CUDA: batched+noncont MMQ, refactor bs>1 MoE code (llama/13199)

vulkan: use uint array index to avoid glslang bug (llama/13193)

ggml : fix ppc64le build (llama/13176)

feat(ggml-cpu): enable z17 compile (llama/13182)

CUDA: fix non-cont. inputs for batched mat mul (llama/13155)

fix(rpc): Improve input validation and error handling (llama/13069)

SYCL: Add all missing unary kernels (llama/13074)

musa: fix typo in cc control (llama/13144)

CUDA: fix q_nope_absorbed prec for DS 2 Lite f16 (llama/13137)

musa: fix build warning (llama/13129)

ggml: move fp16/bf16 conversion optimizations to CPU backend + export conversion APIs (llama/13107)

change the reorder tensor from init to execute OP (llama/13003)

rpc : do not wait for response when sending RPC_CMD_SET_TENSOR (llama/12943)

ggml : fix ggml_gallocr_ptr type (ggml/1205)

whisper : add check that target name exists (#3103)

server : add --no-gpu option to print usage output (#3098)

ruby : ignore "Downloading" output in test_log_suppress (#3106)

make : fix samples glob pattern (#3100)

ggml : suppress Windows compiler warnings (#3075)

whisper : fix grammar advance stack warning (#3087)

examples : expose language detection probabilities to server example (#3044)

whisper : remove empty .gitmodules file [no ci] (#3085)

talk-llama : sync llama.cpp (#3084)

ci : disable publishing of java binding [no ci] (#3086)

build : Add Moore Threads GPU support and update GitHub workflow for MUSA build (#3069)

examples : fix deprecated FFmpeg functions (#3073)

ruby : add encoder begin callback related methods (#3076)

ci : enable bindings java job (#3070)

ruby : add cmake option (#0)

cuda : fix unused variable compile warning (#0)

sync : ggml

opencl : remove obsolete files (skip) (ggml/1200)

sync : ggml

opencl: split ggml-opencl.cl into multiple files and cleanup (llama/12886)

ggml : fix trailing whitespaces (llama/0)

CUDA: use switch statements in constexpr functions (llama/13095)

metal : fix floating-point range of attention scores in FA kernels (llama/13090)

vulkan: matmul gcn tuning (llama/13016)

CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID (llama/13014)

ggml : add SSE 4.2 and x64 base variant for CPUs without AVX (llama/12871)

SYCL: Add non-contiguous support in ROPE (llama/12993)

vulkan: support noncontiguous rms_norm (llama/13031)

metal: add neg operator (llama/13029)

SYCL: Refactor and enable FP16 in binary broadcast OPs (llama/12975)

rpc : add RPC_CMD_HELLO (llama/12955)

graph : make FA compatible with MLA + add initial Metal kernels (llama/12953)

ggml: Re-enable CUDA graphs in presence of CONT and DUP nodes (llama/12970)

CANN: Add support for async operator submission (llama/12864)

opencl: fix incorrect local_size index in profiling log (llama/12868)

vulkan: enable coopmat2 FA gqa and split_k optimizations more often (llama/12931)