Skip to content

hipBLASLt 0.10.0 for ROCm 6.3.0

Latest
Latest
Compare
Choose a tag to compare
@rocm-ci rocm-ci released this 03 Dec 19:49
rocm-6.3.0
9f30df5
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Learn about vigilant mode.

Added

  • Support the V2 CPP extension API for backward compatibility
  • Support for data type Int8 in with Int8 out
  • Support for data type FP32/FP64 for gfx110x
  • Add the Extension API hipblaslt_ext::matmulIsTuned
  • Output atol and rtol for hipblaslt-bench validation
  • Output the bench command for hipblaslt CPP ext API path if HIPBLASLT_LOG_MASK=32 is set
  • Support odd sizes for FP8/BF8 GEMM

Changed

  • Reorganize and add more sample code
  • Add a dependency with the hipblas-common package and remove the dependency with the hipblas package

Optimized

  • Support fused kernel for HIPBLASLT_MATMUL_DESC_AMAX_D_POINTER for FP8/BF8 data type
  • Improve the library loading time
  • Improve the overall performance of first returned solution

Upcoming changes

  • The V1 CPP extension API will be deprecated in a future release of hipBLASLt
Assets 2
Loading