| 2026-08-14 |
Titaniumtown | sycl: fuse the gated-delta-net state writeback cpy... |
commit | commitdiff | tree |
| 2026-08-14 |
Zijun Yu | OpenVINO: Qwen3.5, memory optimization, and test-recurr... |
commit | commitdiff | tree |
| 2026-08-14 |
Neo Zhang | Support host pinned mem to improve SYCL Host-to-Device... |
commit | commitdiff | tree |
| 2026-08-14 |
Georgi Gerganov | metal : add TQ2_0 support (llama/26980) |
commit | commitdiff | tree |
| 2026-08-14 |
jinzihao | ggml-cpu/ops: vectorize flash-attention V-cache F16... |
commit | commitdiff | tree |
| 2026-08-14 |
Ian Faust | sycl: remove separate fp32 type promotion in gemm non... |
commit | commitdiff | tree |
| 2026-08-14 |
Titaniumtown | sycl: fuse UNARY(silu|sigmoid|softplus) + MUL (llama... |
commit | commitdiff | tree |
| 2026-08-14 |
Todd Malsbary | sycl : Add DMMV ESIMD Q3_K kernel (llama/26251) |
commit | commitdiff | tree |
| 2026-08-14 |
Neo Zhang | sycl : enhance concat to support Q4_0, Q4_1, Q5_0,... |
commit | commitdiff | tree |
| 2026-08-14 |
Jim Wu | ggml-hip : remove -funsafe-math-optimizations (llama... |
commit | commitdiff | tree |
| 2026-08-14 |
Georgi Gerganov | ggml : fix arm builds, unused var (llama/26991) |
commit | commitdiff | tree |
| 2026-08-14 |
HarrisonSec | gguf : harden loader against malformed tensor dims... |
commit | commitdiff | tree |
| 2026-08-14 |
Jonathan Clohessy | kleidiai: Add runtime feature detection mechanism for... |
commit | commitdiff | tree |
| 2026-08-14 |
Hongqiang Wang | opencl: default FA c8 cluster width to 16 on X1E (llama... |
commit | commitdiff | tree |
| 2026-08-14 |
michaeltrabalka... | vulkan: add TQ2_0 (ternary) support (llama/25850) |
commit | commitdiff | tree |
| 2026-08-14 |
lhez | opencl: use flat mv q5_k when weight exceeds image1d_bu... |
commit | commitdiff | tree |
| 2026-08-14 |
Rafail Giavrimis | CUDA: only disable CUDA graphs when mul_mat_id actually... |
commit | commitdiff | tree |
| 2026-08-14 |
0 | cuda : add warp-per-row wkv7 kernel for single-token... |
commit | commitdiff | tree |
| 2026-08-14 |
Ruben Ortlam | llama: add default load-mode auto, which avoids mmap... |
commit | commitdiff | tree |
| 2026-08-14 |
Masashi Yoshimura | ggml-webgpu: fix CI errors from #25025 and #25262 ... |
commit | commitdiff | tree |
| 2026-08-14 |
Hongqiang Wang | opencl: transpose the K tile in local memory for FA... |
commit | commitdiff | tree |
| 2026-08-14 |
Hitesh Chopra | ggml-cpu : fix CPU affinity mask being ignored on Andro... |
commit | commitdiff | tree |
| 2026-08-14 |
Yash Raj Pandey | ggml : require contiguous src for ROLL on CUDA and... |
commit | commitdiff | tree |
| 2026-08-14 |
Masashi Yoshimura | ggml-webgpu : refactor several wgsl files and simplify... |
commit | commitdiff | tree |
| 2026-08-14 |
Hao-Chen2337 | ggml-cpu : fix missing Q5_0 dispatch in SpaceMiT backen... |
commit | commitdiff | tree |
| 2026-08-14 |
Rafail Giavrimis | CUDA: fuse rms_norm + mul + rope (+ view + set_rows... |
commit | commitdiff | tree |
| 2026-08-14 |
Rafail Giavrimis | CUDA: fix thread/block count in quantized cpy kernel... |
commit | commitdiff | tree |
| 2026-08-14 |
Titaniumtown | sycl: coalesce the ssm_conv window loads (llama/26612) |
commit | commitdiff | tree |
| 2026-08-14 |
robertomeroni | metal : fix NORM/RMS_NORM for row lengths that leave... |
commit | commitdiff | tree |
| 2026-08-14 |
Daniel Bevenius | cmake : add config version support (ggml/1582) |
commit | commitdiff | tree |
| 2026-08-07 |
Georgi Gerganov | talk-llama : sync llama.cpp |
commit | commitdiff | tree |
| 2026-08-07 |
Georgi Gerganov | sync : ggml |
commit | commitdiff | tree |
| 2026-08-07 |
Georgi Gerganov | ggml : bump version to 0.19.0 (ggml/1581) |
commit | commitdiff | tree |
| 2026-08-07 |
Kilian Hu | ggml : add aarch64 HWCAP fallbacks and fix fp16 variant... |
commit | commitdiff | tree |
| 2026-08-07 |
Chris Lee | sycl: fix UE4M3 parsing (llama/25608) |
commit | commitdiff | tree |
| 2026-08-07 |
Titaniumtown | sycl: *glu flat path (llama/26354) |
commit | commitdiff | tree |
| 2026-08-07 |
Neo Zhang | sycl : Support DSv4 OPs: LIGHTNING_INDEXER,DSV4_HC_COMB... |
commit | commitdiff | tree |
| 2026-08-07 |
Neo Zhang | sycl : fix error Error OP FLASH_ATTN_EXT on arc770... |
commit | commitdiff | tree |
| 2026-08-07 |
Neo Zhang | sycl : enhance OP set_rows to support all missed data... |
commit | commitdiff | tree |
| 2026-08-07 |
David Friehs | cuda: fix warnings for unused variable/function (llama... |
commit | commitdiff | tree |
| 2026-08-07 |
JamePeng | metal : avoid `threadgroup` matrix array instantiation... |
commit | commitdiff | tree |
| 2026-08-07 |
Jim Wu | ci : onboard AMD ROCm CI with gfx1151 fixes (llama... |
commit | commitdiff | tree |
| 2026-08-07 |
Ruben Ortlam | vulkan: fix submission batching size, add debug tools... |
commit | commitdiff | tree |
| 2026-08-07 |
Pascal | mtmd/ggml: add ggml_build_forward_order (llama/26649) |
commit | commitdiff | tree |
| 2026-08-07 |
Pranav Uttarkar | vulkan backend ops: implemented GATED_LINEAR_ATTN ... |
commit | commitdiff | tree |
| 2026-08-07 |
Md Shafiuzzaman | whisper : heap out-of-bounds read in log_mel_spectrogra... |
commit | commitdiff | tree |
| 2026-08-07 |
Md Shafiuzzaman | whisper,parakeet : reject invalid n_dims in tensor... |
commit | commitdiff | tree |
| 2026-08-04 |
Daniel Bevenius | release : v1.9.2 (#3970) upstream/1.9.2 |
commit | commitdiff | tree |
| 2026-08-04 |
Georgi Gerganov | talk-llama : sync llama.cpp |
commit | commitdiff | tree |
| 2026-08-04 |
Georgi Gerganov | sync : ggml |
commit | commitdiff | tree |
| 2026-08-04 |
Georgi Gerganov | ggml : bump version to 0.18.1 (ggml/1578) |
commit | commitdiff | tree |
| 2026-08-04 |
Titaniumtown | sycl: parallelize the non-contiguous concat kernel... |
commit | commitdiff | tree |
| 2026-08-04 |
Ozymandias_EBON | Extended SYCL oneDNN SDPA to non-FP16 KV caches (Q4_0... |
commit | commitdiff | tree |
| 2026-08-04 |
AgoraPete | ggml: use dynamic allocation for split graph inputs... |
commit | commitdiff | tree |
| 2026-08-04 |
Hongqiang Wang | opencl: route large q6_K lm_head to the flat GEMV ... |
commit | commitdiff | tree |
| 2026-08-04 |
Oliver Simons | CUDA: Fix data-races when reusing SMEM in block_reduce... |
commit | commitdiff | tree |
| 2026-08-04 |
Thiago Padilha | metal: implement DSv4 Lightning Indexer (llama/25893) |
commit | commitdiff | tree |
| 2026-08-04 |
Talha Adnan | metal : add SILU_BACK (llama/25982) |
commit | commitdiff | tree |
| 2026-08-04 |
Georgi Gerganov | metal : add F16 support for bin ops (llama/26465) |
commit | commitdiff | tree |
| 2026-08-04 |
mgroeber9110 | opencl: limit local workgroup size for GLU operation... |
commit | commitdiff | tree |
| 2026-08-04 |
Georgi Gerganov | metal: implement DeepSeek V4 hyper-connections (llama... |
commit | commitdiff | tree |
| 2026-08-04 |
akleine | opencl: bugfix increment ref_count in ggml_backend_open... |
commit | commitdiff | tree |
| 2026-08-04 |
KyleHagy | sycl: fix classification of iGPUs (llama/26105) |
commit | commitdiff | tree |
| 2026-08-04 |
Masashi Yoshimura | ggml-webgpu: add support for f16 repeat (llama/26307) |
commit | commitdiff | tree |
| 2026-08-04 |
Jeff Bolz | vulkan: extend topk_moe fusion to support sqrt(softplus... |
commit | commitdiff | tree |
| 2026-08-04 |
Anand Patil | vulkan: add POOL_1D op (llama/25431) |
commit | commitdiff | tree |
| 2026-08-04 |
Masato Nakasaka | vulkan: Introduce driver version check for Windows... |
commit | commitdiff | tree |
| 2026-08-04 |
David Friehs | cuda: extract Q2_0 elements via __byte_perm (llama... |
commit | commitdiff | tree |
| 2026-08-04 |
Ozymandias_EBON | SYCL: add oneMKL GEMM flash attention for XMX-accelerat... |
commit | commitdiff | tree |
| 2026-08-04 |
Neo Zhang | support the missed types in cpy (llama/26005) |
commit | commitdiff | tree |
| 2026-08-04 |
Sachin Sharma | ggml-zendnn : group matmul direct API for mul_mat_id... |
commit | commitdiff | tree |
| 2026-08-04 |
Neo Zhang | sycl : support dev2dev memcpy by DEV2DEV_MEMCPY_FORWARD... |
commit | commitdiff | tree |
| 2026-08-04 |
Neo Zhang | Support q2 mul_mat (llama/26231) |
commit | commitdiff | tree |
| 2026-08-04 |
Titaniumtown | sycl: fuse RMS_NORM + MUL (llama/26015) |
commit | commitdiff | tree |
| 2026-08-04 |
Masashi Yoshimura | ggml-webgpu: improve flash_attn_vec for quantized KV... |
commit | commitdiff | tree |
| 2026-08-04 |
Jeff Bolz | vulkan: Support quantized concat (llama/25684) |
commit | commitdiff | tree |
| 2026-08-04 |
pmaybank | Test support for alternative conv layout (llama/25617) |
commit | commitdiff | tree |
| 2026-08-04 |
Robert Esclapez | ggml-cuda: Allow transpose-free gemmv computation ... |
commit | commitdiff | tree |
| 2026-08-03 |
Daniel Bevenius | ci : set GGML_NATIVE=OFF for build-vad workflow (#3966) |
commit | commitdiff | tree |
| 2026-08-03 |
Igor Koshenskii | examples : fix VAD min silence argument parsing (#3963) |
commit | commitdiff | tree |
| 2026-07-31 |
Georgi Gerganov | sync : ggml (#3962) |
commit | commitdiff | tree |
| 2026-07-30 |
Bhargav Krish | parakeet : verify hparams loaded from parakeet model... |
commit | commitdiff | tree |
| 2026-07-29 |
Todd | Adding ruby bindings change log (#3959) |
commit | commitdiff | tree |
| 2026-07-28 |
KITAITI Makoto | ruby : add VAD speech segments API (#3931) |
commit | commitdiff | tree |
| 2026-07-11 |
Cappuccino | examples : Remove leading space from txt output (#3921) |
commit | commitdiff | tree |
| 2026-07-10 |
Georgi Gerganov | ggml : use ggml_vqtbl1q_u8 for 32-bit compat (#0) |
commit | commitdiff | tree |
| 2026-07-10 |
Georgi Gerganov | talk-llama : sync llama.cpp |
commit | commitdiff | tree |
| 2026-07-10 |
Georgi Gerganov | sync : ggml |
commit | commitdiff | tree |
| 2026-07-10 |
Georgi Gerganov | ggml : bump version to 0.16.0 (ggml/1559) |
commit | commitdiff | tree |
| 2026-07-10 |
Georgi Gerganov | ggml : fix conv 2d dw (llama/25490) |
commit | commitdiff | tree |
| 2026-07-10 |
Georgi Gerganov | common : adapt to q2_0 (ggml/0) |
commit | commitdiff | tree |
| 2026-07-10 |
Oliver Simons | Only index by compile times + always multiply/add ... |
commit | commitdiff | tree |
| 2026-07-10 |
Sou-ly | metal : add CONV_2D_DW (depthwise convolution) support... |
commit | commitdiff | tree |
| 2026-07-10 |
RapidMark | ggml-hip: enable -funsafe-math-optimizations (llama... |
commit | commitdiff | tree |
| 2026-07-10 |
Pascal | cuda: align snake fusion matcher with the other backend... |
commit | commitdiff | tree |
| 2026-07-10 |
Aparna M P | hexagon: add VISION RoPE support (llama/25216) |
commit | commitdiff | tree |
| 2026-07-10 |
Masashi Yoshimura | ggml-webgpu: tune subgroup split (d_split) in flash_att... |
commit | commitdiff | tree |
| 2026-07-10 |
Hongqiang Wang | opencl: Q6_K GEMM/GEMV fix for ne01 of weights that... |
commit | commitdiff | tree |
| 2026-07-10 |
Ruben Ortlam | vulkan: disable FA mask_opt on GCN to improve performan... |
commit | commitdiff | tree |
| 2026-07-10 |
Hongqiang Wang | opencl: ragged-tile MoE prefill FP16 GEMM optimization... |
commit | commitdiff | tree |
| next |