]> git.djapps.eu Git - pkg/ggml/sources/ggml/shortlog
pkg/ggml/sources/ggml
2025-05-25 Georgi Gerganovsync : llama.cpp upstream/latest
2025-05-25 Akarshan BiswasSYCL: revert "sycl: simplify bin_bcast_kernel (#13383...
2025-05-25 Diego Devesaggml-cpu : set openmp wait time if not set (llama/13758)
2025-05-25 Georgi Gerganovsync : llama.cpp
2025-05-25 Xuan-Son Nguyenggml : add ggml_gelu_erf() CUDA kernel (llama/13719)
2025-05-25 Johannes GäßlerCUDA: fix race condition in FA vector kernels (llama...
2025-05-25 Chenguang LiCANN: Support MUL_MAT_ID for q8_0 and q4_0 (llama/13705)
2025-05-25 Xuan-Son Nguyenggml : fix the order of ggml_unary_op (llama/13718)
2025-05-25 Jeff Bolzvulkan: support CPY from any type to itself (llama...
2025-05-25 Jeff Bolzvulkan: Disable coopmat/coopmat2/bfloat extensions...
2025-05-25 Judduse LOG_WARN to replace `std::cerr` (llama/13657)
2025-05-25 Nicolò Scipionesycl : Remove waits from function calls (llama/13702)
2025-05-25 Ewan CrawfordSYCL: Avoid using with SYCL-Graph for unsupported nodes...
2025-05-25 Henry Linjamäkiopencl: Add support for multiple devices (llama/12622)
2025-05-25 Henry Linjamäkiopencl: fix couple crashes (llama/12795)
2025-05-25 Xuan-Son Nguyenggml : add ggml_gelu_erf() (llama/13667)
2025-05-25 R0CKSTARmusa: Upgrade MUSA SDK version to rc4.0.1 and use mudnn...
2025-05-25 Evevulkan: fix warnings (llama/13626)
2025-05-25 Johannes GäßlerCUDA: skip fully masked-out KV in FA vec kernel (llama...
2025-05-25 Svetlozar Georgievsycl: disable reorder for sycl mulmat (llama/13536)
2025-05-25 Georgi Gerganovmetal : fix typo in FA kernel comments (llama/13651)
2025-05-25 Nicolò Scipionesycl : Overcoming workaround for mmap() allocation...
2025-05-25 0cc4mVulkan: Add f32 accumulator support to quantized mul...
2025-05-19 Georgi Gerganovsync : llama.cpp
2025-05-19 Chenguang LiCANN: Support MOE Model MUL_MAT_ID (llama/13042)
2025-05-19 Gilad S.cmake: use the current build config for vulkan-shaders...
2025-05-19 Jeff Bolzvulkan: move common FA code to flash_attn_base.comp...
2025-05-19 Jeff Bolzvulkan: use scalar FA rather than coopmat2 when N=...
2025-05-19 Georgi Gerganovmetal : add FA-vec kernel for head size 64 (llama/13583)
2025-05-19 Łukasz Ślusarczyksycl : fixed compilation warnings (llama/13582)
2025-05-19 Diego Devesagguf : use ggml log system (llama/13571)
2025-05-19 Atharva Dubeysycl: simplify bin_bcast_kernel (llama/13383)
2025-05-19 Svetlozar Georgievsycl: reordered Q4_K MMVQ (llama/13109)
2025-05-19 Łukasz Ślusarczyksycl: use oneDNN for matrices multiplication (llama...
2025-05-19 Yibo Caiarm64: optimize q6_k_q8_k kernel with i8mm (llama/13519)
2025-05-19 Johannes GäßlerCUDA: fix crash on large batch size for quant. MoE...
2025-05-19 Johannes GäßlerCUDA: faster Deepseek FA, add Turing support (llama...
2025-05-19 bandoticmake: simplify vulkan shader test logic (llama/13263)
2025-05-19 Jeff Bolzvulkan: KHR_coopmat flash attention (llama/13506)
2025-05-19 Jeff Bolzvulkan: workaround FA compile failures on macos (llama...
2025-05-19 Georgi Gerganovmetal : use FA-vec kernel up to batch size 20 (llama...
2025-05-19 Georgi Gerganovmetal : optimize multi-sequence FA vec kernel (llama...
2025-05-19 Dan Johanssonggml-cpu: Update KleidiAI to v1.6 and fix include direc...
2025-05-19 Johannes Gäßlermnist: fix segmentation fault (#1227)
2025-05-19 Diego Devesaggml : fix apple OS check in ggml_print_backtrace ...
2025-05-17 Daniel Tangggml : Fix missing backtrace on Linux (#1228)
2025-05-13 Georgi Gerganovsync : whisper.cpp
2025-05-13 Daniel Beveniusexamples : update link to Paul Tol's color scheme ...
2025-05-13 Georgi Gerganovexamples : update to ggml-opt and ggml-backend changes...
2025-05-13 Georgi Gerganovsync : llama.cpp
2025-05-13 lhezopencl: remove unnecessary assert for `add` (llama...
2025-05-13 Johannes Gäßlerllama/ggml: add LLM training support (llama/10544)
2025-05-13 Dan Johanssonggml-cpu: Integrate fp32=bf16xbf16 SME KleidiAI kernel...
2025-05-13 Johannes GäßlerCUDA: fix misaligned synchronization in FA (llama/13469)
2025-05-13 Xuan-Son Nguyenggml : add mrope kernel for metal (llama/13457)
2025-05-13 Atharva Dubeyenable dpcpp nightly builds with libraries (llama/13406)
2025-05-13 Johannes GäßlerCUDA: fix crash with partial offloading of MoE (llama...
2025-05-13 David HuangAdd `--no-op-offload` to improve `-ot` pp perf in MoE...
2025-05-13 Johannes GäßlerCUDA: fix race conditions FlashAttention kernels (llama...
2025-05-13 Johannes GäßlerCUDA: fix FlashAttention on Turing (llama/13415)
2025-05-13 Jeff Bolzvulkan: scalar flash attention implementation (llama...
2025-05-13 Alberto Cabrera... sycl : implementation of reordered Q4_0 MMVQ for Intel...
2025-05-13 Georgi Gerganovmetal : optimize MoE for large batches (llama/13388)
2025-05-13 Johannes GäßlerCUDA: FA support for Deepseek (Ampere or newer) (llama...
2025-05-13 Johannes GäßlerCUDA: fix crash on large batch size for MoE models...
2025-05-13 Radoslav Gerganovrpc : add rpc_msg_set_tensor_hash_req (llama/13353)
2025-05-13 Jeff Bolzvulkan: Allow up to 4096 elements for mul_mat_id row_id...
2025-05-13 Alberto Cabrera... sycl: addressing non-contiguous src1 mul_mats (nc and...
2025-05-08 Taylorsam : support box prompt (#1206)
2025-05-07 Georgi Gerganovsync : llama.cpp
2025-05-07 R0CKSTARcuda : remove nrows_x in mul_mat_q_process_tile (llama...
2025-05-07 Johannes GäßlerCUDA: mix virt/real CUDA archs for GGML_NATIVE=OFF...
2025-05-07 Akarshan BiswasSYCL: Disable reorder optimize by default and stop...
2025-05-07 Johannes GäßlerCUDA: fix bad asserts for partial offload (llama/13337)
2025-05-07 Johannes GäßlerCUDA: fix --split-mode row for MMQ (llama/13323)
2025-05-07 Johannes GäßlerCUDA: fix logic for clearing padding with -ngl 0 (llama...
2025-05-07 Akarshan BiswasSYCL: Disable mul_mat kernels for noncontiguous tensor...
2025-05-07 Diego Devesarpc : use backend registry, support dl backends (llama...
2025-05-07 Aaron Teoggml : activate s390x simd for Q3_K (llama/13301)
2025-05-07 Johannes GäßlerCUDA: fix race condition in MMQ stream-k fixup (llama...
2025-05-07 Johannes GäßlerCUDA: fix race condition in MMQ ids_dst (llama/13294)
2025-05-07 Jeff Bolzvulkan: Additional type support for unary, binary,...
2025-05-07 Georgi Gerganovsync : whisper.cpp
2025-05-07 Daniel Beveniuswhisper: remove MSVC warnings pragmas (whisper/3090)
2025-05-07 Jared Tweedcmake : removed stdc++fs (whisper/3097)
2025-05-02 Georgi Gerganovsync : llama.cpp upstream/0.0.2015
2025-05-02 Georgi Gerganovvulkan : fix lint (llama/0)
2025-05-02 shalinib-ibmggml : Enable MMA for BF16 in llamafile_sgemm (llama...
2025-05-02 Justin Santa... rpc : avoid uninitialized memory in serialize_tensor...
2025-05-02 Jesse Grossggml: Don't assert fail when tensor data changes (llama...
2025-05-02 Diego Devesabuild : fix build info on windows (llama/13239)
2025-05-02 Jeff Bolzvulkan: Add bfloat16 support (llama/12554)
2025-05-02 Jeff Bolzvulkan: Handle src1 batch dimension in non-contiguous...
2025-05-02 Johannes Gäßlertest: non-cont. b in test-backend-ops -o MUL_MAT (llama...
2025-05-02 Aclyvulkan : kernels for depthwise 2D convolution (CONV_2D_...
2025-05-01 Georgi Gerganovsync : whisper.cpp
2025-05-01 Daniel Beveniuswhisper : add check that target name exists (whisper...
2025-05-01 Daniel Beveniusggml : suppress Windows compiler warnings (whisper...
2025-05-01 Georgi Gerganovsync : llama.cpp
2025-05-01 Johannes GäßlerCUDA: batched+noncont MMQ, refactor bs>1 MoE code ...
next