maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/AMDGPU/hsa.ll

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

87 lines
3.4 KiB
LLVM
Raw Normal View History

AMDGPU: Update AMDHSA code object version handling Differential Revision: https://reviews.llvm.org/D89076
2020-10-15 01:03:37 +08:00
; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=kaveri --amdhsa-code-object-version=2 | FileCheck --check-prefix=HSA %s
; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=kaveri --amdhsa-code-object-version=2 -mattr=-flat-for-global | FileCheck --check-prefix=HSA-CI %s
; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=carrizo --amdhsa-code-object-version=2 | FileCheck --check-prefix=HSA %s
; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=carrizo --amdhsa-code-object-version=2 -mattr=-flat-for-global | FileCheck --check-prefix=HSA-VI %s
; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=kaveri -filetype=obj --amdhsa-code-object-version=2 | llvm-readobj -symbols -s -sd - | FileCheck --check-prefix=ELF %s
; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=kaveri --amdhsa-code-object-version=2 | llvm-mc -filetype=obj -triple amdgcn--amdhsa -mcpu=kaveri --amdhsa-code-object-version=2 | llvm-readobj -symbols -s -sd - | FileCheck %s --check-prefix=ELF
; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=gfx1010 --amdhsa-code-object-version=2 -mattr=+wavefrontsize32,-wavefrontsize64 | FileCheck --check-prefix=HSA --check-prefix=GFX10 --check-prefix=GFX10-W32 %s
; RUN: llc < %s -mtriple=amdgcn--amdhsa -mcpu=gfx1010 --amdhsa-code-object-version=2 -mattr=-wavefrontsize32,+wavefrontsize64 | FileCheck --check-prefix=HSA --check-prefix=GFX10 --check-prefix=GFX10-W64 %s
AMDGPU/SI: Add hsa code object directives Reviewers: arsenm Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D10757 llvm-svn: 240831
2015-06-27 05:15:07 +08:00
; The SHT_NOTE section contains the output from the .hsa_code_object_*
; directives.
AMDGPU/SI: Use .hsatext section instead of .text for HSA Reviewers: arsenm, grosbach, rafael Subscribers: arsenm, llvm-commits Differential Revision: http://reviews.llvm.org/D12424 llvm-svn: 248619
2015-09-26 05:41:28 +08:00
; ELF: Section {
AMDGPU/SI: Add support for AMD code object version 2. Summary: Version 2 is now the default. If you want to emit version 1, use the amdgcn--amdhsa-amdcov1 triple. Reviewers: arsenm, kzhuravl Subscribers: arsenm, llvm-commits Differential Revision: http://reviews.llvm.org/D19283 llvm-svn: 268647
2016-05-06 01:03:33 +08:00
; ELF: Name: .text
AMDGPU/SI: Use .hsatext section instead of .text for HSA Reviewers: arsenm, grosbach, rafael Subscribers: arsenm, llvm-commits Differential Revision: http://reviews.llvm.org/D12424 llvm-svn: 248619
2015-09-26 05:41:28 +08:00
; ELF: Type: SHT_PROGBITS (0x1)
AMDGPU/SI: Add support for AMD code object version 2. Summary: Version 2 is now the default. If you want to emit version 1, use the amdgcn--amdhsa-amdcov1 triple. Reviewers: arsenm, kzhuravl Subscribers: arsenm, llvm-commits Differential Revision: http://reviews.llvm.org/D19283 llvm-svn: 268647
2016-05-06 01:03:33 +08:00
; ELF: Flags [ (0x6)
AMDGPU/SI: Use .hsatext section instead of .text for HSA Reviewers: arsenm, grosbach, rafael Subscribers: arsenm, llvm-commits Differential Revision: http://reviews.llvm.org/D12424 llvm-svn: 248619
2015-09-26 05:41:28 +08:00
; ELF: SHF_ALLOC (0x2)
; ELF: SHF_EXECINSTR (0x4)
; ELF: }
AMDGPU/SI: Add hsa code object directives Reviewers: arsenm Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D10757 llvm-svn: 240831
2015-06-27 05:15:07 +08:00
; ELF: SHT_NOTE
Revert "[AMDGPU] Don’t marke the .note section as ALLOC" This reverts commit 977cd661cf019039dec7ffdd15bf0ac500828c87. It breaks OpenCL testing. OpenCL Runtime is using PT_LOAD information to calculate memory for global variables. This commit should be relanded once the OpenCL runtime stops relying on PT_LOAD information for calculating global variable memory size. Differential Revision: https://reviews.llvm.org/D74995
2020-02-22 07:26:51 +08:00
; ELF: Flags [ (0x2)
; ELF: SHF_ALLOC (0x2)
[AMDGPU] Mark .note section SHF_ALLOC so lld creates a segment for it Differential Revision: https://reviews.llvm.org/D25694 llvm-svn: 284435
2016-10-18 06:40:15 +08:00
; ELF: ]
; ELF: SectionData (
AMDGPU/SI: Add hsa code object directives Reviewers: arsenm Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D10757 llvm-svn: 240831
2015-06-27 05:15:07 +08:00
; ELF: 0000: 04000000 08000000 01000000 414D4400
AMDGPU/SI: Add support for R_AMDGPU_GOTPCREL Reviewers: rafael, ruiu, tony-tye, arsenm, kzhuravl Subscribers: arsenm, llvm-commits, kzhuravl Differential Revision: http://reviews.llvm.org/D21484 llvm-svn: 275268
2016-07-13 22:23:33 +08:00
; ELF: 0010: 02000000 01000000 04000000 1B000000
AMDGPU/SI: Add hsa code object directives Reviewers: arsenm Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D10757 llvm-svn: 240831
2015-06-27 05:15:07 +08:00
; ELF: 0020: 03000000 414D4400 04000700 07000000
; ELF: 0030: 00000000 00000000 414D4400 414D4447
; ELF: 0040: 50550000
[AMDGPU] Mark .note section SHF_ALLOC so lld creates a segment for it Differential Revision: https://reviews.llvm.org/D25694 llvm-svn: 284435
2016-10-18 06:40:15 +08:00
; ELF: )
AMDGPU/SI: Add hsa code object directives Reviewers: arsenm Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D10757 llvm-svn: 240831
2015-06-27 05:15:07 +08:00
AMDGPU/SI: Emit HSA kernels with symbol type STT_AMDGPU_HSA_KERNEL Reviewers: arsenm Subscribers: arsenm, llvm-commits Differential Revision: http://reviews.llvm.org/D13804 llvm-svn: 252291
2015-11-06 19:45:14 +08:00
; ELF: Symbol {
; ELF: Name: simple
AMDGPU: Set flat_scratch from flat_scratch_init reg This was hardcoded to the static private size, but this would be missing the offset and additional size for someday when we have dynamic sizing. Also stops always initializing flat_scratch even when unused. In the future we should stop emitting this unless flat instructions are used to access private memory. For example this will initialize it almost always on VI because flat is used for global access. llvm-svn: 260658
2016-02-12 14:31:30 +08:00
; ELF: Size: 288
AMDGPU/SI: Emit HSA kernels with symbol type STT_AMDGPU_HSA_KERNEL Reviewers: arsenm Subscribers: arsenm, llvm-commits Differential Revision: http://reviews.llvm.org/D13804 llvm-svn: 252291
2015-11-06 19:45:14 +08:00
; ELF: Type: AMDGPU_HSA_KERNEL (0xA)
; ELF: }
[AMDGPU] Do not emit .AMDGPU.config section for amdhsa Differential Revision: https://reviews.llvm.org/D27732 llvm-svn: 291245
2017-01-07 01:02:10 +08:00
; HSA-NOT: .AMDGPU.config
; HSA: .text
AMDGPU/SI: Add support for R_AMDGPU_GOTPCREL Reviewers: rafael, ruiu, tony-tye, arsenm, kzhuravl Subscribers: arsenm, llvm-commits, kzhuravl Differential Revision: http://reviews.llvm.org/D21484 llvm-svn: 275268
2016-07-13 22:23:33 +08:00
; HSA: .hsa_code_object_version 2,1
AMDPGU/SI: Use correct resource descriptors for VI on HSA Summary: We need to set MTYPE = 2 for VI shaders when targeting the HSA runtime. Reviewers: arsenm Differential Revision: http://reviews.llvm.org/D10777 llvm-svn: 240841
2015-06-27 05:58:42 +08:00
; HSA-CI: .hsa_code_object_isa 7,0,0,"AMD","AMDGPU"
; HSA-VI: .hsa_code_object_isa 8,0,1,"AMD","AMDGPU"
R600/SI: Set the ATC bit on all resource descriptors for the HSA runtime llvm-svn: 223125
2014-12-03 01:05:41 +08:00
AMDGPU: Detect kernarg segment pointer This is necessary to pass the kernarg segment pointer to callee functions. Also don't unconditionally enable for kernels. llvm-svn: 307978
2017-07-14 08:11:13 +08:00
; HSA-LABEL: .amdgpu_hsa_kernel simple
AMDGPU/SI: Emit amd_kernel_code_t in EmitFunctionBodyStart() Summary: This way the function symbol points to the start of amd_kernel_code_t rather than the start of the function. Reviewers: arsenm Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D10705 llvm-svn: 240829
2015-06-27 05:14:58 +08:00
; HSA: {{^}}simple:
AMDGPU/SI: Update amd_kernel_code_t definition and add assembler support Reviewers: arsenm Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D10772 llvm-svn: 240839
2015-06-27 05:58:31 +08:00
; HSA: .amd_kernel_code_t
AMDGPU: Rework how private buffer passed for HSA If we know we have stack objects, we reserve the registers that the private buffer resource and wave offset are passed and use them directly. If not, reserve the last 5 SGPRs just in case we need to spill. After register allocation, try to pick the next available registers instead of the last SGPRs, and then insert copies from the inputs to the reserved registers in the progloue. This also only selectively enables all of the input registers which are really required instead of always enabling them. llvm-svn: 254331
2015-12-01 05:16:03 +08:00
; HSA: enable_sgpr_private_segment_buffer = 1
; HSA: enable_sgpr_kernarg_segment_ptr = 1
[AMDGPU] gfx10 tests. NFC. llvm-svn: 363946
2019-06-21 00:29:40 +08:00
; PRE-GFX10: enable_wavefront_size32 = 0
; GFX10-W32: enable_wavefront_size32 = 1
; GFX10-W64: enable_wavefront_size32 = 0
; PRE-GFX10: wavefront_size = 6
; GFX10-W32: wavefront_size = 5
; GFX10-W64: wavefront_size = 6
AMDGPU: Set call_convention bit in kernel_code_t According to the documentation this is supposed to be -1 if indirect calls are not supported. llvm-svn: 293081
2017-01-26 04:21:57 +08:00
; HSA: call_convention = -1
AMDGPU/SI: Update amd_kernel_code_t definition and add assembler support Reviewers: arsenm Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D10772 llvm-svn: 240839
2015-06-27 05:58:31 +08:00
; HSA: .end_amd_kernel_code_t
AMDGPU: Rework how private buffer passed for HSA If we know we have stack objects, we reserve the registers that the private buffer resource and wave offset are passed and use them directly. If not, reserve the last 5 SGPRs just in case we need to spill. After register allocation, try to pick the next available registers instead of the last SGPRs, and then insert copies from the inputs to the reserved registers in the progloue. This also only selectively enables all of the input registers which are really required instead of always enabling them. llvm-svn: 254331
2015-12-01 05:16:03 +08:00
; HSA: s_load_dwordx2 s[{{[0-9]+:[0-9]+}}], s[4:5], 0x0
AMDPGU/SI: Use correct resource descriptors for VI on HSA Summary: We need to set MTYPE = 2 for VI shaders when targeting the HSA runtime. Reviewers: arsenm Differential Revision: http://reviews.llvm.org/D10777 llvm-svn: 240841
2015-06-27 05:58:42 +08:00
R600/SI: Set the ATC bit on all resource descriptors for the HSA runtime llvm-svn: 223125
2014-12-03 01:05:41 +08:00
; Make sure we are setting the ATC bit:
AMDPGU/SI: Use correct resource descriptors for VI on HSA Summary: We need to set MTYPE = 2 for VI shaders when targeting the HSA runtime. Reviewers: arsenm Differential Revision: http://reviews.llvm.org/D10777 llvm-svn: 240841
2015-06-27 05:58:42 +08:00
; HSA-CI: s_mov_b32 s[[HI:[0-9]]], 0x100f000
; On VI+ we also need to set MTYPE = 2
; HSA-VI: s_mov_b32 s[[HI:[0-9]]], 0x1100f000
AMDGPU/SI: Use flat for global load/store when targeting HSA Summary: For some reason doing executing an MUBUF instruction with the addr64 bit set and a zero base pointer in the resource descriptor causes the memory operation to be dropped when the shader is executed using the HSA runtime. This kind of MUBUF instruction is commonly used when the pointer is stored in VGPRs. The base pointer field in the resource descriptor is set to zero and and the pointer is stored in the vaddr field. This patch resolves the issue by only using flat instructions for global memory operations when targeting HSA. This is an overly conservative fix as all other configurations of MUBUF instructions appear to work. NOTE: re-commit by fixing a failure in Codegen/AMDGPU/llvm.dbg.value.ll Reviewers: tstellarAMD Subscribers: arsenm, llvm-commits Differential Revision: http://reviews.llvm.org/D15543 llvm-svn: 256282
2015-12-23 04:55:23 +08:00
; Make sure we generate flat store for HSA
[AMDGPU] gfx10 tests. NFC. llvm-svn: 363946
2019-06-21 00:29:40 +08:00
; PRE-GFX10: flat_store_dword v{{\[[0-9]+:[0-9]+\]}}, v{{[0-9]+}}
AMDGPU: Select global saddr mode from SGPR pointer Use the 64-bit SGPR base with a 0 offset, since it's 1 fewer instruction to materialize the 0 vs. the 64-bit copy.
2020-11-11 00:06:59 +08:00
; GFX10: global_store_dword v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]}}
R600/SI: Set the ATC bit on all resource descriptors for the HSA runtime llvm-svn: 223125
2014-12-03 01:05:41 +08:00
AMDGPU: Emit functions sizes Reviewers: arsenm Subscribers: arsenm, llvm-commits Differential Revision: http://reviews.llvm.org/D15951 llvm-svn: 257172
2016-01-08 22:50:23 +08:00
; HSA: .Lfunc_end0:
; HSA: .size simple, .Lfunc_end0-simple
AMDGPU/SI: Add amdgpu_kernel calling convention. Part 1. Summary: This will be used for AMDGPU_HSA_KERNEL symbol type in output ELF. Also, in the future unused non-kernels may be optimized. For now, also accept SPIR_KERNEL for HCC frontend. Also, add bitcode compatibility tests for missing calling conventions except AVR_BUILTIN which doesn't have parse code. Reviewers: tstellarAMD, arsenm Subscribers: arsenm, joker.eph, llvm-commits llvm-svn: 268717
2016-05-06 17:07:29 +08:00
define amdgpu_kernel void @simple(i32 addrspace(1)* %out) {
R600/SI: Set the ATC bit on all resource descriptors for the HSA runtime llvm-svn: 223125
2014-12-03 01:05:41 +08:00
entry:
store i32 0, i32 addrspace(1)* %out
ret void
}
AMDGPU: Detect kernarg segment pointer This is necessary to pass the kernarg segment pointer to callee functions. Also don't unconditionally enable for kernels. llvm-svn: 307978
2017-07-14 08:11:13 +08:00
; HSA-LABEL: .amdgpu_hsa_kernel simple_no_kernargs
; HSA: enable_sgpr_kernarg_segment_ptr = 0
define amdgpu_kernel void @simple_no_kernargs() {
entry:
store volatile i32 0, i32 addrspace(1)* undef
ret void
}