Merge requests · Kleidi / KleidiAI

Merge trains

Matmul Micro-kernel(MxN) F16 <- (QSI8D32) LHS x (QAI4C32) RHS

!389 · created May 29, 2025 by Anitha Raj
feature
- Merged
- 6
- Approved
updated Jun 04, 2025
Matmul Micro-kernels F32 <- QAI8DXP(LHS) x QSI8CXP(RHS) optimized for SME

!378 · created May 19, 2025 by Anitha Raj
feature
- Merged
- 29
- Approved
updated May 30, 2025
Add FP16 IGEMM support

!360 · created Apr 16, 2025 by Emil Ohlsson v1.8.0
feature
- Merged
- +3
- 49
- 1
- Approved
updated Apr 23, 2025
Add support for FP32 Indirect GEMM with SME

!357 · created Apr 14, 2025 by Mohammed Suhail Munshi v1.8.0
feature
- Merged
- +2
- 36
- Approved
updated Apr 24, 2025
Matmul Micro-kernels BF16 <- (QAI8DXP) LHS x (QSI4CXP) RHS

!354 · created Apr 14, 2025 by Congregate Migrate
feature
- Merged
- +5
- 90
- Approved
updated Jul 08, 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
Add QAI8 IGEMM kernels

!335 · created Apr 02, 2025 by Emil Ohlsson
feature
- Merged
- +3
- 24
- Approved
updated Apr 09, 2025
Draft: Add QAI8 IGEMM LHS pack

!329 · created Mar 27, 2025 by Emil Ohlsson
feature
- Closed
- 0 of 1 Approvals
updated Apr 09, 2025
Matmul Micro-kernels F32/F16 <- (QSI8D32) LHS x (QAI4C32) RHS

!311 · created Feb 21, 2025 by Anitha Raj
feature
- Merged
- 105
- Approved
updated Apr 09, 2025
Optimize F32 <- QAI8DXP (LHS) x QSI4C32P (RHS) for 4x8 i8mm

!302 · created Feb 17, 2025 by Michael Kozlov
feature
- Merged
- Approved
updated Feb 20, 2025
Optimize F32 <- QAI8DXP 1x8 (LHS) x QSI4C32P 8x8 (RHS) for 1x8 sdot

!300 · created Feb 13, 2025 by Michael Kozlov 1.4.0
feature
- Merged
- 10
- Approved
updated Feb 17, 2025
Add SME2 kernel for QAI8 GEMV

!299 · created Feb 13, 2025 by Emil Ohlsson
feature
- Merged
- +3
- 19
- Approved
updated Feb 17, 2025
Optimize F32 <- QAI8DXP (LHS) x QSI4C32P (RHS) for 1x8 SDOT

!294 · created Feb 07, 2025 by Michael Kozlov
feature
- Merged
- 3
- Approved
updated Feb 12, 2025
Optimize F32 <- QAI8DXP (LHS) x QSI4C32P (RHS) for 4x8 SDOT

!288 · created Jan 30, 2025 by Michael Kozlov
feature
- Merged
- 6
- 1
- Approved
updated Feb 06, 2025
Add MSVC support for BFloat16

!226 · created Dec 10, 2024 by Viet-Hoa Do
feature
- Merged
- +3
- 25
- Approved
updated Jan 09, 2025
Add 16x4 GEMM dotprod microkernel

!211 · created Dec 03, 2024 by Michael Kozlov Next release
feature
- Merged
- 44
- Approved
updated Dec 24, 2024
Add 1x4 GEMV dotprod microkernels

!210 · created Dec 02, 2024 by Michael Kozlov Next release
feature
- Merged
- 26
- Approved
updated Dec 24, 2024
Add initial MSVC build

!207 · created Dec 02, 2024 by Viet-Hoa Do
feature
- Merged
- +3
- 35
- Approved
updated Dec 09, 2024
Enable RHS get_stride API

!199 · created Nov 28, 2024 by Felix Johnny Thomasmathibalan 0.5.0 release
feature
- Merged
- 1
- Approved
updated Nov 29, 2024
Add transposed RHS packing for SME F16 GEMM

!197 · created Nov 27, 2024 by Viet-Hoa Do 0.5.0 release
feature
- Merged
- +3
- 9
- Approved
updated Nov 29, 2024