ELS-RD/kernl

Issues

• bug: Fails running dynamic shapes

  #338 opened a year ago by michaelfeil
  0

• Confuse about block_delta

  #337 opened a year ago by zhanglei1172
  0

• bug: optimize_model() fails on HF's GPT2 with "RuntimeError: CUDA error: operation not permitted when stream is capturing"

  #336 opened a year ago by CorentinJ
  4

• bug: Llama model optimization failing

  #317 opened 2 years ago by AndrewMead10
  4

• bug: Experimental/Whisper notebook (speedup.ipynb) is not working

  #331 opened a year ago by Artyom17
  0

• bug: Torch.dynamo is not working on H100 due to obsolete triton & pytorch

  #330 opened a year ago by Artyom17
  0

• bug: start_position support for the fused attention kernel

  #329 opened a year ago by ipoletaev
  0

• bug: Bart speedup only 1.6x

  #327 opened 2 years ago by sinking-point
  0

• bug: Llama reproduce error with kernl

  #321 opened 2 years ago by yychen016
  1

• bug: How to save the optimized model to file?

  #313 opened 2 years ago by aaronchan90
  3

• bug: does kernl support pipeline parallel?

  #323 opened 2 years ago by ninisy
  0

• bug: Could not get kernl running on CodeT5

  #283 opened 2 years ago by TheSeamau5
  6

• proposal: Write GEMM (matrice mulitplication) triton optimization animation

  #318 opened 2 years ago by pommedeterresautee
  0

• Review the doc so that it displays correctly on the site ?

  #290 opened 2 years ago by white-gorilla
  0

• bug: AttributeError: module 'torch._utils' has no attribute 'is_compiling'

  #266 opened 2 years ago by gilljon
  2

• docs: automatic code reference generation

  #280 opened 2 years ago by jonathlela
  1

• feature: using TorchBench to test the coverage

  #305 opened 2 years ago by xuzhao9
  2

• bug: Triton 2.0 makes attention kernel crash

  #314 opened 2 years ago by pommedeterresautee
  1

• bug: Trying to run T5 tutorial and getting `free(): invalid pointer` error.

  #310 opened 2 years ago by gilljon
  1

• [FRONT] Linking kernl and the blog

  #281 opened 2 years ago by white-gorilla
  0

• [FRONT] Upgrade kernl's M4M configuration with the Insiders version.

  #306 opened 2 years ago by white-gorilla
  0

• Where can I install the software (on which hosting I can try it)

  #253 opened 2 years ago by Oxi84
  5

• [FRONT] Remove/comment empty sections or put something more engaging, less crude.

  #259 opened 2 years ago by white-gorilla
  0

• Torchdynamo + Inductor faster than kernl in t5 e2e example?

  #232 opened 2 years ago by wangjunhaoumich
  5

• bug: tests failing at nvidia-driver-530

  #304 opened 2 years ago by christallire
  0

• Accelerate warmup IRL?

  #242 opened 2 years ago by JaheimLee
  4

• feature: non verbose CI

  #302 opened 2 years ago by pommedeterresautee
  1

• Installation problem

  #293 opened 2 years ago by p-christ
  7

• feature: run tests on CI

  #289 opened 2 years ago by pommedeterresautee
  2

• I run the bert e2e example, if batch is not 1, I get an error!!!

  #286 opened 2 years ago by lichun-wang
  2

• bug: Whisper "Segmentation fault (core dumped)" / inference got stuck

  #272 opened 2 years ago by philschmid
  6

• feature: introduce int8 quant kernel

  #288 opened 2 years ago by pommedeterresautee
  0

• [M4M Insider] Upgrade version (kernl + blog) to 4.30.2.

  #258 opened 2 years ago by white-gorilla
  2

• bug: ERROR: Package 'kernl' requires a different Python: 3.8.10 not in '==3.9.*'

  #282 opened 2 years ago by silvacarl2
  1

• version is still 0.1.0

  #273 opened 2 years ago by JaheimLee
  2

• [M4M | BLOG] Updated kernl/docs CI to support M4M Insiders.

  #247 opened 2 years ago by white-gorilla
  0

• feature: reduce memory overhead in CG

  #267 opened 2 years ago by pommedeterresautee
  0

• bug: test test_optimized_model fails since HF transformer 4.26

  #264 opened 2 years ago by pommedeterresautee
  0

• bug: when executed in a specific order, tests crash

  #263 opened 2 years ago by pommedeterresautee
  0

• feature: replace Whisper script by a notebook

  #262 opened 2 years ago by pommedeterresautee
  0

• bug: memory leak

  #256 opened 2 years ago by pommedeterresautee
  1

• [M4M Insiders] Upgrade version.

  #254 opened 2 years ago by white-gorilla
  1

• [M4MInsiders] Targeting a specific version of the parent repository

  #252 opened 2 years ago by white-gorilla
  1

• [M4MInsiders] Fix automatic docker image update

  #248 opened 2 years ago by white-gorilla
  1

• [FRONT] The contribution guide is broken.

  #249 opened 2 years ago by white-gorilla
  0

• Add conventions to triton kernel writing

  #229 opened 2 years ago by gaetansnl
  0

• bug: RuntimeError("GPU compute capability 8.0 (Ampere) or higher is required to use Kernl")

  #246 opened 2 years ago by Oxi84
  1

• bug: memory leak in Pytest / CUDA Graph

  #244 opened 2 years ago by pommedeterresautee
  0

• Cache transformed dynamo graph to speedup warmup

  #234 opened 2 years ago by pommedeterresautee
  1

• Linter and formatter configuration incompatibility

  #230 opened 2 years ago by gaetansnl
  0