBJDUMP-NEXT: 0070 05f02fe4 811f007f 000c0000 00000000
1919

2020
.text
@@ -53,6 +53,7 @@ expr_defined:
5353
.amdhsa_workgroup_processor_mode defined_boolean
5454
.amdhsa_memory_ordered defined_boolean
5555
.amdhsa_forward_progress defined_boolean
56+
.amdhsa_inst_pref_size defined_value+6
5657
.amdhsa_exception_fp_ieee_invalid_op defined_boolean
5758
.amdhsa_exception_fp_denorm_src defined_boolean
5859
.amdhsa_exception_fp_ieee_div_zero defined_boolean
@@ -89,6 +90,7 @@ expr_defined:
8990
.amdhsa_workgroup_processor_mode defined_boolean
9091
.amdhsa_memory_ordered defined_boolean
9192
.amdhsa_forward_progress defined_boolean
93+
.amdhsa_inst_pref_size defined_value+6
9294
.amdhsa_exception_fp_ieee_invalid_op defined_boolean
9395
.amdhsa_exception_fp_denorm_src defined_boolean
9496
.amdhsa_exception_fp_ieee_div_zero defined_boolean
@@ -132,6 +134,7 @@ expr_defined:
132134
// ASM-NEXT: .amdhsa_workgroup_processor_mode (((((((((((((((((((((1611399168|(defined_2_bits<<12))&(~49152))|(defined_2_bits<<14))&(~196608))|(defined_2_bits<<16))&(~786432))|(defined_2_bits<<18))&(~67108864))|(defined_boolean<<26))&(~536870912))|(defined_boolean<<29))&(~1073741824))|(defined_boolean<<30))&(~2147483648))|(defined_boolean<<31))&(~2097152))|(defined_boolean<<21))&(~63))|(((alignto(max(defined_value+4, 1), 8))/8)-1))&(~960))&536870912)>>29
133135
// ASM-NEXT: .amdhsa_memory_ordered (((((((((((((((((((((1611399168|(defined_2_bits<<12))&(~49152))|(defined_2_bits<<14))&(~196608))|(defined_2_bits<<16))&(~786432))|(defined_2_bits<<18))&(~67108864))|(defined_boolean<<26))&(~536870912))|(defined_boolean<<29))&(~1073741824))|(defined_boolean<<30))&(~2147483648))|(defined_boolean<<31))&(~2097152))|(defined_boolean<<21))&(~63))|(((alignto(max(defined_value+4, 1), 8))/8)-1))&(~960))&1073741824)>>30
134136
// ASM-NEXT: .amdhsa_forward_progress (((((((((((((((((((((1611399168|(defined_2_bits<<12))&(~49152))|(defined_2_bits<<14))&(~196608))|(defined_2_bits<<16))&(~786432))|(defined_2_bits<<18))&(~67108864))|(defined_boolean<<26))&(~536870912))|(defined_boolean<<29))&(~1073741824))|(defined_boolean<<30))&(~2147483648))|(defined_boolean<<31))&(~2097152))|(defined_boolean<<21))&(~63))|(((alignto(max(defined_value+4, 1), 8))/8)-1))&(~960))&2147483648)>>31
137+
// ASM-NEXT: .amdhsa_inst_pref_size (((defined_value+6)<<4)&4080)>>4
135138
// ASM-NEXT: .amdhsa_round_robin_scheduling (((((((((((((((((((((1611399168|(defined_2_bits<<12))&(~49152))|(defined_2_bits<<14))&(~196608))|(defined_2_bits<<16))&(~786432))|(defined_2_bits<<18))&(~67108864))|(defined_boolean<<26))&(~536870912))|(defined_boolean<<29))&(~1073741824))|(defined_boolean<<30))&(~2147483648))|(defined_boolean<<31))&(~2097152))|(defined_boolean<<21))&(~63))|(((alignto(max(defined_value+4, 1), 8))/8)-1))&(~960))&2097152)>>21
136139
// ASM-NEXT: .amdhsa_exception_fp_ieee_invalid_op (((((((((((((((((((((((((((128|(defined_2_bits<<11))&(~128))|(defined_boolean<<7))&(~256))|(defined_boolean<<8))&(~512))|(defined_boolean<<9))&(~1024))|(defined_boolean<<10))&(~16777216))|(defined_boolean<<24))&(~33554432))|(defined_boolean<<25))&(~67108864))|(defined_boolean<<26))&(~134217728))|(defined_boolean<<27))&(~268435456))|(defined_boolean<<28))&(~536870912))|(defined_boolean<<29))&(~1073741824))|(defined_boolean<<30))&(~1))|defined_boolean)&(~62))&16777216)>>24
137140
// ASM-NEXT: .amdhsa_exception_fp_denorm_src (((((((((((((((((((((((((((128|(defined_2_bits<<11))&(~128))|(defined_boolean<<7))&(~256))|(defined_boolean<<8))&(~512))|(defined_boolean<<9))&(~1024))|(defined_boolean<<10))&(~16777216))|(defined_boolean<<24))&(~33554432))|(defined_boolean<<25))&(~67108864))|(defined_boolean<<26))&(~134217728))|(defined_boolean<<27))&(~268435456))|(defined_boolean<<28))&(~536870912))|(defined_boolean<<29))&(~1073741824))|(defined_boolean<<30))&(~1))|defined_boolean)&(~62))&33554432)>>25
@@ -177,6 +180,7 @@ expr_defined:
177180
// ASM-NEXT: .amdhsa_workgroup_processor_mode 1
178181
// ASM-NEXT: .amdhsa_memory_ordered 1
179182
// ASM-NEXT: .amdhsa_forward_progress 1
183+
// ASM-NEXT: .amdhsa_inst_pref_size 47
180184
// ASM-NEXT: .amdhsa_round_robin_scheduling 1
181185
// ASM-NEXT: .amdhsa_exception_fp_ieee_invalid_op 1
182186
// ASM-NEXT: .amdhsa_exception_fp_denorm_src 1

llvm/test/tools/llvm-objdump/ELF/AMDGPU/kd-gfx11.s

+7-4
Original file line numberDiff line numberDiff line change
@@ -13,7 +13,7 @@
1313
; CHECK-NEXT: .amdhsa_private_segment_fixed_size 0
1414
; CHECK-NEXT: .amdhsa_kernarg_size 0
1515
; CHECK-NEXT: ; SHARED_VGPR_COUNT 0
16-
; CHECK-NEXT: ; INST_PREF_SIZE 0
16+
; CHECK-NEXT: .amdhsa_inst_pref_size 0
1717
; CHECK-NEXT: ; TRAP_ON_START 0
1818
; CHECK-NEXT: ; TRAP_ON_END 0
1919
; CHECK-NEXT: ; IMAGE_OP 0
@@ -70,7 +70,7 @@
7070
; CHECK-NEXT: .amdhsa_private_segment_fixed_size 0
7171
; CHECK-NEXT: .amdhsa_kernarg_size 0
7272
; CHECK-NEXT: .amdhsa_shared_vgpr_count 0
73-
; CHECK-NEXT: ; INST_PREF_SIZE 0
73+
; CHECK-NEXT: .amdhsa_inst_pref_size 0
7474
; CHECK-NEXT: ; TRAP_ON_START 0
7575
; CHECK-NEXT: ; TRAP_ON_END 0
7676
; CHECK-NEXT: ; IMAGE_OP 0
@@ -114,6 +114,7 @@
114114
.amdhsa_next_free_vgpr 32
115115
.amdhsa_next_free_sgpr 32
116116
.amdhsa_shared_vgpr_count 0
117+
.amdhsa_inst_pref_size 0
117118
.end_amdhsa_kernel
118119

119120
;--- 3.s
@@ -127,7 +128,7 @@
127128
; CHECK-NEXT: .amdhsa_private_segment_fixed_size 0
128129
; CHECK-NEXT: .amdhsa_kernarg_size 0
129130
; CHECK-NEXT: .amdhsa_shared_vgpr_count 1
130-
; CHECK-NEXT: ; INST_PREF_SIZE 0
131+
; CHECK-NEXT: .amdhsa_inst_pref_size 63
131132
; CHECK-NEXT: ; TRAP_ON_START 0
132133
; CHECK-NEXT: ; TRAP_ON_END 0
133134
; CHECK-NEXT: ; IMAGE_OP 0
@@ -171,6 +172,7 @@
171172
.amdhsa_next_free_vgpr 32
172173
.amdhsa_next_free_sgpr 32
173174
.amdhsa_shared_vgpr_count 1
175+
.amdhsa_inst_pref_size 63
174176
.end_amdhsa_kernel
175177

176178
;--- 4.s
@@ -184,7 +186,7 @@
184186
; CHECK-NEXT: .amdhsa_private_segment_fixed_size 0
185187
; CHECK-NEXT: .amdhsa_kernarg_size 0
186188
; CHECK-NEXT: .amdhsa_shared_vgpr_count 1
187-
; CHECK-NEXT: ; INST_PREF_SIZE 0
189+
; CHECK-NEXT: .amdhsa_inst_pref_size 63
188190
; CHECK-NEXT: ; TRAP_ON_START 0
189191
; CHECK-NEXT: ; TRAP_ON_END 0
190192
; CHECK-NEXT: ; IMAGE_OP 0
@@ -228,5 +230,6 @@
228230
.amdhsa_next_free_vgpr 32
229231
.amdhsa_next_free_sgpr 32
230232
.amdhsa_shared_vgpr_count 1
233+
.amdhsa_inst_pref_size 63
231234
.amdhsa_wavefront_size32 0
232235
.end_amdhsa_kernel

llvm/test/tools/llvm-objdump/ELF/AMDGPU/kd-gfx12.s

+3-2
Original file line numberDiff line numberDiff line change
@@ -12,7 +12,7 @@
1212
; CHECK-NEXT: .amdhsa_group_segment_fixed_size 0
1313
; CHECK-NEXT: .amdhsa_private_segment_fixed_size 0
1414
; CHECK-NEXT: .amdhsa_kernarg_size 0
15-
; CHECK-NEXT: ; INST_PREF_SIZE 0
15+
; CHECK-NEXT: .amdhsa_inst_pref_size 0
1616
; CHECK-NEXT: ; GLG_EN 0
1717
; CHECK-NEXT: ; IMAGE_OP 0
1818
; CHECK-NEXT: .amdhsa_next_free_vgpr 32
@@ -66,7 +66,7 @@
6666
; CHECK-NEXT: .amdhsa_group_segment_fixed_size 0
6767
; CHECK-NEXT: .amdhsa_private_segment_fixed_size 0
6868
; CHECK-NEXT: .amdhsa_kernarg_size 0
69-
; CHECK-NEXT: ; INST_PREF_SIZE 0
69+
; CHECK-NEXT: .amdhsa_inst_pref_size 255
7070
; CHECK-NEXT: ; GLG_EN 0
7171
; CHECK-NEXT: ; IMAGE_OP 0
7272
; CHECK-NEXT: .amdhsa_next_free_vgpr 32
@@ -108,4 +108,5 @@
108108
.amdhsa_next_free_vgpr 32
109109
.amdhsa_next_free_sgpr 32
110110
.amdhsa_wavefront_size32 0
111+
.amdhsa_inst_pref_size 255
111112
.end_amdhsa_kernel

0 commit comments

Comments
 (0)