Commit Graph

  • a6d672166c [Triton-MLIR][OPTIMIZER] Add ExtElemwiseOp to expensive_to_remat list fix-extelemwise-in-combine-ops Qingyi Liu 2022-11-04 15:23:58 +08:00
  • 1f552308c4 finish porting the original logic Superjomn 2022-11-04 13:35:49 +08:00
  • 4218e68d74 [Triton-MLIR] [Frontend] Return a scalar if all input args are scalar (#839) Keren Zhou 2022-11-03 20:27:47 -07:00
  • 61f2ff98df [triton-mlir] add flag "Link only needed" for external libs. (#838) ben-zhang-609 2022-11-03 18:50:20 +08:00
  • 77bc5187b5 Better NVIDIA Pascal GPU Support (#827) Shintaro Iwasaki 2022-11-03 00:11:52 -07:00
  • 91a9773b38 [OPTIMIZER] Minor bugfixes that affected matmul codegen performance (#834) Philippe Tillet 2022-11-02 22:58:09 -07:00
  • 847a318a03 [CI] macos-latest -> macos-10.15 (#836) Philippe Tillet 2022-11-02 22:22:02 -07:00
  • da2993e1c7 init code Superjomn 2022-11-02 18:02:49 +08:00
  • 5feb6e24f9 [Triton-MLIR]Add ptx vprintf support (#825) ben-zhang-609 2022-11-02 16:39:09 +08:00
  • 12d60cb4a3 [BACKEND] Added support for 1D conversion blocked -> slice (#831) Philippe Tillet 2022-11-01 13:19:58 -07:00
  • 9a9fabbba9 Merge pull request #22 from ROCmSoftwarePlatform/IFU_11_1_2022 Michael Melesse 2022-11-01 14:27:33 -04:00
  • 15886b5ffc skip segfault Michael Melesse 2022-11-01 17:52:18 +00:00
  • f16138d447 [Frontend] Interface fixes for libdevice (#830) Chenggang Zhao 2022-11-02 01:51:58 +08:00
  • c9d84237e8 [Triton-MLIR][Frontend] Interface fixes for libdevice (#829) Chenggang Zhao 2022-11-02 01:51:32 +08:00
  • d5830b4b6a Merge branch 'master' into IFU_11_1_2022 Michael Melesse 2022-11-01 17:29:10 +00:00
  • bba1579485 remove scripts Michael Melesse 2022-11-01 17:24:35 +00:00
  • cc6b5180c7 Merge pull request #19 from ROCmSoftwarePlatform/unskip_test_reduce rsanthanam-amd 2022-11-01 11:05:18 -05:00
  • dfad6bdf36 reduce the skips for test_reduce functions Michael Melesse 2022-11-01 15:00:12 +00:00
  • f3bcbcfde6 Merge pull request #18 from ROCmSoftwarePlatform/fix_test_dot rsanthanam-amd 2022-11-01 09:34:37 -05:00
  • 7ec29a7453 revert scripts Michael Melesse 2022-11-01 14:22:33 +00:00
  • 4fb9d4904e fix 6/7 dot tests Michael Melesse 2022-11-01 14:18:06 +00:00
  • cdc0ec5077 [Triton-MLIR][Backend] Fix reduce conversion and unit tests for int dtypes (#826) Qingyi Liu 2022-11-01 17:42:59 +08:00
  • 031c2ae77b [Triton-MLIR][BACKEND] Port the mma<v1> conversion (#815) Yan Chunwei 2022-11-01 09:42:14 +08:00
  • 4f3e2d6ed7 Merge branch 'rocm52_fixes_IFU' into fix_test_dot Michael Melesse 2022-10-31 19:24:45 +00:00
  • fecc7ce248 Fix for test_bitwise subtests for ROCm. (#16) rsanthanam-amd 2022-10-31 14:24:08 -05:00
  • 277b712284 save changes Michael Melesse 2022-10-31 19:11:58 +00:00
  • d024f0cfb8 update test_dot to use float 32 Michael Melesse 2022-10-31 18:58:10 +00:00
  • 1811791665 add failures in report Michael Melesse 2022-10-31 18:39:58 +00:00
  • 9b3f2487b5 fix minor bug Michael Melesse 2022-10-31 18:33:47 +00:00
  • 14730a2352 Merge pull request #15 from ROCmSoftwarePlatform/bfloat_enable rsanthanam-amd 2022-10-31 13:10:30 -05:00
  • 578ada7740 [DOCS] Add install from source instructions to README (#821) Mark Saroufim 2022-10-31 11:08:18 -07:00
  • 15683986cd unskip most bfloat tests Michael Melesse 2022-10-31 18:04:54 +00:00
  • cb1b87a688 [FRONTEND] Made test_if/test_default pass (#823) Philippe Tillet 2022-10-30 15:32:55 -07:00
  • e61dc75942 [FRONTEND] Fixed inliner and got more tests to pass (#822) Philippe Tillet 2022-10-30 14:10:02 -07:00
  • 6311d70406 Revert "[BUILD] Now using cibuildwheel default" Phil Tillet 2022-10-29 17:15:47 -07:00
  • 584086f08c [BUILD] Now using cibuildwheel default Phil Tillet 2022-10-29 16:59:06 -07:00
  • 71428194a1 [BUILD] Add Back Test Target (#820) Ian Bearman 2022-10-29 10:38:50 -07:00
  • 7dfab26a39 [FRONTEND][BACKEND] Fixed various bugs (#819) Philippe Tillet 2022-10-28 23:34:14 -07:00
  • 3ca667dfa8 [Frontend] Return a scalar if all input args are scalar (#816) Keren Zhou 2022-10-28 23:27:06 -07:00
  • 82834d34f9 [BUILD] No longer use include((HandleLLVMOptions) (#818) Philippe Tillet 2022-10-28 17:02:49 -07:00
  • 48fcd8c987 Merge pull request #14 from ROCmSoftwarePlatform/fix_vectorization rsanthanam-amd 2022-10-28 16:12:57 -05:00
  • 8d9572bc63 add similar fixes two addition tests Michael Melesse 2022-10-28 20:34:58 +00:00
  • ffb30cdc52 skip ptx assert Michael Melesse 2022-10-28 20:23:11 +00:00
  • 7fce2bc5f1 add print_llvm_module Michael Melesse 2022-10-28 20:07:35 +00:00
  • f2106d0aa2 [BUILD] Fix Warnings and Enable Warnings as Errors (#794) Ian Bearman 2022-10-28 12:36:09 -07:00
  • 531ef18cb6 Fix for binop % (mod) unit test failures. (#13) rsanthanam-amd 2022-10-28 14:06:17 -05:00
  • 5f0d90db7e tab prints Michael Melesse 2022-10-28 19:05:42 +00:00
  • 03ae41b310 add print helper Michael Melesse 2022-10-28 17:55:28 +00:00
  • bd61338b31 update scripts Michael Melesse 2022-10-28 17:48:26 +00:00
  • 6e50f8b2c0 print irs Michael Melesse 2022-10-28 17:46:52 +00:00
  • ac0f6793cc [BACKEND] Added support for scalars in LoadOp / StoreOp / ElementwiseOp (#814) Philippe Tillet 2022-10-28 01:17:55 -07:00
  • 3685194456 [Triton-MLIR][BACKEND] Add elementwise ops and tests (#804) ben-zhang-609 2022-10-28 13:26:29 +08:00
  • 3b80801dff [Triton-MLIR][Backend] Fix many problems to get the pipeline working (#809) Keren Zhou 2022-10-27 22:09:06 -07:00
  • 42db3538e4 [Triton-MLIR][Backend] Add ReduceOpConversion into TritonGPUToLLVM conversion (#774) Qingyi Liu 2022-10-28 11:07:45 +08:00
  • 3e6cc6d66c [FRONTEND] Made more tests pass (#805) Philippe Tillet 2022-10-26 17:47:33 -07:00
  • aa556d4f1b update script Michael Melesse 2022-10-26 21:51:15 +00:00
  • 61e88efb23 ignore logs Michael Melesse 2022-10-26 21:42:41 +00:00
  • ed9638801a fix for test_cast Michael Melesse 2022-10-26 21:34:58 +00:00
  • 8ecab462f6 skip segfaults on ROCM Michael Melesse 2022-10-26 20:46:47 +00:00
  • bb7008651a [Backend] Hacky fix of missing barrier in ConvertLayout blocked->shared (#803) goostavz 2022-10-27 04:39:38 +08:00
  • 648e4cfe89 skip test_atomic_rmw on rocm Michael Melesse 2022-10-26 18:22:23 +00:00
  • abe0d3e1b1 cast to amd device when as_nvidia shows up Michael Melesse 2022-10-26 18:12:18 +00:00
  • 4464dfcc18 save scripts Michael Melesse 2022-10-26 17:42:58 +00:00
  • 0cae0168ec fix bfloat failure Michael Melesse 2022-10-26 17:40:28 +00:00
  • 88d57ef9c9 add cache print Michael Melesse 2022-10-26 17:19:30 +00:00
  • 39381d99f8 send amdgcn to cache Michael Melesse 2022-10-26 17:18:33 +00:00
  • 4dc2396ca0 [Triton-MLIR][BACKEND] Support $c from mma layout in dot (#798) Yan Chunwei 2022-10-26 10:33:04 +08:00
  • df925f7187 add cache print script Michael Melesse 2022-10-25 20:48:36 +00:00
  • e84297ca79 print cache Michael Melesse 2022-10-25 20:44:42 +00:00
  • 61c85c18b2 try to load binary Michael Melesse 2022-10-25 20:29:43 +00:00
  • da5c24ffcb just clean cache Michael Melesse 2022-10-25 20:27:13 +00:00
  • 09302f0106 fix linking bug Michael Melesse 2022-10-25 18:31:10 +00:00
  • a2cbe7af91 [FRONTEND] Enhanced support for binary operators (#801) Philippe Tillet 2022-10-24 19:47:01 -07:00
  • 5ca1ed0101 Add bf16/fp16/fp64 support for ty_to_cpp (#800) Yanbo Liang 2022-10-24 19:41:25 -07:00
  • fcb228d1d4 Merge select commits from master branch into triton-mlir (#799) Philippe Tillet 2022-10-24 14:52:37 -07:00
  • 9184b5cf65 add prints Michael Melesse 2022-10-24 18:28:28 +00:00
  • 8da4323514 write hipmodule bytes Michael Melesse 2022-10-24 17:58:25 +00:00
  • eb89e9bdd9 fix generator.cc: generator::visit_function: segfault Michael Melesse 2022-10-24 17:41:20 +00:00
  • 877844de4f [Triton-MLIR][BACKEND] add convert_layout[shared->dot_op] converstion to adapt DotOperand layout (#786) Yan Chunwei 2022-10-24 11:40:13 +08:00
  • baab18e1d1 Improve Jokeren 2022-10-23 20:32:25 -07:00
  • 3aa8296b06 [BUILD] Download pybind11 in setup.py (#703) (#797) Philippe Tillet 2022-10-23 18:52:48 -07:00
  • 1bf59d315c [Triton-MLIR][FRONTEND] Remove the dangling check-triton call in setup.py (#796) Yan Chunwei 2022-10-24 09:26:18 +08:00
  • bb0f9235d1 [OPTIMIZER] Made layout simplification pass efficient for fused attention kernels (#790) Philippe Tillet 2022-10-21 16:52:15 -07:00
  • 56a06f7a06 add debug steps Michael Melesse 2022-10-21 20:17:30 +00:00
  • 6a31c43774 update batcktrace Michael Melesse 2022-10-21 19:56:19 +00:00
  • 8785793445 fix typo Michael Melesse 2022-10-21 17:58:38 +00:00
  • d022f5cf2c add compiling back to gcn Michael Melesse 2022-10-21 17:54:31 +00:00
  • c4726333bf [Triton-MLIR] Minor fixes related with scf/swizzling support (#791) goostavz 2022-10-21 11:46:28 +08:00
  • dc0588a898 [OPTIMIZER] Improved layout simplification pass so it handles swizzled layouts better (#789) Philippe Tillet 2022-10-20 19:03:37 -07:00
  • 4624fd4e1d save compiler Michael Melesse 2022-10-19 20:39:32 +00:00
  • 0d22d2bc03 [TritonMLIR] Disallow 0D tensor (#788) Shintaro Iwasaki 2022-10-19 10:34:32 -07:00
  • 4464646efb [Triton-MLIR][BACKEND] Fix masked load store op vector size (#785) Yan Chunwei 2022-10-18 11:43:50 +08:00
  • 41144f927f fix hip launch Michael Melesse 2022-10-17 20:41:28 +00:00
  • 4d6d4c9431 hip src Michael Melesse 2022-10-17 20:18:44 +00:00
  • 32dbc08c05 fix llvm build errors Michael Melesse 2022-10-17 18:29:15 +00:00
  • 4f21501def add fixes Michael Melesse 2022-10-17 18:21:14 +00:00
  • 5c548fb57e Merge branch 'master' into rcom52_fixes Michael Melesse 2022-10-17 17:53:48 +00:00
  • fa4d0fd1ef add scripts Michael Melesse 2022-10-17 17:28:48 +00:00
  • 38a80664b5 [OPTIMIZER] Updated TritonGPU-combine pass (#784) Philippe Tillet 2022-10-16 21:19:42 -07:00
  • e948a618b3 [Triton-MLIR] fix a tiny bug in coalesce pass (#782) goostavz 2022-10-17 11:29:55 +08:00