66559e04 · F32 copysign subgraph API · Jun 18, 2024

a0f7fae5 · Add CopySign op · Jun 18, 2024

144ffb8b · F16-GEMM combine M loops to allow multiple ports co-issue instructions · Jun 18, 2024

4502b983 · Internal config change · Jun 18, 2024

45deebbe · Clean up the `CMakeLists.txt` file a bit. · Jun 18, 2024

cdde0641 · Add wrappers for the KleidiAI `qp8-qc4w` GEMM microkernels. · Jun 18, 2024

8e5f0b83 · Hook up the new KleidiAI GEMM microkernels to the `fully-connected` operator. · Jun 18, 2024

73c406a0 · Add SIMD copysign microkernels · Jun 18, 2024

24ed1f07 · Internal config change · Jun 18, 2024

9c807a84 · Add Log op · Jun 18, 2024

f19965e5 · F16 F32acc reduce microkernels write to F32 · Jun 14, 2024

ca8ee064 · Added f32-hvx.h to support unified unary simd interface. · Jun 13, 2024

4091800b · gemm-benchmark remove unused variables · Jun 11, 2024

aee6a2f2 · Reduce repeated expression of error-prone specifications (compiler flags) in the bazel build · Jun 11, 2024

569b3cff · Add dependencies to the KleidiAI library to both the `BUILD` and `CMakeLists.txt` files. · Jun 11, 2024

f4628f4f · Add `copy_bias` helper functions and use it to refactor packing.c. · Jun 11, 2024

46cdc3c2 · Run CI when an external PR is approved. · Jun 11, 2024

406bf2ca · Clean up dependencies. · Jun 11, 2024

bb1650b7 · Enable AMX using pytorch/cpuinfo cpuinfo_has_x86_amx_int8() · Jun 09, 2024

96f1f2df · LLM decode benchmarks fill the cache with a predefined number of tokens before starting decoding. · Jun 06, 2024