Skip to content

v3.6-rc

Pre-release
Pre-release
Compare
Choose a tag to compare
@vpirogov vpirogov released this 20 Sep 23:29
· 46 commits to rls-v3.6 since this release

Performance Optimizations

Intel Architecture Processors

  • Improved performance for 4th generation Intel Xeon Scalable processors (formerly Sapphire Rapids).
  • Improved performance for Intel Xeon 6 processors (formerly Granite Rapids).
  • Improved performance of group normalization primitive.
  • Improved bf16 matmul performance with int4 compressed weights on processors with Intel AMX instruction set support.
  • Improved performance of fp8 matmul, pooling, and eltwise primitives on processors with Intel AMX instruction set support.
  • Improved fp32 RNN primitive performance on processors with Intel AVX2 instruction set support.
  • Improved performance of the following subgraphs with Graph API:
    • convolution and binary operation fusions with better layout selection in Graph API.
    • fp8 convolution and unary or binary on processors with Intel AMX instruction set.
    • Scaled Dot Product Attention (SDPA) without scale, Multi-Query Attention (MQA), and Grouped Query Attention (GQA) patterns.
    • LayerNorm, GroupNorm, and SoftMax with int8 quantized output and zero-points.

Intel Graphics Products

  • Improved performance for the Intel Data Center GPU Max Series (formerly Ponte Vecchio).
  • Introduced broad production quality optimizations for Intel Arc Graphics for Intel Core Ultra Processors (Series 2) (formerly Lunar Lake).
  • Introduced broad production quality optimizations for future discrete GPU based on Xe2 architecture (code name Battlemage).
  • Introduced support for Intel Arc Graphics for future Intel Core Ultra Processor (code name Arrow Lake-H).
  • Improved performance of fp8_e5m2 primitives on Intel Data Center GPU Max Series (formerly Ponte Vecchio).
  • Improved matmul and inner product primitives performance for shapes relevant to large language models (LLMs) on GPUs with Intel XMX support.
  • Improved int8 convolution performance with weight zero points.
  • Reduced primitive creation time for softmax, layer normalization, and concat primitives via kernel reuse.
  • Improved performance of the following subgraphs with Graph API:
    • SDPA without scale, MQA, and GQA patterns. f16 variants of these patterns significantly benefit from Intel(R) Xe Matrix Extensions (Intel(R) XMX) support.
    • fp8 convolution and unary or binary on Intel Data Center GPU Max Series.
    • LayerNorm, GroupNorm, and SoftMax with int8 quantized output and zero-points.

AArch64-based Processors

  • Improved fp32 convolution backpropagation performance on processors with SVE support.
  • Improved reorder performance for blocked format on processors with SVE support.
  • Improved bf16 softmax performance on processors with SVE support.
  • Improved batch normalization performance on processors with SVE support.
  • Improved matmul performance on processors with SVE support.
  • Improved fp16 convolution with Arm Compute Library (ACL).
  • Improved matmul performance with ACL.
  • Switched matmul and convolution implementation with ACL to stateless API significantly improving primitive creation time and increasing caching efficiency and performance for these operators.

Functionality

  • Introduced generic GPU support. This implementation relies on portable SYCL kernels and can be used as a starting point to enable new devices in oneDNN.
  • Extended functionality supported on NVIDIA GPUs and AMD GPUs with SYCL based implementations.
  • Enabled support for int8 activations with grouped scales and int8 or int4 compressed weights in matmul primitive. This functionality is implemented on Intel GPUs.
  • Introduces support for stochastic rounding for fp8 data type functionality.
  • [experimental] Extended microkernel API:
    • Introduced int8 quantization support.
    • Extended transform microkernel with transposition support and support for arbitrary strides.
    • Introduced verbose diagnostics support.
  • [experimental] Extended sparse API:
    • Introduced support for sparse memory with coordinate (COO) storage format.
    • Extended matmul primitive to work with sparse memory in COO format. This functionality is implemented on CPUs and Intel GPUs.
  • Introduced int8 support in eltwise primitive with 'clip' algorithm. This functionality is implemented on CPUs.
  • Graph API:
    • Introduced GroupNorm operation and fusions in Graph API.
    • Introduced support for standalone StaticReshape and StaticTranspose operations.

Usability

  • Added examples for SDPA, MQA, and GQA patterns implementation with Graph API.
  • Added an example for deconvolution primitive.
  • Added examples for Vanilla RNN and LBR GRU RNN cells.
  • Introduced support for Intel DPC++/C++ Compiler 2025.0.
  • Introduced interoperability with SYCL Graph record/replay mode.
  • Removed dependency on OpenCL runtime for NVIDIA and AMD GPUs.
  • [experimental] Introduced logging mechanism based on spdlog library.
  • Introduced support for ONEDNN_ENABLE_WORKLOAD build knob for Graph API.
  • Improved performance of get_partitions() function in Graph API.

Validation

  • Introduced protection from out of memory scenarios in benchdnn Graph API driver.

Breaking Changes

Thanks to these Contributors

This release contains contributions from the project core team as well as Abdel @quickwritereader, Adam Jackson @nwnk, Aleksandr Voron @alvoron, Alexey Makarevich @amakarev, Annop Wongwathanarat @annop-w, Daniel Kuts @apach301, @deepeshfujitsu, Fadi Arafeh @fadara01, Fritz Heckel @fwph, Gorokhov Dmitriy @dmitry-gorokhov, Deeksha Kasture @kasturedeeksha, Kentaro Kawakami @kawakami-k, Marek Michalowski @michalowski-arm, @matthias-bonne, @Menooker, Michael Froelich @MichaelFroelich, Nicolas Miller @npmiller, Nikhil Sharma @nikhilfujitsu, @nishith-fujitsu, Permanence AI Coder @Permanence-AI-Coder, Radu Salavat @Radu2k, Renato Barros Arantes @renato-arantes, Robert Cohn @rscohn2, Robert Hardwick @robert-hardwick, Ryo Suzuki @Ryo-not-rio, Shreyas-fuj @Shreyas-fuj, Shu Chen @shu1chen, Siddhartha Menon @Sqvid, Song Jiaming @Litchilitchy, Vladimir Paramuzov @vladimir-paramuzov, Yifei Zhang @yifeizh2. We would also like to thank everyone who asked questions and reported issues.