Qianshui-Jiang opened a new pull request, #11638: URL: https://github.com/apache/tvm/pull/11638
We now have `USE_MKLDNN` (for cblas with matmul/dense) and `USE_DNNL_CODEGEN` (for DNNL/OneDNN with matmal/dense/conv2d .etc in BYOC ). Essentially `MKL-DNN` was discarded by Intel and updated as `DNNL` (aka Intel OneDNN), as part of Intel OneAPI , so this PR mainly focus on deprecate MKL-DNN and use DNNL for both cases (e.g., -libs and BYOC), include: - Unified the config.cmake flags USE_MKLDNN and USE_DNNL_CODEGEN to USE_DNNL， setup USE_DNNL can make both -libs and dnnl_codegen(json runtime) enabled. - Update all the symbol like mkldnn to dnnl in codes. Some issue in Rust code not been updated, since tvm_build 0.2.4 should be updated first, we also updated [here in tvm_build](https://github.com/octoml/tvm-build/pull/15) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
