| 2026-02-27 |
Jeff Bolz | vulkan: check for memory overlap before doing fusion... |
commit | commitdiff | tree |
| 2026-02-27 |
Georgi Gerganov | ggml/gguf : prevent integer overflows (llama/19856) |
commit | commitdiff | tree |
| 2026-02-27 |
Ruben Ortlam | Vulkan Scalar Flash Attention Refactor (llama/19625) |
commit | commitdiff | tree |
| 2026-02-27 |
Jeff Bolz | vulkan: fix coopmat1 without bf16 support (llama/19793) |
commit | commitdiff | tree |
| 2026-02-27 |
Jeff Bolz | vulkan: fix data race in mul_mat_id shader (llama/19790) |
commit | commitdiff | tree |
| 2026-02-27 |
Max Krasnyansky | hexagon refactor all Ops to use local context struct... |
commit | commitdiff | tree |
| 2026-02-27 |
Alberto Cabrera... | ggml-cpu: arm64: q5_K repack gemm and gemv (and generic... |
commit | commitdiff | tree |
| 2026-02-27 |
Gaurav Garg | Improve CUDA graph capture (llama/19754) |
commit | commitdiff | tree |
| 2026-02-27 |
Taimur Ahmad | ggml-cpu: add RVV vec dot kernels for quantization... |
commit | commitdiff | tree |
| 2026-02-27 |
Masashi Yoshimura | ggml-webgpu: Add unary op (SQR, SQRT, SIN, COS) support... |
commit | commitdiff | tree |
| 2026-02-27 |
Ruben Ortlam | vulkan: fix MMQ shader push constants and multi-dispatc... |
commit | commitdiff | tree |
| 2026-02-27 |
Johannes Gäßler | CUDA: fix kernel selection logic for tile FA (llama... |
commit | commitdiff | tree |
| 2026-02-27 |
shalinib-ibm | llamafile: powerpc: add FP16 MMA path for Q4/Q8 matmul... |
commit | commitdiff | tree |
| 2026-02-27 |
Reese Levine | ggml webgpu: Fix bug in dispatching large matrix-vector... |
commit | commitdiff | tree |
| 2026-02-27 |
Reese Levine | ggml webgpu: shader library organization (llama/19530) |
commit | commitdiff | tree |
| 2026-02-27 |
Jeff Bolz | vulkan: split mul_mat into multiple dispatches to avoid... |
commit | commitdiff | tree |
| 2026-02-27 |
shaofeiqi | opencl: refactor expm1 and softplus (llama/19404) |
commit | commitdiff | tree |
| 2026-02-27 |
shaofeiqi | opencl: optimize mean and sum_row kernels (llama/19614) |
commit | commitdiff | tree |
| 2026-02-27 |
Talha Can Havadar | ggml: ggml-cpu: force-no-lto-for-cpu-feats (llama/19609) |
commit | commitdiff | tree |
| 2026-02-27 |
Georgi Gerganov | cuda : enable CUDA graphs for MMID 1 <= BS <= 4 (llama... |
commit | commitdiff | tree |
| 2026-02-27 |
Judd | ggml : make `ggml_is_view` as API (llama/19539) |
commit | commitdiff | tree |
| 2026-02-27 |
Mario Limonciello | Adjust workaround for ROCWMMA_FATTN/GFX9 to only newer... |
commit | commitdiff | tree |
| 2026-02-27 |
abhijain1204fujitsu | ggml: aarch64: Implement SVE in Gemm q4_k 8x8 q8_k... |
commit | commitdiff | tree |
| 2026-02-27 |
David Friehs | cuda: optimize iq2xxs/iq2xs/iq3xxs dequantization ... |
commit | commitdiff | tree |
| 2026-02-27 |
Daniel Bevenius | cmake : check if KleidiAI API has been fetched (llama... |
commit | commitdiff | tree |
| 2026-02-27 |
Georgi Gerganov | ggml : avoid UB in gemm ukernel (llama/19642) |
commit | commitdiff | tree |
| 2026-02-27 |
Aaron Teo | ggml-cpu: optimize ggml_vec_dot_bf16 for s390x (llama... |
commit | commitdiff | tree |
| 2026-02-27 |
Aman Gupta | ggml-cpu: FA add GEMM microkernel (llama/19422) |
commit | commitdiff | tree |
| 2026-02-27 |
SamareshSingh | cmake : fix KleidiAI install target failure with EXCLUD... |
commit | commitdiff | tree |
| 2026-02-27 |
Georgi Gerganov | ggml : bump version to 0.9.7 (ggml/1425) |
commit | commitdiff | tree |
| 2026-02-27 |
Dmitry Atamanov | examples : update miniaudio library to 0.11.24 (#3672) |
commit | commitdiff | tree |
| 2026-02-19 |
Maxime Grenu | docs : fix duplicate word typo in VAD section (#3670) |
commit | commitdiff | tree |
| 2026-02-15 |
Georgi Gerganov | talk-llama : sync llama.cpp upstream/1.8.3+155 |
commit | commitdiff | tree |
| 2026-02-15 |
Georgi Gerganov | sync : ggml |
commit | commitdiff | tree |
| 2026-02-15 |
Georgi Gerganov | models : optimize qwen3next graph (llama/19375) |
commit | commitdiff | tree |
| 2026-02-15 |
Adrien Gallouët | ggml : fix GGML_DEBUG with OpenMP (llama/19599) |
commit | commitdiff | tree |
| 2026-02-15 |
Georgi Gerganov | metal : fix ACC op (llama/19427) |
commit | commitdiff | tree |
| 2026-02-15 |
Jeff Bolz | vulkan: support L2_NORM with contiguous rows (llama... |
commit | commitdiff | tree |
| 2026-02-15 |
Jeff Bolz | vulkan: support GGML_OP_SET (llama/19584) |
commit | commitdiff | tree |
| 2026-02-15 |
Sophon | vulkan: Add vendor id for Qualcomm drivers (llama/19569) |
commit | commitdiff | tree |
| 2026-02-15 |
Max Krasnyansky | hexagon: further optimizations and refactoring for... |
commit | commitdiff | tree |
| 2026-02-15 |
Jeff Bolz | vulkan: restore -inf check in FA shaders (llama/19582) |
commit | commitdiff | tree |
| 2026-02-15 |
Alberto Cabrera... | Fix wrong memcpy length for block_interleave == 4 ... |
commit | commitdiff | tree |
| 2026-02-15 |
ymcki | fix vulkan ggml_acc only works in 3d but not 4d (llama... |
commit | commitdiff | tree |
| 2026-02-15 |
Aman Gupta | CUDA: loop over ne2*ne3 in case it overflows (llama... |
commit | commitdiff | tree |
| 2026-02-15 |
Oliver Simons | CUDA: Do not mutate cgraph for fused ADDs (llama/19566) |
commit | commitdiff | tree |
| 2026-02-15 |
Georgi Gerganov | metal : improve concurrency (llama/19555) |
commit | commitdiff | tree |
| 2026-02-15 |
Georgi Gerganov | metal : support GGML_OP_SET (llama/19548) |
commit | commitdiff | tree |
| 2026-02-15 |
Shupei Fan | hexagon: fix typo in vtcm_needs_release (llama/19545) |
commit | commitdiff | tree |
| 2026-02-15 |
lhez | opencl: add basic support for q4_1 (llama/19534) |
commit | commitdiff | tree |
| 2026-02-15 |
Georgi Gerganov | metal : update sum_rows kernel to support float4 (llama... |
commit | commitdiff | tree |
| 2026-02-15 |
Mario Limonciello | Add a workaround for compilation with ROCWMMA_FATTN... |
commit | commitdiff | tree |
| 2026-02-15 |
Max Krasnyansky | hexagon: further optimization and tuning of matmul... |
commit | commitdiff | tree |
| 2026-02-15 |
lhez | opencl: add general Q6_K mm and Q4_K mv (llama/19347) |
commit | commitdiff | tree |
| 2026-02-15 |
Georgi Gerganov | ggml : unary ops support non-cont src0 + metal F16... |
commit | commitdiff | tree |
| 2026-02-15 |
Georgi Gerganov | metal : extend l2_norm support for non-cont src0 (llama... |
commit | commitdiff | tree |
| 2026-02-15 |
Max Krasnyansky | hexagon: Add ARGSORT, DIV, SQR, SQRT, SUM_ROWS, GEGLU... |
commit | commitdiff | tree |
| 2026-02-15 |
Georgi Gerganov | ggml : extend bin bcast for permuted src1 (llama/19484) |
commit | commitdiff | tree |
| 2026-02-15 |
Georgi Gerganov | metal : consolidate unary ops (llama/19490) |
commit | commitdiff | tree |
| 2026-02-15 |
Oliver Simons | CUDA : Update CCCL-tag for 3.2 to final release from... |
commit | commitdiff | tree |
| 2026-02-15 |
Nikhil Jain | Plug memory leaks and free resources on shutdown (llama... |
commit | commitdiff | tree |
| 2026-02-15 |
Alberto Cabrera... | ggml-cpu: arm64: q6_K repack gemm and gemv (and generic... |
commit | commitdiff | tree |
| 2026-02-15 |
k4ss4n | ggml : use noexcept overload for is_regular_file in... |
commit | commitdiff | tree |
| 2026-02-15 |
Raul Torres | CANN: Remove unnecessary wrapper for `gml_backend_buft_... |
commit | commitdiff | tree |
| 2026-02-15 |
hipudding | CANN: implement quantized MUL_MAT_ID for MoE models... |
commit | commitdiff | tree |
| 2026-02-15 |
Georgi Gerganov | cuda : extend GGML_OP_PAD to work with non-cont src0... |
commit | commitdiff | tree |
| 2026-02-15 |
Oliver Simons | CUDA: Fix non-contig rope (llama/19338) |
commit | commitdiff | tree |
| 2026-02-09 |
Nuno | ci: add vulkan docker image (#3644) |
commit | commitdiff | tree |
| 2026-02-09 |
Pádraic Slattery | chore: Update outdated GitHub Actions versions (#3646) |
commit | commitdiff | tree |
| 2026-02-09 |
Christian Kastner | cmake: Drop obsolete build-time configuration of backen... |
commit | commitdiff | tree |
| 2026-02-09 |
Sid Mohan | server : fix hardcoded /inference path in default HTML... |
commit | commitdiff | tree |
| 2026-02-09 |
Georgi Gerganov | ci : try fix mirrors (#3655) |
commit | commitdiff | tree |
| 2026-02-08 |
Georgi Gerganov | talk-llama : sync llama.cpp |
commit | commitdiff | tree |
| 2026-02-08 |
Georgi Gerganov | sync : ggml |
commit | commitdiff | tree |
| 2026-02-08 |
Georgi Gerganov | metal : consolidate bin kernels (llama/19390) |
commit | commitdiff | tree |
| 2026-02-08 |
Georgi Gerganov | metal : fix event synchronization in cpy_tensor_async... |
commit | commitdiff | tree |
| 2026-02-08 |
Abhijit Ramesh | ggml-webgpu: JIT compile binary operators and handle... |
commit | commitdiff | tree |
| 2026-02-08 |
Nechama Krashinski | sycl: add F16 support for GGML_OP_CEIL (llama/19306) |
commit | commitdiff | tree |
| 2026-02-08 |
Jeff Bolz | vulkan: For coopmat2 FA, use fp16 accumulators for... |
commit | commitdiff | tree |
| 2026-02-08 |
Jeff Bolz | vulkan: make FA mask/softcap enables spec constants... |
commit | commitdiff | tree |
| 2026-02-08 |
Georgi Gerganov | metal : skip loading all-zero mask (llama/19337) |
commit | commitdiff | tree |
| 2026-02-08 |
Georgi Gerganov | cuda : cuda graphs now compare all node params (llama... |
commit | commitdiff | tree |
| 2026-02-08 |
Georgi Gerganov | metal : adaptive CPU/GPU interleave based on number... |
commit | commitdiff | tree |
| 2026-02-08 |
Jeff Bolz | vulkan: Preprocess FA mask to detect all-neg-inf and... |
commit | commitdiff | tree |
| 2026-02-08 |
Georgi Gerganov | metal : add diag (llama/19330) |
commit | commitdiff | tree |
| 2026-02-08 |
Oleksandr Kuvshynov | vulkan: fix GPU deduplication logic. (llama/19222) |
commit | commitdiff | tree |
| 2026-02-08 |
Jeff Bolz | vulkan: Set k_load_shmem to false when K is too large... |
commit | commitdiff | tree |
| 2026-02-08 |
Jeff Bolz | vulkan: fix non-contig rope (llama/19299) |
commit | commitdiff | tree |
| 2026-02-08 |
will-lms | metal : add missing includes (llama/19348) |
commit | commitdiff | tree |
| 2026-02-08 |
Kevin Pouget | ggml-virtgpu: make the code thread safe (llama/19204) |
commit | commitdiff | tree |
| 2026-02-08 |
Aman Gupta | ggml-cpu: use LUT for converting e8->f32 scales on... |
commit | commitdiff | tree |
| 2026-02-08 |
Georgi Gerganov | metal : add solve_tri (llama/19302) |
commit | commitdiff | tree |
| 2026-02-08 |
Ruben Ortlam | vulkan: disable coopmat1 fa on Nvidia Turing (llama... |
commit | commitdiff | tree |
| 2026-02-08 |
Aman Gupta | CUDA: use mmvq for mul-mat-id for small batch sizes... |
commit | commitdiff | tree |
| 2026-02-08 |
Georgi Gerganov | metal : minor cleanup (llama/19251) |
commit | commitdiff | tree |
| 2026-02-08 |
Oliver Simons | CUDA: Fix loop unrolling for BW in mul_mat_q_stream_k_f... |
commit | commitdiff | tree |
| 2026-02-08 |
George | ggml: added cleanups in ggml_quantize_free (llama/19278) |
commit | commitdiff | tree |
| 2026-02-08 |
Gaurav Garg | cuda : revert CUDA_SCALE_LAUNCH_QUEUES override until... |
commit | commitdiff | tree |
| 2026-02-08 |
lhez | opencl: refactor some ops, concat, repeat, tanh and... |
commit | commitdiff | tree |
| 2026-02-08 |
Aman Gupta | ggml-cpu: FA split across kv for faster TG (llama/19209) |
commit | commitdiff | tree |
| next |
overview / pkg / ggml / sources / whisper.cpp / shortlog