Merge requests · Kleidi / KleidiAI

Merge trains

Draft: Implement clamping in SME kernels to match naming

!526 · created Nov 13, 2025 by Hugo OKeeffe
- 1
- 0 of 1 Approvals
updated Nov 13, 2025
GEMV kernel <- f16 LHS x QSI4c32p RHS

!524 · created Nov 11, 2025 by Hugo OKeeffe
- 13
- 0 of 1 Approvals
updated Nov 12, 2025
Add NULL Bias support for rhs_pack_kxn_x32p16x1b_x32_x32_neon

!523 · created Nov 11, 2025 by John McLoughlin
feature
- +5
- 29
- 0 of 1 Approvals
updated Nov 14, 2025
Extend packed buffer description

!521 · created Nov 06, 2025 by Emil Ohlsson
documentation
- 12
- 0 of 1 Approvals
updated Nov 13, 2025
Add 6x4VL SVE MatMul Kernels

!520 · created Nov 05, 2025 by Mohammed Suhail Munshi
feature
- 38
- 0 of 1 Approvals
updated Nov 13, 2025
Add depthwise convolution (dwconv) benchmarking support.

!518 · created Nov 05, 2025 by Cathal Lawlor
enhancement
- +2
- 2
- 0 of 1 Approvals
updated Nov 14, 2025
Add caching to some of the unit tests.

!515 · created Oct 30, 2025 by Jens Elofsson
enhancement
- 19
- 0 of 1 Approvals
updated Nov 12, 2025
Add data generators

!513 · created Oct 28, 2025 by Jakub Sujak
enhancement
- 20
- 0 of 1 Approvals
updated Nov 13, 2025
Add shape generator

!512 · created Oct 28, 2025 by Jakub Sujak
enhancement
- 7
- 0 of 1 Approvals
updated Nov 11, 2025
Add quantize_dynamic API for quantizing reference data in test infrastructure

!511 · created Oct 23, 2025 by james-gross-arm
enhancement
- +4
- 69
- 0 of 1 Approvals
updated Nov 14, 2025
Use a global test seed for all unit tests

!504 · created Oct 20, 2025 by Jakub Sujak
enhancement
- +2
- 18
- 0 of 1 Approvals
updated Nov 11, 2025
Improve FVP-based test runs by enabling parallel execution

!503 · created Oct 20, 2025 by Anton Bondarenko
enhancement
- 14
- 0 of 1 Approvals
updated Nov 07, 2025
Add SME2 3x3 depthwise int8 kernel (stride=1, per-ch sym weight quant, float scale requant)

!489 · created Oct 06, 2025 by Declan Cox
feature
- 144
- 0 of 1 Approvals
updated Nov 04, 2025
Add negative values to test float random generator

!437 · created Aug 21, 2025 by John McLoughlin
enhancement
- +2
- 7
- 0 of 1 Approvals
updated Oct 24, 2025
Draft: Add example to demonstrate micro-kernel fusion at cache level

!388 · created May 28, 2025 by Gian Marco Iodice
support
- 0 of 1 Approvals
updated Jun 10, 2025
Draft: Matmul Micro-kernels F32<- (QSI8DX) LHS x (QAI4CX) RHS

!348 · created Apr 10, 2025 by Anitha Raj
feature
- 17
- 0 of 1 Approvals
updated Sep 12, 2025
Draft: Move matmul_clamp_fp32_bf16p_bf16p into matmul_clamp_f32_bf16p_bf16p

!346 · created Apr 10, 2025 by Anton Bondarenko
major support
- 3
- 0 of 1 Approvals
updated Apr 11, 2025
Draft: Rename kai_lhs_quant_pack_qsi8d32p_f32 to kai_lhs_quant_pack_qsi8d32p_f32_neon

!341 · created Apr 04, 2025 by Anitha Raj
major
- 0 of 1 Approvals
updated Apr 11, 2025
Draft: Rename kai_lhs_quant_pack_qsi8d32p_f32_neon to kai_lhs_quant_pack_qsi8d32pmrx4_f32_neon

!336 · created Apr 02, 2025 by Anitha Raj
major
- 2
- 0 of 1 Approvals
updated Apr 11, 2025
Rename and Optimize QSI4CXP RHS pack micro-kernel

!296 · created Feb 11, 2025 by Anitha Raj
enhancement major
- 7
- 1
- 0 of 1 Approvals
updated Apr 11, 2025