sync : whisper.cpp#3239

Merged

ggerganov merged 30 commits intomasterfrom

sync-whisper.cpp-25-06-10

Jun 10, 2025

Commits on Jun 10, 2025

Add in-build ggml::ggml ALIAS library (ggml/1260)

dg0yt
authored and
ggerganov
committed
ggml: check if non-native endian model is being loaded (llama/13943)

taronaeo
authored and
ggerganov
committed
gguf: fix failure on version == 0 (llama/13956)

JohannesGaessler
authored and
ggerganov
committed
sycl: quantize and reorder the input to q8_1 when reorder is enabled (llama/13826)

authored and
ggerganov
committed
cmake : Handle mixed-case 'Power' strings in POWER CPU detection (llama/13966)

authored and
ggerganov
committed
metal : use F32 accumulators in FA kernels (llama/13975)
ggerganov
committed
OpenCL: Add concat, tsembd, upscale, tanh, pad and repeat (llama/13840)

rmatif
authored and
ggerganov
committed
opencl: add `backend_synchronize` (llama/13939)

lhez
authored and
ggerganov
committed
vulkan: fix warnings in perf logger querypool code (llama/13937)

jeffbolznv
authored and
ggerganov
committed
CUDA: fix FTZ in FA for Gemma 3 (llama/13991)

JohannesGaessler
authored and
ggerganov
committed
releases : use dl backend for linux release, remove arm64 linux release (llama/13996)

slaren
authored and
ggerganov
committed
ggml-vulkan: adds support for op CONV_TRANSPOSE_1D (llama/13813)

etasnadi
authored and
ggerganov
committed
vulkan: automatically deduce size of push constants (llama/13936)

jeffbolznv
authored and
ggerganov
committed
llama : allow using mmap without PrefetchVirtualMemory, apply GGML_WIN_VER to llama.cpp sources (llama/14013)

slaren
authored and
ggerganov
committed
vulkan: Enable VK_KHR_cooperative_matrix extension for Intel Xe2 GPUs (llama/14001)

rillomas
authored and
ggerganov
committed
SYCL: Implement few same quantized type copy kernels (llama/13739)

qnixsynapse
authored and
ggerganov
committed
cuda : fix buffer type check with integrated GPUs (llama/14069)

slaren
authored and
ggerganov
committed
sycl: Add reorder to Q6_K mmvq implementation (llama/13885)

s-Nick
authored and
ggerganov
committed
CANN: Simplify the environment variable setting(#13104)

bachelor-dou
authored and
ggerganov
committed
cuda : fix device sync on buffer clear (llama/14033)

slaren
authored and
ggerganov
committed
ggml-cpu : split arch-specific implementations (llama/13892)

xctan
and
ggerganov
committed
metal : use less stack memory in FA kernel (llama/14088)
ggerganov
committed
sync : ggml
ggerganov
committed
talk-llama : sync llama.cpp
ggerganov
committed
files : remove old sources
ggerganov
committed
sync : ggml
ggerganov
committed
files : remove old sources (part 2)
ggerganov
committed
sync : ggml
ggerganov
committed
android : fix builds (#0)
ggerganov
committed
ggml : fix weak alias win32 (#0)
ggerganov
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

sync : whisper.cpp#3239

sync : whisper.cpp#3239
ggerganov merged 30 commits intomasterfrom
sync-whisper.cpp-25-06-10

Commits on Jun 10, 2025

Add in-build ggml::ggml ALIAS library (ggml/1260)

ggml: check if non-native endian model is being loaded (llama/13943)

gguf: fix failure on version == 0 (llama/13956)

sycl: quantize and reorder the input to q8_1 when reorder is enabled (llama/13826)

cmake : Handle mixed-case 'Power' strings in POWER CPU detection (llama/13966)

metal : use F32 accumulators in FA kernels (llama/13975)

OpenCL: Add concat, tsembd, upscale, tanh, pad and repeat (llama/13840)

opencl: add `backend_synchronize` (llama/13939)

vulkan: fix warnings in perf logger querypool code (llama/13937)

CUDA: fix FTZ in FA for Gemma 3 (llama/13991)

releases : use dl backend for linux release, remove arm64 linux release (llama/13996)

ggml-vulkan: adds support for op CONV_TRANSPOSE_1D (llama/13813)

vulkan: automatically deduce size of push constants (llama/13936)

llama : allow using mmap without PrefetchVirtualMemory, apply GGML_WIN_VER to llama.cpp sources (llama/14013)

vulkan: Enable VK_KHR_cooperative_matrix extension for Intel Xe2 GPUs (llama/14001)

SYCL: Implement few same quantized type copy kernels (llama/13739)

cuda : fix buffer type check with integrated GPUs (llama/14069)

sycl: Add reorder to Q6_K mmvq implementation (llama/13885)

CANN: Simplify the environment variable setting(#13104)

cuda : fix device sync on buffer clear (llama/14033)

ggml-cpu : split arch-specific implementations (llama/13892)

metal : use less stack memory in FA kernel (llama/14088)

sync : ggml

talk-llama : sync llama.cpp

files : remove old sources

sync : ggml

files : remove old sources (part 2)

sync : ggml

android : fix builds (#0)

ggml : fix weak alias win32 (#0)

sync : whisper.cpp#3239ggerganov merged 30 commits intomasterggml-org/whisper.cpp:masterfrom sync-whisper.cpp-25-06-10ggml-org/whisper.cpp:sync-whisper.cpp-25-06-10Copy head branch name to clipboard

Commits

Commits on Jun 10, 2025

sync : whisper.cpp#3239
ggerganov merged 30 commits intomasterfrom
sync-whisper.cpp-25-06-10