github llama.cpp · just now · inference

llama.cpp b10772 adds F16 ABS support to Hexagon HTP backend

from English

The llama.cpp release b10772 extends the Hexagon Tensor Processor (HTP) backend's F16 unary operation coverage by adding support for the absolute value (ABS) function.

The update introduces hvx_abs_f16 kernels and dispatchers in hvx-arith.h, mirroring the existing sqr_f16 structure.
ABS is registered as a supported HTP_OP_UNARY operation, allowing ggml_hexagon_precompute_unary_params to fill kernel parameters for F16 nodes.
The codebase merges separate execute_op_unary_f32 and f16 functions into a single execute_op_unary function to avoid duplicating VTCM/DMA plumbing.
On-device verification on QRD8850 (Hexagon v81) confirmed 8/8 passing tests for F16 and F32 ABS with no CPU fallback.

This change enables efficient absolute value computations in FP16 precision on compatible Hexagon hardware, expanding the range of supported unary operations for mobile inference.

Importance 1/3 Trust 1/3 llama.cpp Inference efficiency

Read original

github llama.cpp · 1h ago · 1 view Live

llama.cpp b10773 accepts data: URLs for video and audio input

The llama.cpp server in build b10773 now accepts data: URLs for the `input_video` and `input_audio` fields, resolving a previous issue where such inputs were incorrectly treated as raw base64 strings.

github llama.cpp · 6h ago · 2 views

llama.cpp b10766 adds DeepSeek 4 vision support

The llama.cpp project has released build b10766, which includes a fix to correctly support input vision for the DeepSeek 4 model.

github llama.cpp · 7h ago · 4 views

llama.cpp b10762 adds DeepSeek-V4-Flash-Vision-Exp support

The llama.cpp project has released build b10762, which introduces support for the DeepSeek-V4-Flash-Vision-Exp model. This update also includes handling of minimum and maximum token counts from the command line interface.

github llama.cpp · 7h ago · 5 views

llama.cpp b10763 enables preserve_reasoning by default

The llama.cpp project released build b10763, which changes the default behavior of the `preserve_reasoning` chat template keyword argument. The server now logs the effective state of this kwarg and issues a warning when it is enabled by default on templates that support it.

github llama.cpp · 14h ago · 1 view

llama.cpp b10760 fixes Qwen3-tts-0.6b inference stability

The llama.cpp project released build b10760, which includes specific fixes for the Qwen3-tts-0.6B model to prevent numerical instability during inference.