ROCm / rocBLAS

Next generation BLAS implementation for ROCm platform

https://rocm.docs.amd.com/projects/rocBLAS/en/latest/

ROCm/rocBLAS Issues

[Question] The source code of highly optimized gemm kernel
Updated 9 days ago2
[Bug]: A few gemm_ex solutions allocate an unusually large workspace.
Updated 19 days ago
[Issue]: Build error with different cpp files
Updated a month ago1
[Bug]: rocblas_gemm_ex with m==1 fp16 inputs/outputs f32 compute slower than a quite naive gemv kernel on MI100
Updated a month ago18
rocBLAS performance differs with nonuniform data
Closed a month ago3
[Bug]: rocblas link fails with relocation R_X86_64_PC32 out of range
Updated a month ago27
[Question]: rocBLAS determinism with GPU atomics
Closed 3 months ago4
Missing functions
Closed 3 months ago12
[Bug]: rocblas-bench `--verify` / `-v` option seems to be ignored.
Closed 3 months ago1
Is there a way to attach rocgdb on kernel?
Closed 4 months ago2
[HIPIFY][#1281][ROC][CUDA 12.0][feature] Support of "pure int64_t" APIs in ROC, since they are introduced in cuBLAS 12.0
Closed 4 months ago3
[Feature]: Two tile gemm enabling for MI250
Closed 4 months ago2
Documentation clarification about non-default stream switching
Closed 4 months ago1
[Bug]: Error when compiling "ModuleNotFoundError: No module named 'Tensile.TensileCreateLibrary'"
Closed 5 months ago11
[Bug]: rocBLAS fails tests badly in FP16 for distro packages
Closed 5 months ago9
[Feature]: old Tensile_LOGIC source is need, like navi21 navi22 navi23 navi24 for diy
Closed 5 months ago9
Please provide pkg-config file
Closed 5 months ago6
[Bug]: Failed to build rocblas-5.6.0 with Tensile from source
Closed a year ago6
rocblas thread safety and level3 initialization overhead
Closed 6 months ago5
[Bug]: rdeps.py dont seems work anymore.
Closed a year ago23
TypeError: unsupported operand type(s) for //: 'str' and 'str'
Closed 6 months ago
[Bug]: Illegal Instruction Crash in librocblas.so.3 due to AVX Instructions on Non-AVX CPU
Closed 6 months ago11
[Bug]: Unable to build rocBLAS 6.0.2 from source
Closed 7 months ago4
[Feature]: Support for gfx1036
Closed 7 months ago3
[Feature]: allow data to be nullptr when querying for workspace buffer size
Closed 8 months ago5
[Bug]: Tensile Library Crashes when trying to implement llama_index RAG with AMD GPU using ROCm 6.0.0
Closed 8 months ago2
[Bug]: Incorrect results when using GPUs with different architectures
Closed 8 months ago18
Cmake compiler rocBLAS errorr[Feature]:
Closed 8 months ago3
gfx90c
Closed 8 months ago2
[Issue]: Missing input file and output command on rocBLAS documentation
Closed 9 months ago4
[Bug]: Stop breaking backwards compatibility or at least warn
Updated 10 months ago5
[Bug]: Issue with CMake Configuration for ROCm BLAS
Closed a year ago3
[Bug]: rocBLAS error: Cannot read TensileLibrary.dat: Illegal seek (AMD Radeon integrated)
Closed a year ago4
[Feature]: licensing, use a SPDX license
Closed a year ago4
atomics_allowed always true
Closed a year ago1
[Bug]: Failed to build rocblas-5.7.0 with Tensile from source
Closed a year ago8
[Feature]: clarify the supported devices
Closed a year ago2
[Bug]: rocblas cannot load TensileLibrary.dat
Closed a year ago3
[Bug]: build is breaking with hip_lite.
Closed a year ago4
[Bug]: rocBLAS gemm in python-interface produces wrong results when called more than once
Closed a year ago11
[Bug]: TensileCreateLibrary prepAsm error
Closed a year ago5
TensileLibrary.dat not found
Closed a year ago4
[Feature]: Support equivalent of cublasCherkEx()
Updated a year ago6
[Bug]: Different behaviour depending on include order
Closed 2 years ago1
[Bug]: Is latest dev branch compatible with pytorch, ROCm5.4.1?
Closed 2 years ago3
[Bug]: rocBLAS error: Cannot read TensileLibrary.dat
Closed 2 years ago6
How to use roc-obj with rocblas-bench
Closed 2 years ago16
How to perform a rocBLAS test installation in my home directory
Closed 2 years ago13
[Bug]: pre_checkin_blas1_nan_check failed on Radeon 6700XT
Closed 2 years ago2
[HIPIFY][ROC][question] Is typecasting of `int` to `rocblas_int` needed?
Closed 2 years ago6