15428e0d78
Add target feature for point sample acceleration and enable it for relevant targets. Also add support to insert waitcnts where required when point sample accel may have occurred. This has implications for out of order returns, which is why extra waitcnts are required. Add a VMEM_NOSAMPLER bit in the register masks to determine when waitcnt is required.
81 lines
6.3 KiB
YAML
81 lines
6.3 KiB
YAML
# RUN: llc -mtriple=amdgcn -mcpu=gfx1100 -verify-machineinstrs -run-pass si-insert-waitcnts -o - %s | FileCheck -check-prefixes=GCN,GFX11 %s
|
|
# RUN: llc -mtriple=amdgcn -mcpu=gfx1150 -verify-machineinstrs -run-pass si-insert-waitcnts -o - %s | FileCheck -check-prefixes=GCN,GFX1150 %s
|
|
# RUN: llc -mtriple=amdgcn -mcpu=gfx1200 -verify-machineinstrs -run-pass si-insert-waitcnts -o - %s | FileCheck -check-prefixes=GCN,GFX12 %s
|
|
|
|
---
|
|
name: waitcnt-gather-sample
|
|
machineFunctionInfo:
|
|
isEntryFunction: true
|
|
body: |
|
|
bb.0:
|
|
; GCN-LABEL: name: waitcnt-gather-sample
|
|
; GCN: $vgpr10_vgpr11_vgpr12_vgpr13 = IMAGE_GATHER4_LZ_O_V4_V3 $vgpr0_vgpr1_vgpr2, $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $sgpr8_sgpr9_sgpr10_sgpr11, 1, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (load (s128))
|
|
; GFX1150-NEXT: S_WAITCNT 1015
|
|
; GFX12-NEXT: S_WAIT_SAMPLECNT 0
|
|
; GCN-NEXT: $vgpr13_vgpr14_vgpr15_vgpr16 = IMAGE_SAMPLE_V4_V2 $vgpr20_vgpr21, $sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11, $sgpr0_sgpr1_sgpr2_sgpr3, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
|
|
$vgpr10_vgpr11_vgpr12_vgpr13 = IMAGE_GATHER4_LZ_O_V4_V3 $vgpr0_vgpr1_vgpr2, $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $sgpr8_sgpr9_sgpr10_sgpr11, 1, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (load (s128))
|
|
$vgpr13_vgpr14_vgpr15_vgpr16 = IMAGE_SAMPLE_V4_V2 $vgpr20_vgpr21, $sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11, $sgpr0_sgpr1_sgpr2_sgpr3, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
|
|
S_ENDPGM 0
|
|
...
|
|
---
|
|
name: waitcnt-gather-sample-o
|
|
machineFunctionInfo:
|
|
isEntryFunction: true
|
|
body: |
|
|
bb.0:
|
|
; GCN-LABEL: name: waitcnt-gather-sample-o
|
|
; GCN: $vgpr10_vgpr11_vgpr12_vgpr13 = IMAGE_GATHER4_LZ_O_V4_V3 $vgpr0_vgpr1_vgpr2, $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $sgpr8_sgpr9_sgpr10_sgpr11, 1, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (load (s128))
|
|
; GFX12-NEXT: S_WAIT_SAMPLECNT 0
|
|
; GCN-NEXT: $vgpr13_vgpr14_vgpr15_vgpr16 = IMAGE_SAMPLE_O_V4_V2 $vgpr20_vgpr21, $sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11, $sgpr0_sgpr1_sgpr2_sgpr3, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
|
|
$vgpr10_vgpr11_vgpr12_vgpr13 = IMAGE_GATHER4_LZ_O_V4_V3 $vgpr0_vgpr1_vgpr2, $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $sgpr8_sgpr9_sgpr10_sgpr11, 1, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (load (s128))
|
|
$vgpr13_vgpr14_vgpr15_vgpr16 = IMAGE_SAMPLE_O_V4_V2 $vgpr20_vgpr21, $sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11, $sgpr0_sgpr1_sgpr2_sgpr3, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
|
|
S_ENDPGM 0
|
|
...
|
|
---
|
|
name: waitcnt-sample-gather
|
|
machineFunctionInfo:
|
|
isEntryFunction: true
|
|
body: |
|
|
bb.0:
|
|
; GCN-LABEL: name: waitcnt-sample-gather
|
|
; GCN: $vgpr13_vgpr14_vgpr15_vgpr16 = IMAGE_SAMPLE_V4_V2 $vgpr20_vgpr21, $sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11, $sgpr0_sgpr1_sgpr2_sgpr3, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
|
|
; GFX1150-NEXT: S_WAITCNT 1015
|
|
; GFX12-NEXT: S_WAIT_SAMPLECNT 0
|
|
; GCN-NEXT: $vgpr10_vgpr11_vgpr12_vgpr13 = IMAGE_GATHER4_LZ_O_V4_V3 $vgpr0_vgpr1_vgpr2, $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $sgpr8_sgpr9_sgpr10_sgpr11, 1, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (load (s128))
|
|
$vgpr13_vgpr14_vgpr15_vgpr16 = IMAGE_SAMPLE_V4_V2 $vgpr20_vgpr21, $sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11, $sgpr0_sgpr1_sgpr2_sgpr3, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
|
|
$vgpr10_vgpr11_vgpr12_vgpr13 = IMAGE_GATHER4_LZ_O_V4_V3 $vgpr0_vgpr1_vgpr2, $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, $sgpr8_sgpr9_sgpr10_sgpr11, 1, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (load (s128))
|
|
S_ENDPGM 0
|
|
...
|
|
---
|
|
name: waitcnt-sample-load
|
|
machineFunctionInfo:
|
|
isEntryFunction: true
|
|
body: |
|
|
bb.0:
|
|
; GCN-LABEL: name: waitcnt-sample-load
|
|
; GCN: $vgpr13_vgpr14_vgpr15_vgpr16 = IMAGE_SAMPLE_V4_V2 $vgpr20_vgpr21, $sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11, $sgpr0_sgpr1_sgpr2_sgpr3, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
|
|
; GFX11: S_WAITCNT 1015
|
|
; GFX1150-NEXT: S_WAITCNT 1015
|
|
; GFX12-NEXT: S_WAIT_SAMPLECNT 0
|
|
; GCN-NEXT: renamable $vgpr10_vgpr11_vgpr12_vgpr13 = IMAGE_MSAA_LOAD_V4_V2_gfx11 killed renamable $vgpr4_vgpr5, killed renamable $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, 4, 7, -1, 0, 0, -1, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), addrspace 8)
|
|
$vgpr13_vgpr14_vgpr15_vgpr16 = IMAGE_SAMPLE_V4_V2 $vgpr20_vgpr21, $sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11, $sgpr0_sgpr1_sgpr2_sgpr3, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
|
|
renamable $vgpr10_vgpr11_vgpr12_vgpr13 = IMAGE_MSAA_LOAD_V4_V2_gfx11 killed renamable $vgpr4_vgpr5, killed renamable $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, 4, 7, -1, 0, 0, -1, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), addrspace 8)
|
|
S_ENDPGM 0
|
|
...
|
|
---
|
|
name: waitcnt-load-sample
|
|
machineFunctionInfo:
|
|
isEntryFunction: true
|
|
body: |
|
|
bb.0:
|
|
; GCN-LABEL: name: waitcnt-load-sample
|
|
; GCN: renamable $vgpr10_vgpr11_vgpr12_vgpr13 = IMAGE_MSAA_LOAD_V4_V2_gfx11 killed renamable $vgpr4_vgpr5, killed renamable $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, 4, 7, -1, 0, 0, -1, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), addrspace 8)
|
|
; GFX11: S_WAITCNT 1015
|
|
; GFX1150-NEXT: S_WAITCNT 1015
|
|
; GFX12-NEXT: S_WAIT_LOADCNT 0
|
|
; GCN-NEXT: $vgpr13_vgpr14_vgpr15_vgpr16 = IMAGE_SAMPLE_V4_V2 $vgpr20_vgpr21, $sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11, $sgpr0_sgpr1_sgpr2_sgpr3, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
|
|
renamable $vgpr10_vgpr11_vgpr12_vgpr13 = IMAGE_MSAA_LOAD_V4_V2_gfx11 killed renamable $vgpr4_vgpr5, killed renamable $sgpr0_sgpr1_sgpr2_sgpr3_sgpr4_sgpr5_sgpr6_sgpr7, 4, 7, -1, 0, 0, -1, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), addrspace 8)
|
|
$vgpr13_vgpr14_vgpr15_vgpr16 = IMAGE_SAMPLE_V4_V2 $vgpr20_vgpr21, $sgpr4_sgpr5_sgpr6_sgpr7_sgpr8_sgpr9_sgpr10_sgpr11, $sgpr0_sgpr1_sgpr2_sgpr3, 15, 0, 0, 0, 0, 0, 0, 0, implicit $exec :: (dereferenceable load (s128), align 4, addrspace 4)
|
|
S_ENDPGM 0
|
|
...
|