rsanthanam-amd
46fd581b0a
Merge pull request #29 from ROCmSoftwarePlatform/parse_amdgcn_from_rocminfo
...
Changes to eliminate the need for the MI_GPU_ARCH environment variable.
2022-11-18 12:53:25 -06:00
Rohit Santhanam
8cc448d92e
Changes to eliminate the need for the MI_GPU_ARCH environment variable.
...
The AMDGPU arch is now parsed out of the rocminfo dump.
2022-11-18 18:51:57 +00:00
Michael Melesse
9a9fabbba9
Merge pull request #22 from ROCmSoftwarePlatform/IFU_11_1_2022
...
IFU 11/1/2022
2022-11-01 14:27:33 -04:00
Michael Melesse
15886b5ffc
skip segfault
2022-11-01 17:52:18 +00:00
Michael Melesse
d5830b4b6a
Merge branch 'master' into IFU_11_1_2022
2022-11-01 17:29:10 +00:00
Michael Melesse
bba1579485
remove scripts
2022-11-01 17:24:35 +00:00
rsanthanam-amd
cc6b5180c7
Merge pull request #19 from ROCmSoftwarePlatform/unskip_test_reduce
...
reduce the skips for test_reduce functions
2022-11-01 11:05:18 -05:00
Michael Melesse
dfad6bdf36
reduce the skips for test_reduce functions
2022-11-01 15:00:12 +00:00
rsanthanam-amd
f3bcbcfde6
Merge pull request #18 from ROCmSoftwarePlatform/fix_test_dot
...
Fix 6/7 test dot
2022-11-01 09:34:37 -05:00
Michael Melesse
7ec29a7453
revert scripts
2022-11-01 14:22:33 +00:00
Michael Melesse
4fb9d4904e
fix 6/7 dot tests
2022-11-01 14:18:06 +00:00
Michael Melesse
4f3e2d6ed7
Merge branch 'rocm52_fixes_IFU' into fix_test_dot
2022-10-31 19:24:45 +00:00
rsanthanam-amd
fecc7ce248
Fix for test_bitwise subtests for ROCm. ( #16 )
...
The issue was that the kernel names were colliding with each other in
the cache. Since the kernel names were based on the date and time, the
kernels were getting compiled so fast that a subsequent kernel would end
up with the same name as the previous one and would therefore overwrite
it in the cache.
It seems to run the same test multiple times but the subsequent runs
would end up using the wrong kernel because of the collisions.
It is fixed by appending a randomly generated alphanumeric string to
keep the kernel names unique.
2022-10-31 15:24:08 -04:00
Michael Melesse
277b712284
save changes
2022-10-31 19:11:58 +00:00
Michael Melesse
d024f0cfb8
update test_dot to use float 32
2022-10-31 18:58:10 +00:00
Michael Melesse
1811791665
add failures in report
2022-10-31 18:39:58 +00:00
Michael Melesse
9b3f2487b5
fix minor bug
2022-10-31 18:33:47 +00:00
rsanthanam-amd
14730a2352
Merge pull request #15 from ROCmSoftwarePlatform/bfloat_enable
...
unskip most bfloat tests
2022-10-31 13:10:30 -05:00
Mark Saroufim
578ada7740
[DOCS] Add install from source instructions to README ( #821 )
2022-10-31 11:08:18 -07:00
Michael Melesse
15683986cd
unskip most bfloat tests
2022-10-31 18:04:54 +00:00
Phil Tillet
6311d70406
Revert "[BUILD] Now using cibuildwheel default"
...
This reverts commit 584086f08c
2022-10-29 17:15:47 -07:00
Phil Tillet
584086f08c
[BUILD] Now using cibuildwheel default
2022-10-29 16:59:06 -07:00
Keren Zhou
3ca667dfa8
[Frontend] Return a scalar if all input args are scalar ( #816 )
2022-10-28 23:27:06 -07:00
rsanthanam-amd
48fcd8c987
Merge pull request #14 from ROCmSoftwarePlatform/fix_vectorization
...
fix test_vectorization and test_load_cache_modifier
2022-10-28 16:12:57 -05:00
Michael Melesse
8d9572bc63
add similar fixes two addition tests
2022-10-28 20:34:58 +00:00
Michael Melesse
ffb30cdc52
skip ptx assert
2022-10-28 20:23:11 +00:00
Michael Melesse
7fce2bc5f1
add print_llvm_module
2022-10-28 20:07:35 +00:00
rsanthanam-amd
531ef18cb6
Fix for binop % (mod) unit test failures. ( #13 )
...
If the either data type if fp, then fmod should be used for the
reference computation.
2022-10-28 15:06:17 -04:00
Michael Melesse
5f0d90db7e
tab prints
2022-10-28 19:05:42 +00:00
Michael Melesse
03ae41b310
add print helper
2022-10-28 17:55:28 +00:00
Michael Melesse
bd61338b31
update scripts
2022-10-28 17:48:26 +00:00
Michael Melesse
6e50f8b2c0
print irs
2022-10-28 17:46:52 +00:00
Michael Melesse
aa556d4f1b
update script
2022-10-26 21:51:15 +00:00
Michael Melesse
61e88efb23
ignore logs
2022-10-26 21:42:41 +00:00
Michael Melesse
ed9638801a
fix for test_cast
2022-10-26 21:34:58 +00:00
Michael Melesse
8ecab462f6
skip segfaults on ROCM
2022-10-26 20:46:47 +00:00
Michael Melesse
648e4cfe89
skip test_atomic_rmw on rocm
2022-10-26 18:22:23 +00:00
Michael Melesse
abe0d3e1b1
cast to amd device when as_nvidia shows up
2022-10-26 18:12:18 +00:00
Michael Melesse
4464dfcc18
save scripts
2022-10-26 17:42:58 +00:00
Michael Melesse
0cae0168ec
fix bfloat failure
2022-10-26 17:40:28 +00:00
Michael Melesse
88d57ef9c9
add cache print
2022-10-26 17:19:30 +00:00
Michael Melesse
39381d99f8
send amdgcn to cache
2022-10-26 17:18:33 +00:00
Michael Melesse
df925f7187
add cache print script
2022-10-25 20:48:36 +00:00
Michael Melesse
e84297ca79
print cache
2022-10-25 20:44:42 +00:00
Michael Melesse
61c85c18b2
try to load binary
2022-10-25 20:29:43 +00:00
Michael Melesse
da5c24ffcb
just clean cache
2022-10-25 20:27:13 +00:00
Michael Melesse
09302f0106
fix linking bug
2022-10-25 18:31:10 +00:00
Yanbo Liang
5ca1ed0101
Add bf16/fp16/fp64 support for ty_to_cpp ( #800 )
...
In ```torch._inductor```, we [convert 0d CPU tensor to scalar during
triton codegen](https://github.com/pytorch/pytorch/pull/87329 ), so need
add missing triton support for bf16/fp16/fp64.
2022-10-24 19:41:25 -07:00
Michael Melesse
9184b5cf65
add prints
2022-10-24 18:28:28 +00:00
Michael Melesse
8da4323514
write hipmodule bytes
2022-10-24 17:58:25 +00:00
