fb30ba9a6c
hexagon: tiling, tracing and optimizations for unary ops ( #25474 )
...
* hexagon: tile wide rows in pointwise unary ops to avoid VTCM overflow
* unary: reject permuted tensors for now (not used by models)
* hex-unary: replace divs with fastdiv
* hex-unary: add vtcm layout and host computed kernel params
* hex-unary: move fastdiv init into kernel params
* hex-unary: add specialized thread functions to improve generated code
* hex-unary: tracing instrumentation for unary ops
* hex-unary: factor out hvx kernels, streamline and remove more duplication
* ggml-hexagon: fix std::min collision with Windows min macro
* hex-cmake: make lto build happy
---------
Co-authored-by: Max Krasnyansky <maxk@qti.qualcomm.com >
2026-07-09 10:15:47 -07:00
Aparna M P and GitHub
f2d1c2f398
hexagon: add VISION RoPE support ( #25216 )
...
* hexagon: add VISION RoPE support
* hexagon: support RoPE on strided half-dim views for all modes
* hex-rope: decouple src0 DMA copy size from row stride
* hex-rope: support non-contiguous dst for RoPE
* hex-rope: fix dst spad pitch for non-contiguous dst
2026-07-08 21:55:00 -07:00
Aparna M P and GitHub
cec51c7a7d
snapdragon: update windows toolchain to use hsdk v6.6.0.0 ( #23552 )
2026-05-23 19:56:41 -07:00
Aparna M P and GitHub
17d22a35b2
hexagon: add MROPE and IMROPE support in HTP rope op ( #23317 )
2026-05-19 14:10:13 -07:00
Aparna M P and GitHub
ac76808e4d
hexagon: enable support for NORM op ( #23319 )
2026-05-19 09:48:21 -07:00
Aparna M P and GitHub
2098fd6169
hexagon: enable non-contiguous row tensor support for unary ops ( #22574 )
2026-05-01 10:09:23 -07:00
0dedb9ef7a
hexagon: add support for FILL op ( #22198 )
...
Co-authored-by: Max Krasnyansky <maxk@qti.qualcomm.com >
2026-04-21 16:24:20 -07:00
8710e5f9b9
hexagon: improve RMS_NORM and DIV accuracy ( #21251 )
...
* hexagon-rms_norm: fix RMS_NORM for non-aligned tensor sizes
Co-authored-by: Krishna Sridhar <srsr@qti.qualcomm.com >
* hexagon-div: perform DIV in fp16 domain for lower dsp archs
---------
Co-authored-by: Krishna Sridhar <srsr@qti.qualcomm.com >
2026-04-01 08:43:08 -07:00