Todor BoinovskiandMax Krasnyansky 0278d8362d hexagon: activation ops update (#25974) ...

* hex-geglu: optimized all-in-one geglu microkernel

* hex-geglu: enable non-contiguous src and strided DMA

* hex-act: enable non-contiguous srs and strided DMA for rest of ACT ops

* hex-act: generalize GLU per-thread functions via DEFINE_GLU_PER_THREAD macro

* hexagon: move UNARY_SILU and UNARY_GELU to unary-ops

* hex-act: replace the generic ops_context scratchpad usage with a local htp_vtcm_layout computation per act op.

---------

Co-authored-by: Max Krasnyansky <maxk@qti.qualcomm.com>

2026-07-22 09:25:04 -07:00

..

act-ops.c

hexagon: activation ops update (#25974)

2026-07-22 09:25:04 -07:00

argsort-ops.c

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

binary-ops.c

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

cmake-toolchain.cmake

hexagon: MUL_MAT and MUL_MAT_ID rework : 32x32 tiled weight repack, kernel-params, cached graphs (#24954)

2026-06-24 12:14:25 -07:00

CMakeLists.txt

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

concat-ops.c

hexagon: new vtcm layouts and improved pipelines for MUL_MAT, MUL_MAT_ID and FLASH_ATTN_EXT (#25425)

2026-07-08 07:38:27 -07:00

cpy-ops.c

hexagon: add support for CONCAT op (#23648)

2026-05-26 06:20:05 -07:00

cumsum-ops.c

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

diag-ops.c

Hexagon: DAIG op (#22195)

2026-04-21 14:16:04 -07:00

dma-queue.c

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

dma-queue.h

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

fill-ops.c

hexagon: add support for FILL op (#22198)

2026-04-21 16:24:20 -07:00

flash-attn-ops.c

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

flash-attn-ops.h

hexagon: new vtcm layouts and improved pipelines for MUL_MAT, MUL_MAT_ID and FLASH_ATTN_EXT (#25425)

2026-07-08 07:38:27 -07:00

gated-delta-net-ops.c

Remove padding and multiple D2D copies for MTP (#24086)

2026-06-10 23:21:16 +05:30

get-rows-ops.c

hexagon: add support for CONCAT op (#23648)

2026-05-26 06:20:05 -07:00

hex-bitmap.h

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

hex-common.h

hexagon: MUL_MAT and MUL_MAT_ID rework : 32x32 tiled weight repack, kernel-params, cached graphs (#24954)

2026-06-24 12:14:25 -07:00

hex-dma.h

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

hex-dump.h

hexagon: general DMA and Binary Op fixes for large strides (#20918)

2026-03-23 15:33:49 -07:00

hex-fastdiv.h

hexagon: support for OP_CPY, host buffers now optional, hvx-utils refactoring and optimizations (#18822)

2026-01-14 21:46:12 -08:00

hex-profile.h

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

hex-utils.h

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

hmx-fa-kernels.h

hexagon: new vtcm layouts and improved pipelines for MUL_MAT, MUL_MAT_ID and FLASH_ATTN_EXT (#25425)

2026-07-08 07:38:27 -07:00

hmx-mm-kernels-tiled.h

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

hmx-queue.c

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

hmx-queue.h

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

hmx-utils.h

hexagon: new vtcm layouts and improved pipelines for MUL_MAT, MUL_MAT_ID and FLASH_ATTN_EXT (#25425)

2026-07-08 07:38:27 -07:00

htp_iface.idl

hexagon: MUL_MAT and MUL_MAT_ID rework : 32x32 tiled weight repack, kernel-params, cached graphs (#24954)

2026-06-24 12:14:25 -07:00

htp-ctx.h

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

htp-ops.h

hexagon: add CLAMP op (#25934)

2026-07-20 16:12:09 -07:00

htp-tensor.c

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

htp-tensor.h

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

htp-vtcm.h

hexagon: new vtcm layouts and improved pipelines for MUL_MAT, MUL_MAT_ID and FLASH_ATTN_EXT (#25425)

2026-07-08 07:38:27 -07:00

hvx-arith.h

hexagon: add fp16 support for binary ops: add,sub,mul,div (#20139)

2026-03-05 18:29:13 -08:00

hvx-base.h

hexagon: flash attention rework (optimizations, accuracy improvements, etc) (#25085)

2026-07-01 06:59:19 -07:00

hvx-copy.h

hexagon: hmx flash attention (#22347)

2026-05-01 20:29:13 -07:00

hvx-div.h

hexagon: improve RMS_NORM and DIV accuracy (#21251)

2026-04-01 08:43:08 -07:00

hvx-dump.h

ggml-hexagon: flash-attention and reduce-sum optimizations (#19141)

2026-01-30 21:14:20 -08:00

hvx-exp.h

hexagon: flash attention rework (optimizations, accuracy improvements, etc) (#25085)

2026-07-01 06:59:19 -07:00

hvx-fa-kernels.h

hexagon: flash attention rework (optimizations, accuracy improvements, etc) (#25085)

2026-07-01 06:59:19 -07:00

hvx-flash-attn.h

hexagon: MUL_MAT, MUL_MAT_ID, FLASH_ATTN and GDN cleanup and optimizations for latest models (#23989)

2026-06-01 23:40:08 -07:00

hvx-floor.h

hexagon: support for OP_CPY, host buffers now optional, hvx-utils refactoring and optimizations (#18822)

2026-01-14 21:46:12 -08:00

hvx-inverse.h

hexagon: add fp16 support for binary ops: add,sub,mul,div (#20139)

2026-03-05 18:29:13 -08:00

hvx-log.h

hexagon: MUL_MAT, MUL_MAT_ID, FLASH_ATTN and GDN cleanup and optimizations for latest models (#23989)

2026-06-01 23:40:08 -07:00

hvx-mm-kernels-flat.h

hexagon: flash attention rework (optimizations, accuracy improvements, etc) (#25085)

2026-07-01 06:59:19 -07:00

hvx-mm-kernels-tiled.h

hexagon: flash attention rework (optimizations, accuracy improvements, etc) (#25085)

2026-07-01 06:59:19 -07:00

hvx-norm.h

hexagon: tiling, tracing and optimizations for unary ops (#25474)

2026-07-09 10:15:47 -07:00

hvx-pow.h

hexagon: MUL_MAT, MUL_MAT_ID, FLASH_ATTN and GDN cleanup and optimizations for latest models (#23989)

2026-06-01 23:40:08 -07:00

hvx-reduce.h

hexagon: Flash Attention optimizations (dma, mpyacc, multi-row) and MatMul updates (#20118)

2026-03-04 21:55:29 -08:00

hvx-repl.h

hexagon: eliminate scalar VTCM loads via HVX splat helpers (#22993)

2026-05-12 17:28:02 -07:00

hvx-scale.h

hexagon: support for OP_CPY, host buffers now optional, hvx-utils refactoring and optimizations (#18822)

2026-01-14 21:46:12 -08:00

hvx-sigmoid.h

hexagon: flash attention rework (optimizations, accuracy improvements, etc) (#25085)

2026-07-01 06:59:19 -07:00

hvx-sin-cos.h

hexagon: add support for CONCAT op (#23648)

2026-05-26 06:20:05 -07:00

hvx-sqrt.h

hexagon: Add ARGSORT, DIV, SQR, SQRT, SUM_ROWS, GEGLU (#19406)

2026-02-10 23:21:12 -08:00

hvx-types.h

hexagon: support for OP_CPY, host buffers now optional, hvx-utils refactoring and optimizations (#18822)

2026-01-14 21:46:12 -08:00

hvx-utils.h

hexagon: tiling, tracing and optimizations for unary ops (#25474)

2026-07-09 10:15:47 -07:00

main.c

hexagon: activation ops update (#25974)

2026-07-22 09:25:04 -07:00

matmul-ops.c

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

matmul-ops.h

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

pad-ops.c

hexagon: MUL_MAT, MUL_MAT_ID, FLASH_ATTN and GDN cleanup and optimizations for latest models (#23989)

2026-06-01 23:40:08 -07:00

repeat-ops.c

hexagon: improved Op queuing, buffer and cache management (#21705)

2026-04-10 15:47:43 -07:00

rope-ops.c

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

set-rows-ops.c

hexagon: add support for CONCAT op (#23648)

2026-05-26 06:20:05 -07:00

softmax-ops.c

hexagon: improved Op queuing, buffer and cache management (#21705)

2026-04-10 15:47:43 -07:00

solve-tri-ops.c

hexagon: add SOLVE_TRI op (#21974)

2026-04-23 18:39:13 -07:00

ssm-conv.c

fix(hexagon): use padded stride for ssm-conv weights (#24470)

2026-06-20 14:58:49 -07:00

sum-rows-ops.c

hexagon: improved Op queuing, buffer and cache management (#21705)

2026-04-10 15:47:43 -07:00

unary-ops.c

hexagon: activation ops update (#25974)

2026-07-22 09:25:04 -07:00

unary-ops.h

hexagon: activation ops update (#25974)

2026-07-22 09:25:04 -07:00

work-queue.c

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00

work-queue.h

hexagon: L2 cache handling rework (dirty bit tracking with lazy flushing) and more MUL_MAT updates (#25762)

2026-07-16 09:28:04 -07:00