Hello,

The v24.04 release of Compute Library is out and comes with a collection of improvements and new features.

Source code and prebuilt binaries are available at: https://github.com/ARM-software/ComputeLibrary/releases/tag/v24.04

Highlights of the release:

* Add Bfloat16 data type support for NEMatMulhttps://arm-software.github.io/ComputeLibrary/v24.04/classarm__compute_1_1_n_e_mat_mul.xhtml. * Add support for SoftMax in SME2 for FP32 and FP16. * Add support for in place accumulation to CPU GEMM kernels. * Add low-precision Int8 * Int8 -> FP32 CPU GEMM which dequantizes after multiplication * Add is_dynamic flag to QuantizationInfohttps://arm-software.github.io/ComputeLibrary/v24.04/classarm__compute_1_1_quantization_info.xhtml to signal to operators that it may change after configuration * Performance optimizations: * Optimize start-up time of NEConvolutionLayerhttps://arm-software.github.io/ComputeLibrary/v24.04/classarm__compute_1_1_n_e_convolution_layer.xhtml for some input configurations where GeMM is selected as the convolution algorithm * Optimize NEConvolutionLayerhttps://arm-software.github.io/ComputeLibrary/v24.04/classarm__compute_1_1_n_e_convolution_layer.xhtml for input tensor size > 1e7 bytes and weight tensor height > 7 * Optimize NESoftmaxLayerhttps://arm-software.github.io/ComputeLibrary/v24.04/namespacearm__compute.xhtml#a4df2143ca0a3bdbbbc54b440a52541cd for axis != 0 by natively supporting higher axes up to axis 3.

IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.