oneMKL

oneMKL copied to clipboard

# Summary The oneMKL interface library supports linking the the DFT backends (to allow for compile-time dispatch) instead of linking to the main library itself and using the dynamic dispatch...

hjabird

Test CI

3

toxicscum

#include <oneapi/mkl.hpp>

2

hello, i'm windows10, oneapi user. when i write oneapi code with onemkl, i can't use oneapi::mkl::rng::device namespace. cmakelists.txt include compile options /Qmkl:sequential /Qmkl-ilp64:sequential https://github.com/oneapi-src/oneMKL/blob/develop/include/oneapi/mkl.hpp https://github.com/oneapi-src/oneMKL/blob/develop/include/oneapi/mkl/rng.hpp in there, header did not...

tb5874

[CUDA][HIP] Use device to get native context

5

Since https://github.com/oneapi-src/unified-runtime/pull/999 it is no longer valid to get the native context from the SYCL context on a multi GPU system. The get native func for contexts has been deprecated...

hdelan

# Summary CUDA backend for BLAS domain (and potentially others as well) does not build with hipSYCL (AdaptiveCpp) # Version oneMKL: 402a7a4 AdaptiveCpp: 7041cdf # Environment Tested on Nivida GPU....

mmeterel

Some tests hang/slow down significantly when running with portBLAS

1

It was found that some tests hang/slow down significantly when running with portBLAS. Tested on Intel CPU/GPU, NVIDIA GPU with DPC++ compiler from oneAPI Base Toolkit 2023.2 and opensource implementation...

toxicscum

Don't use `sycl::detail` namespace

2

`sycl::detail::pi` is used in: https://github.com/oneapi-src/oneMKL/blob/6c5f7ea783a7fe828c0de3c064c3bc837727524d/src/blas/backends/cublas/cublas_scope_handle.cpp#L123 Please note there `detail` sub-namespace is reserved for implementations and there is no guarantees about its content. Not only that is non-portable, but it is...

AlexeySachkov

Could the oneMKL functions "omat" and "imat" be combined ? The cublas Sgeam supports in-place and out-of-place matrix operations.

jinz2014

What is the supported precision for gemm in oneMKL open source project ?

2

As far as I understand, the oneMKL open source project provides implementation for the Intel MKL product, with multiple hardware support. I was checking the documentation for both of them...

FatmaElbadry2

// cuBLAS does not support int64_t as return type for the data Please see https://docs.nvidia.com/cuda/cublas/index.html#bit-integer-interface Thanks

jinz2014