ggml-cpu: x86 feature detection is specific to x86 (llama/13811)

ggml : allow CUDA graphs when using pipeline parallelism (llama/13814)

cuda : avoid cuGetErrorString (llama/13791)

SYCL: Add non contiguous support in RMS_NORM and NORM kernels (llama/13611)

sycl: Add more debug prints (llama/13640)

vulkan: mark IM2COL as supporting non-contig (llama/13783)

CANN: Add the basic supports of Flash Attention kernel (llama/13627)

SYCL: revert "sycl: simplify bin_bcast_kernel (ggml/13383)" (llama/13752)

ggml-cpu : set openmp wait time if not set (llama/13758)

ggml : add ggml_gelu_erf() CUDA kernel (llama/13719)

CUDA: fix race condition in FA vector kernels (llama/13742)

CANN: Support MUL_MAT_ID for q8_0 and q4_0 (llama/13705)

ggml : fix the order of ggml_unary_op (llama/13718)

vulkan: support CPY from any type to itself (llama/13695)

vulkan: Disable coopmat/coopmat2/bfloat extensions if glslc doesn't support it (llama/13696)

use LOG_WARN to replace `std::cerr` (llama/13657)

sycl : Remove waits from function calls (llama/13702)

SYCL: Avoid using with SYCL-Graph for unsupported nodes (llama/13587)

opencl: Add support for multiple devices (llama/12622)

opencl: fix couple crashes (llama/12795)

ggml : add ggml_gelu_erf() (llama/13667)

musa: Upgrade MUSA SDK version to rc4.0.1 and use mudnn::Unary::IDENTITY op to accelerate D2D memory copy (llama/13647)

vulkan: fix warnings (llama/13626)

CUDA: skip fully masked-out KV in FA vec kernel (llama/13584)

sycl: disable reorder for sycl mulmat (llama/13536)

metal : fix typo in FA kernel comments (llama/13651)

sycl : Overcoming workaround for mmap() allocation on Windows (llama/13482)

Vulkan: Add f32 accumulator support to quantized mul mat to fix GLM4 32B incoherence (llama/13607)

sync : ggml

docs : convert README_sycl.md to utf8 format [no ci] (#3191)

node : enable no_prints to suppress all output (#3189)

talk-llama : fix for swedish umlauts + expose model inference settings in talk-llama.cpp (#3187)

docs : fix VAD section heading levels (#3186)

ci : use dynamic libopenblas.dll for window-blas (#3177)

server : Add k6 Load Testing Script (#3175)

docs : add VAD model download instructions [no ci] (#3180)

docs : replace typo "]"with ")" in README (#3179)

whisper : remove redundant assignments (#3178)

whisper : update CMakeLists.txt to handle deprecated gpu Warnings (#3163)

ruby : add GGML_SYCL_DNN option to ruby bindings (#3172)

talk-llama : sync llama.cpp

sync : ggml

CANN: Support MOE Model MUL_MAT_ID (llama/13042)

cmake: use the current build config for vulkan-shaders-gen (llama/13595)

vulkan: move common FA code to flash_attn_base.comp (llama/13556)

vulkan: use scalar FA rather than coopmat2 when N==1 (llama/13554)

metal : add FA-vec kernel for head size 64 (llama/13583)

sycl : fixed compilation warnings (llama/13582)

gguf : use ggml log system (llama/13571)

sycl: simplify bin_bcast_kernel (llama/13383)