ci : add apt-get clean to musa Dockerfile (#3275)

ruby : specify Apple frameworks explicitly on build (#3270)

talk-llama : sync llama.cpp

sync : ggml

CUDA: add conv_2d_transpose (llama/14287)

sycl: add usage of enqueue_functions extension (llama/14244)

Implement GGML_CPU_ALL_VARIANTS for PowerPC (llama/14286)

cuda : synchronize graph capture and cublas handle destruction (llama/14288)

ggml : fix repack work size for mul_mat_id (llama/14292)

ggml: Update KleidiAI to v1.9.0 (llama/14277)

CUDA: add conv_2d_dw (llama/14265)

ggml-cpu : remove unnecesary arm feature detection (llama/14281)

build : suppress gcc15 compile warnings (llama/14261)

sycl: Cleanup codepaths in Get Rows in sycl backend (llama/14215)

llamafile : support s390x SIMD instruction set (llama/14273)

Vulkan: Set device max size for host memory to avoid OOM warning and fallback to CPU buffer (llama/14249)

metal : add mean kernel (llama/14267)

ggml-cpu: reduce asm calls for hsum (llama/14037)

ggml-cpu: fix uncaught underscore terminators (llama/14023)

ggml: Add Apple support for GGML_CPU_ALL_VARIANTS (llama/14258)

Add `ggml_roll` (ggml/1274)

android : update CMakeLists.txt to use FetchContent for ggml (#3268)

cmake : fix android build (#3265)

examples : add stereo to mono conversion in read_audio_data (#3266)

talk-llama : sync llama.cpp

sync : ggml

cmake: remove shader-gen step-targets from ggml-vulkan (llama/14226)

ggml-cpu : remove the weak alias trick (llama/14221)

musa: fix build warning (unused variable) (llama/14231)

llama : add thread safety test (llama/14035)

cmake: clean up external project logic for vulkan-shaders-gen (llama/14179)

HIP: disable rocwmma on gfx12 by default until rocm 7.0 (llama/14202)

ggml: Add Android support for GGML_CPU_ALL_VARIANTS (llama/14206)

vulkan: mutex around vkQueueSubmit (llama/14127)

ggml-cpu : rework weak alias on apple targets (llama/14146)

CUDA/HIP: fix ssm_scan on devices where warp size is not 32 (llama/14196)

HIP: Replace usage of depricated preprocessor macro __AMDGCN_WAVEFRONT_SIZE__ (llama/14183)

sycl: Adding additional cpy dbg print output (llama/14034)

SYCL: Bump oneMath commit (llama/14152)

sycl: Remove not needed copy f16->f32 for dnnl mul mat (llama/14125)

cmake : handle whitepsaces in path during metal build (llama/14126)

Implement GGML_CPU_ALL_VARIANTS for ARM (llama/14080)

vulkan: Better thread-safety for command pools/buffers (llama/14116)

vulkan: Track descriptor pools/sets per-context (llama/14109)

opencl: add `mul_mv_id_q4_0_f32_8x_flat` (llama/14003)

Vulkan: Don't default to CPU device (like llvmpipe), even if no other device is available, to allow fallback to CPU backend (llama/14099)

rpc : nicer error messages for RPC server crash (llama/14076)

ggml : disable warnings for tests when using MSVC (ggml/1273)

ggml : remove unused ggml_context_container (ggml/1272)

examples : include examples in msvc disable warn (ggml/1270)