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/udivrem64.ll

170 lines
3.6 KiB
LLVM
Raw Normal View History

AMDGPU: Fix high bits after division optimization This is essentially doing a 24-bit signed division with FP. We need to truncate to the N bit result. llvm-svn: 270305
2016-05-21 09:53:33 +08:00
;RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck --check-prefix=SI --check-prefix=GCN --check-prefix=FUNC %s
Enable FeatureFlatForGlobal on Volcanic Islands This switches to the workaround that HSA defaults to for the mesa path. This should be applied to the 4.0 branch. Patch by Vedran Miletić <vedran@miletic.net> llvm-svn: 292982
2017-01-25 06:02:15 +08:00
;RUN: llc -march=amdgcn -mcpu=tonga -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck --check-prefix=VI --check-prefix=GCN --check-prefix=FUNC %s
R600: Try to use lower types for 64bit division if possible v2: add and enable tests for SI Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Matt Arsenault <Matthew.Arsenault@amd.com> llvm-svn: 226881
2015-01-23 07:42:43 +08:00
;RUN: llc -march=r600 -mcpu=redwood < %s | FileCheck --check-prefix=EG --check-prefix=FUNC %s
R600: optimize the UDIVREM 64 algorithm This is a squash of several optimization commits: - calculate DIV_Lo and DIV_Hi separately - use BFE_U32 if we are operating on 32bit values - use precomputed constants instead of shifting in UDVIREM - skip the first 32 iterations of udivrem v2: Check whether BFE is supported before using it Patch by: Jan Vesely Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Tom Stellard <thomas.stellard@amd.com> llvm-svn: 207589
2014-04-30 07:12:46 +08:00
R600: Call EmitFunctionHeader() in the AsmPrinter to populate the ELF symbol table llvm-svn: 218776
2014-10-02 01:15:17 +08:00
;FUNC-LABEL: {{^}}test_udiv:
R600: optimize the UDIVREM 64 algorithm This is a squash of several optimization commits: - calculate DIV_Lo and DIV_Hi separately - use BFE_U32 if we are operating on 32bit values - use precomputed constants instead of shifting in UDVIREM - skip the first 32 iterations of udivrem v2: Check whether BFE is supported before using it Patch by: Jan Vesely Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Tom Stellard <thomas.stellard@amd.com> llvm-svn: 207589
2014-04-30 07:12:46 +08:00
;EG: RECIP_UINT
R600: Try to convert BFE back to standard bit ops when possible. This allows existing DAG combines to work on them, and then we can re-match to BFE if necessary during instruction selection. llvm-svn: 209462
2014-05-23 02:09:12 +08:00
;EG: LSHL {{.*}}, 1,
R600: optimize the UDIVREM 64 algorithm This is a squash of several optimization commits: - calculate DIV_Lo and DIV_Hi separately - use BFE_U32 if we are operating on 32bit values - use precomputed constants instead of shifting in UDVIREM - skip the first 32 iterations of udivrem v2: Check whether BFE is supported before using it Patch by: Jan Vesely Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Tom Stellard <thomas.stellard@amd.com> llvm-svn: 207589
2014-04-30 07:12:46 +08:00
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
R600: Try to use lower types for 64bit division if possible v2: add and enable tests for SI Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Matt Arsenault <Matthew.Arsenault@amd.com> llvm-svn: 226881
2015-01-23 07:42:43 +08:00
[AMDGPU] New 64 bit div/rem expansion Old expansion was 20 VGPRs, 78 SGPRs and ~380 instructions. This expansion is 11 VGPRs, 12 SGPRs and ~120 instructions. Passes OpenCL conformance test_integer_ops quick_[u]long_math Differential Revision: https://reviews.llvm.org/D38607 llvm-svn: 315081
2017-10-07 01:24:45 +08:00
;GCN: v_mac_f32_e32 v{{[0-9]+}}, 0x4f800000,
;GCN: v_rcp_f32_e32
;GCN: v_mul_f32_e32 v{{[0-9]+}}, 0x5f7ffffc
;GCN: v_mul_f32_e32 v{{[0-9]+}}, 0x2f800000
;GCN: v_trunc_f32_e32
;GCN: v_mac_f32_e32 v{{[0-9]+}}, 0xcf800000
R600/SI: Enable a lot of existing tests for VI (squashed commits) This is a union of these commits: * R600/SI: Enable more tests for VI which need no changes * R600/SI: Enable V_BCNT tests for VI Differences: - v_bcnt_..._e32 -> _e64 - s_load_dword* inline offset is in bytes instead of dwords * R600/SI: Enable all tests for VI which use S_LOAD_DWORD The inline offset is changed from dwords to bytes. * R600/SI: Enable LDS tests for VI Differences: - the s_load_dword inline offset changed from dwords to bytes - the tests checked very little on CI, so they have been fixed to check all instructions that "SI" checked * R600/SI: Enable lshr tests for VI * R600/SI: Fix divrem64 tests - "v_lshl_64" was missing "b" before "64" - added VI-NOT checks * R600/SI: Enable the SI.tid test for VI * R600/SI: Enable the frem test for VI Also, the frem_f64 checking is added for CI-VI. * R600/SI: Add VI tests for rsq.clamped llvm-svn: 228830
2015-02-11 22:26:46 +08:00
;GCN: s_endpgm
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_udiv(i64 addrspace(1)* %out, i64 %x, i64 %y) {
R600: optimize the UDIVREM 64 algorithm This is a squash of several optimization commits: - calculate DIV_Lo and DIV_Hi separately - use BFE_U32 if we are operating on 32bit values - use precomputed constants instead of shifting in UDVIREM - skip the first 32 iterations of udivrem v2: Check whether BFE is supported before using it Patch by: Jan Vesely Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Tom Stellard <thomas.stellard@amd.com> llvm-svn: 207589
2014-04-30 07:12:46 +08:00
%result = udiv i64 %x, %y
store i64 %result, i64 addrspace(1)* %out
ret void
}
R600: Call EmitFunctionHeader() in the AsmPrinter to populate the ELF symbol table llvm-svn: 218776
2014-10-02 01:15:17 +08:00
;FUNC-LABEL: {{^}}test_urem:
R600: optimize the UDIVREM 64 algorithm This is a squash of several optimization commits: - calculate DIV_Lo and DIV_Hi separately - use BFE_U32 if we are operating on 32bit values - use precomputed constants instead of shifting in UDVIREM - skip the first 32 iterations of udivrem v2: Check whether BFE is supported before using it Patch by: Jan Vesely Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Tom Stellard <thomas.stellard@amd.com> llvm-svn: 207589
2014-04-30 07:12:46 +08:00
;EG: RECIP_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
;EG: BFE_UINT
R600: Try to convert BFE back to standard bit ops when possible. This allows existing DAG combines to work on them, and then we can re-match to BFE if necessary during instruction selection. llvm-svn: 209462
2014-05-23 02:09:12 +08:00
;EG: AND_INT {{.*}}, 1,
R600: Try to use lower types for 64bit division if possible v2: add and enable tests for SI Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Matt Arsenault <Matthew.Arsenault@amd.com> llvm-svn: 226881
2015-01-23 07:42:43 +08:00
[AMDGPU] New 64 bit div/rem expansion Old expansion was 20 VGPRs, 78 SGPRs and ~380 instructions. This expansion is 11 VGPRs, 12 SGPRs and ~120 instructions. Passes OpenCL conformance test_integer_ops quick_[u]long_math Differential Revision: https://reviews.llvm.org/D38607 llvm-svn: 315081
2017-10-07 01:24:45 +08:00
;GCN: v_mac_f32_e32 v{{[0-9]+}}, 0x4f800000,
;GCN: v_rcp_f32_e32
;GCN: v_mul_f32_e32 v{{[0-9]+}}, 0x5f7ffffc
;GCN: v_mul_f32_e32 v{{[0-9]+}}, 0x2f800000
;GCN: v_trunc_f32_e32
;GCN: v_mac_f32_e32 v{{[0-9]+}}, 0xcf800000
R600/SI: Enable a lot of existing tests for VI (squashed commits) This is a union of these commits: * R600/SI: Enable more tests for VI which need no changes * R600/SI: Enable V_BCNT tests for VI Differences: - v_bcnt_..._e32 -> _e64 - s_load_dword* inline offset is in bytes instead of dwords * R600/SI: Enable all tests for VI which use S_LOAD_DWORD The inline offset is changed from dwords to bytes. * R600/SI: Enable LDS tests for VI Differences: - the s_load_dword inline offset changed from dwords to bytes - the tests checked very little on CI, so they have been fixed to check all instructions that "SI" checked * R600/SI: Enable lshr tests for VI * R600/SI: Fix divrem64 tests - "v_lshl_64" was missing "b" before "64" - added VI-NOT checks * R600/SI: Enable the SI.tid test for VI * R600/SI: Enable the frem test for VI Also, the frem_f64 checking is added for CI-VI. * R600/SI: Add VI tests for rsq.clamped llvm-svn: 228830
2015-02-11 22:26:46 +08:00
;GCN: s_endpgm
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_urem(i64 addrspace(1)* %out, i64 %x, i64 %y) {
R600: optimize the UDIVREM 64 algorithm This is a squash of several optimization commits: - calculate DIV_Lo and DIV_Hi separately - use BFE_U32 if we are operating on 32bit values - use precomputed constants instead of shifting in UDVIREM - skip the first 32 iterations of udivrem v2: Check whether BFE is supported before using it Patch by: Jan Vesely Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Tom Stellard <thomas.stellard@amd.com> llvm-svn: 207589
2014-04-30 07:12:46 +08:00
%result = urem i64 %x, %y
store i64 %result, i64 addrspace(1)* %out
ret void
}
R600: Try to use lower types for 64bit division if possible v2: add and enable tests for SI Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Matt Arsenault <Matthew.Arsenault@amd.com> llvm-svn: 226881
2015-01-23 07:42:43 +08:00
;FUNC-LABEL: {{^}}test_udiv3264:
;EG: RECIP_UINT
;EG-NOT: BFE_UINT
R600/SI: Enable a lot of existing tests for VI (squashed commits) This is a union of these commits: * R600/SI: Enable more tests for VI which need no changes * R600/SI: Enable V_BCNT tests for VI Differences: - v_bcnt_..._e32 -> _e64 - s_load_dword* inline offset is in bytes instead of dwords * R600/SI: Enable all tests for VI which use S_LOAD_DWORD The inline offset is changed from dwords to bytes. * R600/SI: Enable LDS tests for VI Differences: - the s_load_dword inline offset changed from dwords to bytes - the tests checked very little on CI, so they have been fixed to check all instructions that "SI" checked * R600/SI: Enable lshr tests for VI * R600/SI: Fix divrem64 tests - "v_lshl_64" was missing "b" before "64" - added VI-NOT checks * R600/SI: Enable the SI.tid test for VI * R600/SI: Enable the frem test for VI Also, the frem_f64 checking is added for CI-VI. * R600/SI: Add VI tests for rsq.clamped llvm-svn: 228830
2015-02-11 22:26:46 +08:00
;GCN-NOT: s_bfe_u32
;GCN-NOT: v_mad_f32
;SI-NOT: v_lshr_b64
;VI-NOT: v_lshrrev_b64
;GCN: s_endpgm
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_udiv3264(i64 addrspace(1)* %out, i64 %x, i64 %y) {
R600: Try to use lower types for 64bit division if possible v2: add and enable tests for SI Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Matt Arsenault <Matthew.Arsenault@amd.com> llvm-svn: 226881
2015-01-23 07:42:43 +08:00
%1 = lshr i64 %x, 33
%2 = lshr i64 %y, 33
%result = udiv i64 %1, %2
store i64 %result, i64 addrspace(1)* %out
ret void
}
;FUNC-LABEL: {{^}}test_urem3264:
;EG: RECIP_UINT
;EG-NOT: BFE_UINT
R600/SI: Enable a lot of existing tests for VI (squashed commits) This is a union of these commits: * R600/SI: Enable more tests for VI which need no changes * R600/SI: Enable V_BCNT tests for VI Differences: - v_bcnt_..._e32 -> _e64 - s_load_dword* inline offset is in bytes instead of dwords * R600/SI: Enable all tests for VI which use S_LOAD_DWORD The inline offset is changed from dwords to bytes. * R600/SI: Enable LDS tests for VI Differences: - the s_load_dword inline offset changed from dwords to bytes - the tests checked very little on CI, so they have been fixed to check all instructions that "SI" checked * R600/SI: Enable lshr tests for VI * R600/SI: Fix divrem64 tests - "v_lshl_64" was missing "b" before "64" - added VI-NOT checks * R600/SI: Enable the SI.tid test for VI * R600/SI: Enable the frem test for VI Also, the frem_f64 checking is added for CI-VI. * R600/SI: Add VI tests for rsq.clamped llvm-svn: 228830
2015-02-11 22:26:46 +08:00
;GCN-NOT: s_bfe_u32
;GCN-NOT: v_mad_f32
;SI-NOT: v_lshr_b64
;VI-NOT: v_lshrrev_b64
;GCN: s_endpgm
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_urem3264(i64 addrspace(1)* %out, i64 %x, i64 %y) {
R600: Try to use lower types for 64bit division if possible v2: add and enable tests for SI Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Matt Arsenault <Matthew.Arsenault@amd.com> llvm-svn: 226881
2015-01-23 07:42:43 +08:00
%1 = lshr i64 %x, 33
%2 = lshr i64 %y, 33
%result = urem i64 %1, %2
store i64 %result, i64 addrspace(1)* %out
ret void
}
AMDGPU: Fix high bits after division optimization This is essentially doing a 24-bit signed division with FP. We need to truncate to the N bit result. llvm-svn: 270305
2016-05-21 09:53:33 +08:00
;FUNC-LABEL: {{^}}test_udiv2364:
R600: Try to use lower types for 64bit division if possible v2: add and enable tests for SI Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Matt Arsenault <Matthew.Arsenault@amd.com> llvm-svn: 226881
2015-01-23 07:42:43 +08:00
;EG: UINT_TO_FLT
;EG: UINT_TO_FLT
;EG: FLT_TO_UINT
;EG-NOT: RECIP_UINT
;EG-NOT: BFE_UINT
R600/SI: Enable a lot of existing tests for VI (squashed commits) This is a union of these commits: * R600/SI: Enable more tests for VI which need no changes * R600/SI: Enable V_BCNT tests for VI Differences: - v_bcnt_..._e32 -> _e64 - s_load_dword* inline offset is in bytes instead of dwords * R600/SI: Enable all tests for VI which use S_LOAD_DWORD The inline offset is changed from dwords to bytes. * R600/SI: Enable LDS tests for VI Differences: - the s_load_dword inline offset changed from dwords to bytes - the tests checked very little on CI, so they have been fixed to check all instructions that "SI" checked * R600/SI: Enable lshr tests for VI * R600/SI: Fix divrem64 tests - "v_lshl_64" was missing "b" before "64" - added VI-NOT checks * R600/SI: Enable the SI.tid test for VI * R600/SI: Enable the frem test for VI Also, the frem_f64 checking is added for CI-VI. * R600/SI: Add VI tests for rsq.clamped llvm-svn: 228830
2015-02-11 22:26:46 +08:00
;SI-NOT: v_lshr_b64
;VI-NOT: v_lshrrev_b64
;GCN: v_mad_f32
;GCN: s_endpgm
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_udiv2364(i64 addrspace(1)* %out, i64 %x, i64 %y) {
AMDGPU: Fix high bits after division optimization This is essentially doing a 24-bit signed division with FP. We need to truncate to the N bit result. llvm-svn: 270305
2016-05-21 09:53:33 +08:00
%1 = lshr i64 %x, 41
%2 = lshr i64 %y, 41
R600: Try to use lower types for 64bit division if possible v2: add and enable tests for SI Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Matt Arsenault <Matthew.Arsenault@amd.com> llvm-svn: 226881
2015-01-23 07:42:43 +08:00
%result = udiv i64 %1, %2
store i64 %result, i64 addrspace(1)* %out
ret void
}
AMDGPU: Fix high bits after division optimization This is essentially doing a 24-bit signed division with FP. We need to truncate to the N bit result. llvm-svn: 270305
2016-05-21 09:53:33 +08:00
;FUNC-LABEL: {{^}}test_urem2364:
R600: Try to use lower types for 64bit division if possible v2: add and enable tests for SI Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Matt Arsenault <Matthew.Arsenault@amd.com> llvm-svn: 226881
2015-01-23 07:42:43 +08:00
;EG: UINT_TO_FLT
;EG: UINT_TO_FLT
;EG: FLT_TO_UINT
;EG-NOT: RECIP_UINT
;EG-NOT: BFE_UINT
R600/SI: Enable a lot of existing tests for VI (squashed commits) This is a union of these commits: * R600/SI: Enable more tests for VI which need no changes * R600/SI: Enable V_BCNT tests for VI Differences: - v_bcnt_..._e32 -> _e64 - s_load_dword* inline offset is in bytes instead of dwords * R600/SI: Enable all tests for VI which use S_LOAD_DWORD The inline offset is changed from dwords to bytes. * R600/SI: Enable LDS tests for VI Differences: - the s_load_dword inline offset changed from dwords to bytes - the tests checked very little on CI, so they have been fixed to check all instructions that "SI" checked * R600/SI: Enable lshr tests for VI * R600/SI: Fix divrem64 tests - "v_lshl_64" was missing "b" before "64" - added VI-NOT checks * R600/SI: Enable the SI.tid test for VI * R600/SI: Enable the frem test for VI Also, the frem_f64 checking is added for CI-VI. * R600/SI: Add VI tests for rsq.clamped llvm-svn: 228830
2015-02-11 22:26:46 +08:00
;SI-NOT: v_lshr_b64
;VI-NOT: v_lshrrev_b64
;GCN: v_mad_f32
;GCN: s_endpgm
AMDGPU: Mark all unspecified CC functions in tests as amdgpu_kernel Currently the default C calling convention functions are treated the same as compute kernels. Make this explicit so the default calling convention can be changed to a non-kernel. Converted with perl -pi -e 's/define void/define amdgpu_kernel void/' on the relevant test directories (and undoing in one place that actually wanted a non-kernel). llvm-svn: 298444
2017-03-22 05:39:51 +08:00
define amdgpu_kernel void @test_urem2364(i64 addrspace(1)* %out, i64 %x, i64 %y) {
AMDGPU: Fix high bits after division optimization This is essentially doing a 24-bit signed division with FP. We need to truncate to the N bit result. llvm-svn: 270305
2016-05-21 09:53:33 +08:00
%1 = lshr i64 %x, 41
%2 = lshr i64 %y, 41
R600: Try to use lower types for 64bit division if possible v2: add and enable tests for SI Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu> Reviewed-by: Matt Arsenault <Matthew.Arsenault@amd.com> llvm-svn: 226881
2015-01-23 07:42:43 +08:00
%result = urem i64 %1, %2
store i64 %result, i64 addrspace(1)* %out
ret void
}