Benjamin Brock

No member named `task` in namespace `tbb`

1

I'm encountering this compile error when trying to compile oneDPL with the asynchronous algorithm API---not sure if I need to add a compiler flag, different include order, or what. I...

AWS with Intel 82599, ixgbevf?

3

Is it possible to get urdma working on AWS? I have a c5 VM with an Intel 82599 with ixgbevf. Once I get urdma installed, I try: ```Bash ubuntu@ip-172-31-44-242:~$ sudo...

`gemm` throws exception on PVC

16

# Summary I'm trying to use `gemm` on PVC, but it keeps throwing an exception. Please let me know where I'm going wrong. I am attempting to use `gemm` and...

Implement `direct_iterator` and `make_direct_iterator`

3

Implement `direct_iterator` and `make_direct_iterator`, which allow users to wrap device iterators that should be used directly inside SYCL kernels by oneDPL. This PR addresses #855 and #854. This will likely...

I've begun implementing the GraphBLAS C++ Specification on top of GBTL. Creating this as a new PR to keep everything on the cmu-sei GBTL repo, instead of using my fork.

Add distributed ranges as experimental feature.

9

This draft PR adds distributed ranges as an experimental feature, inside the `onedpl::experimental::dr` namespace.

C++ Modules Do Not Work with SYCL

2

C++ modules do not seem to currently work with intel/llvm when SYCL is used. I'm curious if this is likely to be added in the near future, or if there...

Performance bug for odd sizes in `for_each`, likely other algorithms

1

SYCL `parallel_for`, as well as `for_each` and likely other oneDPL algorithms, have significantly degraded performance for odd sizes. I first noticed this with the Black-Scholes benchmark, where I was seeing...

# Description oneMKL's CMake currently checks `CXX_COMPILER_NAME` and uses that to determine whether SYCL is supported. This prevents SYCL support from being detected if the user compiles with their own...