ggml-backend : only offload from host buffers (llama/11120)

rpc : code cleanup (llama/11107)

SYCL: Use get_multi_ptr instead of deprecated get_pointer in wkv6 (llama/11087)

CUDA: add BF16 support (llama/11093)

Vulkan: Add device-specific blacklist for coopmat for the AMD proprietary driver (llama/11074)

Support for models with non-512-aligned tensors over RPC. (llama/11047)

fix: Vulkan shader gen binary path (llama/11037)

ggml : allow loading backend with env variable (ggml/1059)

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

metal : avoid uint (llama/11019)

ggml : fixes for AVXVNNI instruction set with MSVC and Clang (llama/11027)

vulkan: optimize mul_mat for small values of N (llama/10991)

vulkan: im2col and matmul optimizations for stable diffusion (llama/10942)

vulkan: Use push constant offset to handle misaligned descriptors (llama/10987)

vulkan: multi-row k quants (llama/10846)

examples, ggml : fix GCC compiler warnings (llama/10983)

ggml : more perfo with llamafile tinyblas on x86_64 (llama/10714)

ggml : use wstring for backend search paths (llama/10960)

ggml : fix arm enabled features check (llama/10961)

ggml : fix const usage in SSE path (llama/10962)

ggml : fix run-time on FreeBSD in get_executable_path() (llama/10948)

vulkan: build fixes for 32b (llama/10927)

vulkan: optimize coopmat2 dequant functions (llama/10855)

ggml-cpu: replace NEON asm with intrinsics in ggml_gemv_q4_0_4x8_q8_0() (llama/10874)

SYCL: Migrate away from deprecated ggml_tensor->backend (llama/10840)

ggml : add test for SVE and disable when it fails (llama/10906)

ggml: fix arm build with gcc (llama/10895)

ggml : fix arm build (llama/10890)

tts : add OuteTTS support (llama/10784)

tests: add tests for GGUF (llama/10830)

ggml : improve inputs log sched_print_assignments (ggml/1053)

files : remove old sources

ggml : update ggml_backend_cpu_device_supports_op (llama/10867)

vulkan: bugfixes for small subgroup size systems + llvmpipe test (llama/10809)

rwkv6: add wkv6 support for Vulkan backend (llama/10829)

llama : add Qwen2VL support + multimodal RoPE (llama/10361)

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

Fix crash caused by ggml_backend_load_all when launching on Android Activity (llama/10812)

vulkan: small mul_mat_vec optimizations (llama/10665)

SYCL: Reduce most of the compiler warnings (llama/10748)

ggml : Fix compilation issues on ARM platform when building without fp16 (llama/10811)

CUDA: faster non-contiguous concat (llama/10760)

remove CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS (llama/10797)

Vulkan: Use improved q4_k and q5_k dequant code in dequant shaders (llama/10798)

Vulkan: Add VK_EXT_subgroup_size_control support to ensure full subgroups for coopmats (llama/10721)

ggml: load all backends from a user-provided search path (llama/10699)

vulkan: request round-to-even for fp16 in im2col/rope_head (llama/10767)

vulkan: dynamic subgroup size for the remaining k quants (llama/10745)

CUDA: rename macros to avoid conflicts with WinAPI (llama/10736)

vulkan: disable spirv-opt for coopmat shaders (llama/10763)