oneapi-src/oneDNN v1.6.4

on GitHub

This is a patch release containing the following changes to v1.6.3:

• Fixed performance regression in dnnl_sgemm with N=1 (379a216, f35e991)
• Extended matmul to support multiple demensions and broadcast (0728f26)
• Fixed performance regression for convolution weight gradient implementation for Intel AVX2(9ab050b, 6cd0c35)
• Fixed unknown primitive kind assertion on GPU (c95a01c)
• Fixed build issue on Windows for the case when oneDNN is built as submodule (2fceddf)
• Fixed issues with NaN results produced by dnnl_sgemm in some scenarios (5ce95ef)
• Improved performance for convolution backpropagation with 1x1 filter and NHWC activations on systems with Intel AVX2 support (74bfc74)
• Fixed correctness issue for convolution with 3D spatial (bf6ee84)
• Fixed potential segmentation fault when destroying RNN primitive (0d9839b)
• Fixed performance regression for fp32 convolutions Intel AVX512 implementation (668e282)