microsoft/BitBLAS

Issues

• Mismatch output wth INT1 × float16

  #276 opened 4 days ago by 1773226512
  3

• [Feature Request] Implement Split K Tilelang scheduler templates

  #263 opened 9 days ago by LeiWang1999
  2

• [Feature Request] Default Backend Should be changed into `tilelang` instead of `tvm` before v0.0.1 release

  #252 opened 11 days ago by LeiWang1999
  5

• [BUG] Vectorized Bias Add with AtomicAdd may lead to unknown bugs

  #271 opened 11 days ago by LeiWang1999
  3

• [Feature Request] Enhance Database to support reload scheduled tilelang operator

  #269 opened 12 days ago by LeiWang1999
  0

• [Feature Request] Enhance TileLang to auto inject ifthenelse for Out-of-Bounds Memory Access in TileLang

  #265 opened 12 days ago by LeiWang1999
  2

• [Feature Request] LayoutInference pass should be enhanced to analysis vectorize factor cross indices

  #266 opened 12 days ago by LeiWang1999
  1

• [BUG] TVM PopenPoolExecutor may have some bugs on TL Scripts

  #211 opened 3 months ago by LeiWang1999
  3

• Example of bitblas/ladder with dtypes like int3, int5, int6, int7

  #244 opened 2 months ago by yaoyaoding
  4

• [Feature Request] Lazy tvm tensor intrin registration is required to save import time

  #256 opened 14 days ago by LeiWang1999
  2

• [Feature Request] Flash Attention Op should be enhanced with our Scheduler Abstraction

  #264 opened 16 days ago by LeiWang1999
  0

• performance of float16 with fast tuning

  #173 opened 4 months ago by klxy0304
  9

• InternalError: Check failed: func->buffer_map.size() == 0 (3 vs. 0) : This pass must be called after MakePackedAPI

  #259 opened 19 days ago by Cunxiao2002
  4

• Does Bitblas support P40's DP4A for matrix?

  #251 opened 20 days ago by sorasoras
  2

• Issue with integrating with AutoGPTQ

  #257 opened 20 days ago by Steindox
  2

• [Feature Request] Lower Vectorized Loop Pass should be enhanced to adapt layout inference

  #258 opened 23 days ago by LeiWang1999
  2

• Any example to use `float16xfp4_e2m1` matmul?

  #253 opened a month ago by yaoyaoding
  2

• Release Plan of BitBLAS 0.0.1

  #150 opened 4 months ago by LeiWang1999
  5

• [Feature Request] AMD HIP Backend Should be Migrate from Ladder branch

  #208 opened 2 months ago by LeiWang1999
  2

• CUDA error: an illegal memory access was encountered when using BitBlas on multiple GPUs

  #154 opened 4 months ago by mobicham
  12

• [Feature Request]Tail Split Required for Dynamic Symbolic

  #192 opened 2 months ago by LeiWang1999
  4

• TensorIntrin 'mma_i8i8f16_smooth_a_trans_b_smooth_b' is not registered

  #160 opened 2 months ago by huanpengchu
  5

• [Feature Request][TL] Maybe need an annotation to disable part of the ast be translated into TL

  #221 opened 2 months ago by LeiWang1999
  3

• [BUG] Dynamic symoblic may block the lowering phase of async copy

  #218 opened 3 months ago by LeiWang1999
  3

• [Feature Request] Enhance Simplification to remove unused function arguments

  #215 opened 3 months ago by LeiWang1999
  3

• Different outputs based on weight shape

  #157 opened 3 months ago by MekkCyber
  17

• [Feature Request] Parallel Primitive Should be enhanced to improve the performance for irregular shapes

  #209 opened 3 months ago by LeiWang1999
  0

• '

  #206 opened 3 months ago by LeiWang1999
  0

• [BUG] Database or tuning conflict with Multi-GPU Environment

  #204 opened 3 months ago by LeiWang1999
  3

• Installation Failure on CUDA 11.7

  #203 opened 3 months ago by rustic-snob
  2

• How to achieve vLLM Row Parallelism correctly?

  #186 opened 3 months ago by KeremTurgutlu
  5

• Matmul Output Hardware Mismatch

  #187 opened 3 months ago by KeremTurgutlu
  1

• [Bug] Incorrect output for bitnet model

  #184 opened 3 months ago by LeiWang1999
  1

• Segmentation fault when integrated with Ray

  #181 opened 3 months ago by mobicham
  4

• Speed Comparison: BitLinear and nn.Linear

  #118 opened 5 months ago by ZiqingChang
  11

• performance of float16 with fast tuning

  #172 opened 4 months ago by klxy0304
  0

• "nan" in the outputs

  #169 opened 4 months ago by dataparameters
  4

• benchmark error

  #151 opened 4 months ago by MichoChan
  2

• Error in the example

  #156 opened 4 months ago by chenjy11223
  4

• accuracy and performance of bfloat16 with bitblas linear

  #161 opened 4 months ago by AbedKhateeb2
  4

• Speed problem of bitblas.Matmul

  #122 opened 4 months ago by Chenglin-Yang
  1

• Loading BitBlasLinear takes a lot of Time

  #152 opened 4 months ago by MekkCyber
  3

• running issues

  #131 opened 4 months ago by brisker
  27

• [Feature Request] 3-bit support

  #117 opened 5 months ago by mobicham
  6

• Potential Bitblas compat with cuda 12.5

  #120 opened 4 months ago by Qubitium
  6

• [Discussion] Maybe we should use dynamic shared memory by default

  #138 opened 4 months ago by LeiWang1999
  2

• Extracting Operator or CUDA Kernel for TensorRT/ONNX Custom Op Registration

  #123 opened 5 months ago by areddy2022
  3

• Fix code scanning alert - Alert Suppression Report

  #121 opened 5 months ago by LeiWang1999
  1

• RuntimeError: CUDA error: an illegal memory access was encountered

  #116 opened 5 months ago by XA23i
  2

• ValueError: Check failed: lanes <= 4 (8 vs. 4) : Ramp of more than 4 lanes is not allowed.

  #112 opened 5 months ago by XA23i
  2