Code

Sync CentML/hidet -> hidet-org/hidet#476

Code

Merged

vadiklyutiy merged 132 commits intomainhidet-org/hidet:mainfrom vadim/synchidet-org/hidet:vadim/syncCopy head branch name to clipboardDec 20, 2024

+31,183-2,549

Commits

Commits on Dec 19, 2024

• Sync

  vadiklyutiy

  committed
  99de55a

  Copy full SHA for 99de55a

• lint

  vadiklyutiy

  committed
  39287f4

  Copy full SHA for 39287f4

• [Operators] Extend the functionality of einsum to support Ellipsis (#374)

  Show description for f2fc272
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  f2fc272

  Copy full SHA for f2fc272

• [BUG] Fix bug in normalize_launch_dims() (#381)

  Show description for 1431033
  vadiklyutiy

  committed
  1431033

  Copy full SHA for 1431033

• Fix float return when limited by memory (#389)

  Show description for 5e5e5a0
  

  maxyanghu

  authored and

  vadiklyutiy

  committed
  5e5e5a0

  Copy full SHA for 5e5e5a0

• [CI] Adding successfully compiled vision models to the tests/benchmark/run_config.json (#205)

  Show description for 045caed
  

  8 peoplecommitted
  045caed

  Copy full SHA for 045caed

• [Operators] Adding support for the torch.nn.EmbeddingBag (#378)

  Show description for efe3569
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  efe3569

  Copy full SHA for efe3569

• [BUILD] Several changes in wheel building (#392)

  Show description for 9d50fa2
  vadiklyutiy

  committed
  9d50fa2

  Copy full SHA for 9d50fa2

• [DEBUG] Save Task pickle in translations cache (#380)

  Show description for 7746a3a
  vadiklyutiy

  committed
  7746a3a

  Copy full SHA for 7746a3a

• make llama2 work with all ttransformers versions (#385)

  Show description for f7563e6
  

  2 peopleauthored and

  vadiklyutiy

  committed
  f7563e6

  Copy full SHA for f7563e6

• [Bug] Cast dtypes in hidet.where when mismatch (#386)

  Show description for 15ec205
  

  2 peopleauthored and

  vadiklyutiy

  committed
  15ec205

  Copy full SHA for 15ec205

• workaround for gpt-j (#395)

  Show description for 6500c3c
  

  2 peopleauthored and

  vadiklyutiy

  committed
  6500c3c

  Copy full SHA for 6500c3c

• [Bug] Fix hidet.ops.gather, add torch.sign torch.ceil. Disable torch.autograd.function.FunctionCtx (#394)

  Show description for d6dc4c0
  

  2 peopleauthored and

  vadiklyutiy

  committed
  d6dc4c0

  Copy full SHA for d6dc4c0

• Adding accruacy check for huggingface LLMs in Regression (#368)

  Show description for 39186c3
  

  2 peopleauthored and

  vadiklyutiy

  committed
  39186c3

  Copy full SHA for 39186c3

• [Tests] Adding tests for math primitives (#412)

  Show description for 0d2e41a
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  0d2e41a

  Copy full SHA for 0d2e41a

• [CuTe] fix longformer (#411)

  Show description for 838b208
  

  2 peopleauthored and

  vadiklyutiy

  committed
  838b208

  Copy full SHA for 838b208

• [Bug][Enhancement] Correct the behavior of non-parallel build when option parallel_tune is set to 1 (#406)

  Show description for c4bf950
  

  maxyanghu

  authored and

  vadiklyutiy

  committed
  c4bf950

  Copy full SHA for c4bf950

• [Bug] Fixing longformer compilation (#403)

  Show description for 9ac0516
  

  2 peopleauthored and

  vadiklyutiy

  committed
  9ac0516

  Copy full SHA for 9ac0516

• [Fix] Support writing subbyte data to global memory (#415)

  Show description for b1f6177
  

  yaoyaoding

  authored and

  vadiklyutiy

  committed
  b1f6177

  Copy full SHA for b1f6177

• [Update] Updating torch docker image from 24.04 to 24.07 (#418)

  Show description for 850ac79
  

  2 peopleauthored and

  vadiklyutiy

  committed
  850ac79

  Copy full SHA for 850ac79

• [Ir][Primitives] add exp2 (#410)

  Show description for 93a7887
  

  2 peopleauthored and

  vadiklyutiy

  committed
  93a7887

  Copy full SHA for 93a7887

• [Primitives] Add CUDA primitives: prmt, lop3, f16x2 sub and fma, and barrier (#414)

  Show description for d549803
  

  yaoyaoding

  authored and

  vadiklyutiy

  committed
  d549803

  Copy full SHA for d549803

• [Fix] fixed torch.pow (#420)

  Show description for b6e06b7
  

  2 peopleauthored and

  vadiklyutiy

  committed
  b6e06b7

  Copy full SHA for b6e06b7

• [Operators] Adding support for the method torch.Tensor.scatter_add_ (#421)

  Show description for 0b8dbce
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  0b8dbce

  Copy full SHA for 0b8dbce

• [Utility] Add ncu and nsys test utilities (#413)

  Show description for bcce954
  

  yaoyaoding

  authored and

  vadiklyutiy

  committed
  bcce954

  Copy full SHA for bcce954

• [Bug] fixing regression (#422)

  Show description for dd7dab5
  

  2 peopleauthored and

  vadiklyutiy

  committed
  dd7dab5

  Copy full SHA for dd7dab5

• [PERF] Allow prologue fusion for reduce op (#426)

  Show description for f4c2ff9
  vadiklyutiy

  committed
  f4c2ff9

  Copy full SHA for f4c2ff9

• [Version] Update 0.4.0 -> 0.5.0.dev in setup.py (#433)

  Show description for 8faf803
  vadiklyutiy

  committed
  8faf803

  Copy full SHA for 8faf803

• [PERF] Specialize pow(x,2) as x*x. llama-7B (#434)

  Show description for 49780eb
  vadiklyutiy

  committed
  49780eb

  Copy full SHA for 49780eb

• [BUG] Fixing a bug triggered while compiling in-place operator torch.Tensor.scatter_add_ (#429)

  Show description for 081c84d
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  081c84d

  Copy full SHA for 081c84d

• [BUG] Fixing an error triggered from the conv_channel_last_pass while compiling the model sam (#444)

  Show description for c6f0e66
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  c6f0e66

  Copy full SHA for c6f0e66

• [Graph][Ops] disable cublas matmul for parallel k (#431)

  Show description for e807c3d
  

  3 peopleauthored and

  vadiklyutiy

  committed
  e807c3d

  Copy full SHA for e807c3d

• [Scripts] Add scripts of our wheel server (#439)

  Show description for 3420571
  

  yaoyaoding

  authored and

  vadiklyutiy

  committed
  3420571

  Copy full SHA for 3420571

• [Bug] Fixing the ValueError triggered while compiling the model dlrm during operator fusion pass (#437)

  Show description for 123cc00
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  123cc00

  Copy full SHA for 123cc00

• [PERF] Introduce the new IR optimization Pass that spatial(1,47) -> spatial(47) (#452)

  Show description for 9ea795d
  vadiklyutiy

  committed
  9ea795d

  Copy full SHA for 9ea795d

• [BUG] Fixing errors encountered while compiling detectron2_fcos_r_50_fpn (#455)

  Show description for 865500e
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  865500e

  Copy full SHA for 865500e

• [Ir][Primitives] fix #436 via adding missing instructions (#440)

  Show description for c590ffb
  

  2 peopleauthored and

  vadiklyutiy

  committed
  c590ffb

  Copy full SHA for c590ffb

• [BUG] Fixing another error encountered while compiling detectron2_fcos_r_50_fpn (#457)

  Show description for 32d7bf9
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  32d7bf9

  Copy full SHA for 32d7bf9

• [Operators] Adding support for torch.nn.GLU module (#461)

  Show description for d8ecff2
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  d8ecff2

  Copy full SHA for d8ecff2

• [BUG] Fix NotImpelementedError encountered while compiling the model doctr_det_predictor (#462)

  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  a7758db

  Copy full SHA for a7758db

• [CI] Print stderr in run_tests.py (#443)

  Show description for 63eca04
  vadiklyutiy

  committed
  63eca04

  Copy full SHA for 63eca04

• [Dynamic][Enhancement] Convert div and mod including symbolvars to fast int div/mod (#405)

  Show description for 4d3c48f
  

  maxyanghu

  authored and

  vadiklyutiy

  committed
  4d3c48f

  Copy full SHA for 4d3c48f

• Revert "[Dynamic][Enhancement] Convert div and mod including symbolvars to fast int div/mod" (#463)

  Show description for e251184
  

  maxyanghu

  authored and

  vadiklyutiy

  committed
  e251184

  Copy full SHA for e251184

• Added more llms to Regression test (#432)

  Show description for 9308c9f
  

  2 peopleauthored and

  vadiklyutiy

  committed
  9308c9f

  Copy full SHA for 9308c9f

• [PERF] Indexes optimization (#458)

  Show description for cb07596
  vadiklyutiy

  committed
  cb07596

  Copy full SHA for cb07596

• [BUG] Fixing memory issue encountered while compiling the model sam (#466)

  Show description for 3c8c922
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  3c8c922

  Copy full SHA for 3c8c922

• [BUG] Fix ZeroDivisionError triggered wihtin the function parallel_part_heuristic in graph/ops/conv2d/conv2d_gemm.py (#472)

  Show description for 55e29e9
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  55e29e9

  Copy full SHA for 55e29e9

• [BUG] Fix ValueError caused by different operand data types in if_then_else while initializing Conv2dTransposeGemmImageTask (#470)

  Show description for f9fc705
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  f9fc705

  Copy full SHA for f9fc705

• [TOOLS] Attached hash values to function signature in source.cu (#459)

  Show description for c76d7e2
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  c76d7e2

  Copy full SHA for c76d7e2

• [BUG] Support concat empty tensors (#475)

  Show description for b46f65b
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  b46f65b

  Copy full SHA for b46f65b

• remove mpt-7b due to accuracy failure (#477)

  Show description for 2bf6e68
  

  2 peopleauthored and

  vadiklyutiy

  committed
  2bf6e68

  Copy full SHA for 2bf6e68

• Fix masked attention by using fp32 accumulate on first matmul (q and k) part (#468)

  Show description for 0870797
  

  2 peopleauthored and

  vadiklyutiy

  committed
  0870797

  Copy full SHA for 0870797

• Fixed the format change on the new transformers version (#482)

  Show description for dd52728
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  dd52728

  Copy full SHA for dd52728

• [Bug] Resolved multi-threading conflict with save_lower_ir() (#480)

  Show description for f2b1fb8
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  f2b1fb8

  Copy full SHA for f2b1fb8

• [PERF] Сontinue indexes optimisations (#473)

  Show description for eca3695
  vadiklyutiy

  committed
  eca3695

  Copy full SHA for eca3695

• bug fix

  vadiklyutiy

  committed
  e83a072

  Copy full SHA for e83a072

• Revert accidental commit (#484)

  Show description for 1dc3974
  vadiklyutiy

  committed
  1dc3974

  Copy full SHA for 1dc3974

• Empty commit to test out the branch protection rule.

  

  wangshangsam

  authored and

  vadiklyutiy

  committed
  66492e6

  Copy full SHA for 66492e6

• [Dynamic][Enhancement] Convert div and mod including symbolvars to fast int div/mod (#464)

  Show description for e69ec9d
  

  maxyanghu

  authored and

  vadiklyutiy

  committed
  e69ec9d

  Copy full SHA for e69ec9d

• [TOOLS] Task benchmark utilities (#479)

  Show description for 2ce215c
  vadiklyutiy

  committed
  2ce215c

  Copy full SHA for 2ce215c

• [Bug] Rule based simplifier. Fix incorrect rule e/c1/c2 -> e/(c1*c2) (#487)

  Show description for 4f3dccc
  vadiklyutiy

  committed
  4f3dccc

  Copy full SHA for 4f3dccc

• [IR] Bound check for task mapping worker (#483)

  Show description for 36373ff
  vadiklyutiy

  committed
  36373ff

  Copy full SHA for 36373ff

• [Enhancement] Causal attention with fp32 accumulator (#481)

  Show description for 8d7eb10
  

  2 peopleauthored and

  vadiklyutiy

  committed
  8d7eb10

  Copy full SHA for 8d7eb10

• [CI] Update the set of Regression tests (#493)

  Show description for 48d409b
  vadiklyutiy

  committed
  48d409b

  Copy full SHA for 48d409b

• [PERF] Fix for indexes optimization (#488)

  Show description for 5bada76
  vadiklyutiy

  committed
  5bada76

  Copy full SHA for 5bada76

• [Hidet Script] Import externally defined function automatically (#503)

  Show description for 8fdf896
  

  yaoyaoding

  authored and

  vadiklyutiy

  committed
  8fdf896

  Copy full SHA for 8fdf896

• [CI] Make test and publish workflows use built wheel on tests (#492)

  Show description for 5ccd0cc
  

  c-fteixeira

  authored and

  vadiklyutiy

  committed
  5ccd0cc

  Copy full SHA for 5ccd0cc

• [Operators] Support matmul with NT layout (#496)

  Show description for 798f551
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  798f551

  Copy full SHA for 798f551

• [Operators] Support bfloat16 data type in matmul operator (#511)

  Show description for 9e8eeca
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  9e8eeca

  Copy full SHA for 9e8eeca

• [BUG] Fix NT matmul corner case where n or k dimension is odd (#513)

  Show description for 0657eab
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  0657eab

  Copy full SHA for 0657eab

• wgmma instruction support and test for f16 input … (#499)

  Show description for 6764c08
  

  kjiang170

  authored and

  vadiklyutiy

  committed
  6764c08

  Copy full SHA for 6764c08

• [PERF] Rewrite softmax (#516)

  Show description for abe1e53
  vadiklyutiy

  committed
  abe1e53

  Copy full SHA for abe1e53

• [Bug] Fix the incorrect result after merging changes related to matmul_nt (#518)

  Show description for 10e7c3b
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  10e7c3b

  Copy full SHA for 10e7c3b

• refactor wgmma (#521)

  Show description for 80b3bd3
  

  kjiang170

  authored and

  vadiklyutiy

  committed
  80b3bd3

  Copy full SHA for 80b3bd3

• [OP] Support of logsoftmax (#517)

  Show description for be98acf
  vadiklyutiy

  committed
  be98acf

  Copy full SHA for be98acf

• [BUG] Fix distilbert by changing variables names in ops.where (#512)

  Show description for fde166c
  

  2 peopleauthored and

  vadiklyutiy

  committed
  fde166c

  Copy full SHA for fde166c

• [Enhancement] Save running time by using symbolic_run to replace async_run in optimize (#490)

  Show description for daad151
  

  ZichuWu

  and

  vadiklyutiy

  committed
  daad151

  Copy full SHA for daad151

• [Operators] bfloat16 data type support for attention operators (#524)

  Show description for a455026
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  a455026

  Copy full SHA for a455026

• f16 rest options supported and tested (#527)

  Show description for 269a307
  

  kjiang170

  authored and

  vadiklyutiy

  committed
  269a307

  Copy full SHA for 269a307

• [COMPTIME] Added support for run_torch for the rest of transform operation (#525)

  Show description for 3965d85
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  3965d85

  Copy full SHA for 3965d85

• Add more shapes to reduce op in regression (#534)

  Show description for 9450dc5
  

  2 peopleauthored and

  vadiklyutiy

  committed
  9450dc5

  Copy full SHA for 9450dc5

• [Bug] ‘uint32_t’ was not declared in this scope in CI build-wheel for runtime (#545)

  Show description for e824695
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  e824695

  Copy full SHA for e824695

• wgmma bf16 support (#531)

  

  kjiang170

  authored and

  vadiklyutiy

  committed
  75ad7de

  Copy full SHA for 75ad7de

• [OPTIONS] Set mma as default in PassContext() (#530)

  Show description for dfb43fe
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  dfb43fe

  Copy full SHA for dfb43fe

• [Bug] Fix out of memory error occurred while running llama-2-7b (#547)

  Show description for e7fd1b9
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  e7fd1b9

  Copy full SHA for e7fd1b9

• [OPTIONS] Clean Huggingface tokens option (#561)

  Show description for c17766a
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  c17766a

  Copy full SHA for c17766a

• [CI] Tests Workflow. Add manual trigger of tests on different gpu types (#555)

  

  c-fteixeira

  authored and

  vadiklyutiy

  committed
  8eb2501

  Copy full SHA for 8eb2501

• [CI] Shorten build-docs run time (#565)

  Show description for 412dac7
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  412dac7

  Copy full SHA for 412dac7

• [CI] Move import torch inside run_torch() (#570)

  Show description for ceb15fb
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  ceb15fb

  Copy full SHA for ceb15fb

• python3.8 -> python3.9 (#558)

  Show description for 3a06436
  vadiklyutiy

  committed
  3a06436

  Copy full SHA for 3a06436

• [Operators] Allow NT matmul layout for bfloat16 data type (#562)

  Show description for 93c88d2
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  93c88d2

  Copy full SHA for 93c88d2

• fix test_wgmma.py error for illegal warp address (#588)

  Show description for df79733
  

  kjiang170

  authored and

  vadiklyutiy

  committed
  df79733

  Copy full SHA for df79733

• [OPTIONS] Remove unnecessary parallel_k (#572)

  Show description for 9cc1847
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  9cc1847

  Copy full SHA for 9cc1847

• [CI] add new github actions workflow to manually build and push to internal pypi index (#554)

  Show description for ce8bc4f
  

  xinli-centml

  authored and

  vadiklyutiy

  committed
  ce8bc4f

  Copy full SHA for ce8bc4f

• [Tests] Change float16 to bfloat16 for tests/apps (#589)

  Show description for 1a84379
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  1a84379

  Copy full SHA for 1a84379

• [Tests] Adjust test cases for tests/utils for bfloat16. (#597)

  Show description for 2ee866a
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  2ee866a

  Copy full SHA for 2ee866a

• [Tests] Added bfloat16 test cases for tests/cuda (#590)

  Show description for 2df6889
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  2df6889

  Copy full SHA for 2df6889

• [BUG] Fix incorrect converting fxgraph to hidet's flow graph + expand looking for nccl lib with user site packages (#604)

  Show description for 90b3dfe
  vadiklyutiy

  committed
  90b3dfe

  Copy full SHA for 90b3dfe

• [Tests] Adjust test cases for tests/unit-tests for bfloat16. (#596)

  Show description for 38f798b
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  38f798b

  Copy full SHA for 38f798b

• [CI] Exclude tests/unit_tests/test_dynamic_shape.py::test_attention[cuda] (#606)

  Show description for 4db5eb3
  vadiklyutiy

  committed
  4db5eb3

  Copy full SHA for 4db5eb3

• Kaihang/wgmma tf32 u8 i8 support (#549)

  Show description for f20838a
  

  kjiang170

  authored and

  vadiklyutiy

  committed
  f20838a

  Copy full SHA for f20838a

• [Fix] Fixing a minor mistake encountered while adapting test cases for bfloat16 data type (#607)

  Show description for 7cbc208
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  7cbc208

  Copy full SHA for 7cbc208

• Use one global cuda workspace for all the CompiledGraph (#603)

  Show description for 8d6cb44
  

  maxyanghu

  authored and

  vadiklyutiy

  committed
  8d6cb44

  Copy full SHA for 8d6cb44

• [Tests] Adjust test cases for tests/models for bfloat16. (#595)

  Show description for 49014ff
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  49014ff

  Copy full SHA for 49014ff

• [Tests] Adapt tests/ir for bfloat16 test cases (#593)

  Show description for 6921a9b
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  6921a9b

  Copy full SHA for 6921a9b

• [Tests] Adapt tests/frontends to bfloat16 (#592)

  Show description for 4db73ea
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  4db73ea

  Copy full SHA for 4db73ea

• [Tests] Adapt tests/lang for bfloat16 test cases (#594)

  Show description for 3289ed7
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  3289ed7

  Copy full SHA for 3289ed7

• [CI] Turn off search space 2 for tests/lang (#617)

  Show description for 613327e
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  613327e

  Copy full SHA for 613327e

• [BUG] Fix bugs in shared map implementation (#608)

  Show description for ded8809
  vadiklyutiy

  committed
  ded8809

  Copy full SHA for ded8809

• [torchAPI] Inherit cuda stream from torch (#618)

  Show description for c69a298
  vadiklyutiy

  committed
  c69a298

  Copy full SHA for c69a298

• [DISTRIBUTED] Support all_reduce in torch.compile mode (#612)

  Show description for 0fa16fc
  vadiklyutiy

  committed
  0fa16fc

  Copy full SHA for 0fa16fc

• [Tests] Adapt tests/operators for bfloat16 (#615)

  Show description for 48f0dfa
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  48f0dfa

  Copy full SHA for 48f0dfa

• [PERF] Support bf16 in one more place (#623)

  Show description for ec25a0b
  vadiklyutiy

  committed
  ec25a0b

  Copy full SHA for ec25a0b

• [CI]Fix small typoes for building and publishing to internal Hidet PYPI Index (#598)

  Show description for c4e13e5
  

  xinli-centml

  authored and

  vadiklyutiy

  committed
  c4e13e5

  Copy full SHA for c4e13e5

• [BUG] Fix torch2.5 OoM issue (#609)

  Show description for 5c8cfc6
  

  2 peopleauthored and

  vadiklyutiy

  committed
  5c8cfc6

  Copy full SHA for 5c8cfc6

• Revert "[BUG] Fix torch2.5 OoM issue" (#635)

  Show description for 097ce2f
  

  zhumakhan

  authored and

  vadiklyutiy

  committed
  097ce2f

  Copy full SHA for 097ce2f

• [BUG] Fix torch2.5 OoM and docs build fix (#637)

  Show description for 9c7ca36
  

  2 peopleauthored and

  vadiklyutiy

  committed
  9c7ca36

  Copy full SHA for 9c7ca36

• [COMPTIME] Hot start speedup (#625)

  Show description for 7b21b7b
  vadiklyutiy

  committed
  7b21b7b

  Copy full SHA for 7b21b7b

• [Bug] Parallel compilation sync (#616)

  Show description for a5654c9
  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  a5654c9

  Copy full SHA for a5654c9

• Adapt to bfloat16 where necessary (#624)

  

  ZichuWu

  authored and

  vadiklyutiy

  committed
  be933f3

  Copy full SHA for be933f3

• [PERF] Default value for parallel_k is 'disabled' (#634)

  Show description for bb7396f
  vadiklyutiy

  committed
  bb7396f

  Copy full SHA for bb7396f

• Hexcute base branch (All related PRs will be merged into this base PR. ) (#294)

  Show description for 62057b8
  

  2 peopleauthored and

  vadiklyutiy

  committed
  62057b8

  Copy full SHA for 62057b8

• [BUG] fix attach hash to signature (#638)

  Show description for 4fcea21
  

  2 peopleauthored and

  vadiklyutiy

  committed
  4fcea21

  Copy full SHA for 4fcea21

• [IR] Add support for swizzle, interleave and l2Promotion in tensor map creation (#643)

  Show description for 430709e
  

  BolinSNLHM

  authored and

  vadiklyutiy

  committed
  430709e

  Copy full SHA for 430709e

• [BUG] VLLM (and DMWL) compile with hidet backend (#647)

  Show description for 21a5905
  

  2 peopleauthored and

  vadiklyutiy

  committed
  21a5905

  Copy full SHA for 21a5905

• matmul_f16 with wgmma (#627)

  Show description for d2a94e0
  

  kjiang170

  authored and

  vadiklyutiy

  committed
  d2a94e0

  Copy full SHA for d2a94e0

• [BUG] A number of fixes for vllm's TP (#651)

  Show description for 5591f07
  vadiklyutiy

  committed
  5591f07

  Copy full SHA for 5591f07

• [BUG] Add comp server requirements (#661)

  Show description for 53ae60f
  vadiklyutiy

  committed
  53ae60f

  Copy full SHA for 53ae60f

• resolve conflicts

  Show description for 4180123
  vadiklyutiy

  committed
  4180123

  Copy full SHA for 4180123

• [Dependency] Remove the version restriction of transformers and diffuers (#475)

  Show description for 06955a9
  

  yaoyaoding

  authored and

  vadiklyutiy

  committed
  06955a9

  Copy full SHA for 06955a9

• lint

  vadiklyutiy

  committed
  e38da75

  Copy full SHA for e38da75

• Temporary switch test workflow to old runners because new one are not enabled on hidet-org/hidet

  vadiklyutiy

  committed
  d2dd415

  Copy full SHA for d2dd415