ggml: initial IBM zDNN backend (llama/14975)

HIP: add cmake option to enable compiler output of kernel resource usage metrics (llama/15103)

cmake: Add GGML_BACKEND_DIR option (llama/15074)

HIP: add GGML_HIP_MMQ_MFMA option to allow disableing the MFMA path. (llama/14930)

ggml-cpu : disable GGML_NNPA by default due to instability (llama/14880)

musa: upgrade musa sdk to rc4.2.0 (llama/14498)

ggml: Add initial WebGPU backend (llama/14521)

sync : resolve conflicts (#0)

sync : resolve conflicts (ggml/0)

ggml : support bcast ggml_soft_max_ext, ggml_flash_attn_ext (llama/14435)

ggml : add version function to get lib version (ggml/1286)

ggml-cpu: enable IBM NNPA Vector Intrinsics (llama/14317)

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

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

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

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

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

vulkan: use timestamp queries for GGML_VULKAN_PERF (llama/13817)

ggml : riscv: add xtheadvector support (llama/13720)

sycl: use oneDNN for matrices multiplication (llama/12972)

whisper: remove MSVC warnings pragmas (#3090)

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

ggml : suppress Windows compiler warnings (#3075)

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

CUDA/HIP: Share the same unified memory allocation logic. (llama/12934)

ggml : sync/merge cmake,riscv,powerpc, add common.cmake (ggml/0)

SYCL: using graphs is configurable by environment variable and compile option (llama/12371)

opencl: use OpenCL C standard supported by the device (llama/12221)

cmake: Enable specifying exact PowerPC CPU architecture (ggml/1138)

ggml : add logging for native build options/vars (#2935)

ggml-cpu: Faster IQ1 mul_mat_vec on AVX2 using BMI2 instructions (llama/12154)

HIP: implement FlashAttention via rocWMMA for CDNA and RDNA3+ (llama/12032)

CUDA: compress mode option and default to size (llama/12029)

cmake: Fix ggml backend dependencies and installation (llama/11818)

Told cmake to install ggml-cpp.h as a public header file. (ggml/1126)

ggml-cpu: Support s390x SIMD Instruction Set (llama/12019)

CUDA: app option to compile without FlashAttention (llama/12025)

ggml-cpu: Add CPU backend support for KleidiAI library (llama/11390)

cmake: Add ability to pass in GGML_BUILD_NUMBER (ggml/1096)

cmake: add ggml find package (llama/11369)

Hip: disable VMM on hip as it seams that it dosent work in some configurations (llama/11420)

hip : Add hipGraph and VMM support to ROCM (llama/11362)

cmake : avoid -march=native when reproducible build is wanted (llama/11366)

fix: ggml: fix vulkan-shaders-gen build (llama/10448)

GGUF: C++ refactor, backend support, misc fixes (llama/11030)

ggml : do not install metal source when embed library (ggml/1054)

ggml : fix arm build (llama/10890)

Introducing experimental OpenCL backend with support for Qualcomm Adreno GPUs (llama/10693)

remove CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS (llama/10797)

ggml : add predefined list of CPU backend variants to build (llama/10626)