llama.cpp 发布更新，修复性能和稳定性问题 · 跟踪 9 个来源

作者 PulseAugur 编辑部 · [125 个来源] · 2026-09-04 17:57

llama.cpp 项目发布了多个更新，包括 0.4.1 版本，该版本解决了跨不同平台的各种性能和稳定性问题。值得注意的更改包括对 SYCL 后端的优化、CPU 堆损坏的修复以及对缺乏 BF16 硬件加速的 CUDA 设备的后备机制。这些更新旨在提高在各种硬件上运行语言模型的效率和可靠性。 AI

影响这些更新提高了 llama.cpp 的性能和稳定性，可能提高了在各种硬件上运行本地 LLM 的效率。

排序理由该集群包含 llama.cpp 项目的多个发布说明，详细介绍了错误修复和小的改进，而不是新的模型发布或重大的研究突破。

在 llama.cpp — Releases 阅读 →

llama.cpp
Adreno
android
iOS
Linux
macOS
Metal
Microsoft Windows
OpenCL
Apple Silicon
Claude Fable 5.1
CUDA
KleidiAI
OpenEuler
Vulkan
0.4.1
#28900
AMD
b10964
CACHE_LINE_SIZE
ggml-cpu
ggml-cuda
GGML_OP_TOP_K
Nvidia
Qwen4Exp
std::hardware_destructive_interference_size

AI 生成摘要 · Google Gemini · 来自 125 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群包含 llama.cpp 项目的多个发布说明，详细介绍了错误修复和小的改进，而不是新的模型发布或重大的研究突破。
Source corroboration: 125 independent sources; Strong cross-source corroboration — multiple independent publishers covered this within the clustering window.
Topics: product, infra; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: 50 days old; Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.
Coverage growth since scoring: +22 source(s) since last score; New sources have picked up this story since our last re-score. Score will update on the next scoring pass.

完整方法见我们的编辑标准。

报道来源 [125]

llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-16 12:35

b10999

<details open=""> Change max context length for auto-fitting with unified KV (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28849">#28849</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofol…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-16 11:51

b10998

<details open=""> qwen4exp: add hc ops (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28901">#28901</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https://llama.app</a></li> </ul> …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-16 09:06

b10997

<details open=""> HIP: broaden MoE ncols_opt tile heuristic on RDNA3.5 architecture (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28935">#28935</a>) It's found the MoE ncols_opt tile heuristic needs to be broadened to includ…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-16 08:39

b10996

<details open=""> chat : force <code>\n</think></code> on reasoning budget end for qwen3-coder (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28869">#28869</a>) </details> Website: <ul> <li><a href="https…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-16 08:01

b10995

<details open=""> vulkan: make MUL_MAT_ID BN/2 tail unconditional (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28923">#28923</a>) Use BN/2 as the default for BNover2 and as the disabled fallback for BNover4, and remove the enable…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-16 07:27

b10994

<details open=""> metal: fix NaN in mul_mm_id when activations exceed f16 range (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/26223">#26223</a>) <ul> <li>test-backend-ops: reproduce MUL_MAT_ID NaN for activations beyond f16</li> </ul…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-16 06:56

b10993

<details open=""> ci : add self-hosted webgpu to hf-jobs (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28712">#28712</a>) <ul> <li> add self-hosted vulkan and webgpu to hf-jobs </li> <li> try t4-medium </li> <li> con…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-16 06:18

b10992

<details open=""> llama-bench: support --version to print build info (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28971">#28971</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">htt…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-15 23:46

b10991

<details open=""> hexagon: add back missing contiguous fast-path and hvx_copy_uu for each run (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28886">#28886</a>) </details> Website: <ul> <li><a href="https://llam…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-15 23:18

b10990

<details open=""> hex-cpy: use dma if src and dst are contiguous (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28906">#28906</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https:/…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-15 22:27

b10989

<details open=""> HIP: Enable AllReduce for ROCm (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/27825">#27825</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https://llama.app</a></…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-15 18:53

b10988

<details open=""> opencl: choose the MoE expert matmul by batch size for speculative decoding/MTP (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/27637">#27637</a>) <ul> <li> opencl: gate the prebuilt q4_0 MoE GEMM on routing count<…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-15 14:25

b10985

<details open=""> rpc : hash-cache only weights (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28789">#28789</a>) <ul> <li>rpc : hash-cache only weights</li> </ul> ggml_backend_rpc_buffer_set_tensor and ggml_backend_rpc_set_tensor_…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-15 13:46

b10984

<details open=""> cuda: support row-contiguous SUM_ROWS (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/26308">#26308</a>) <ul> <li> cuda: support row-contiguous SUM_ROWS </li> <li> organize the code and add GGML_OP_MEAN to s…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-15 13:13

b10983

<details open=""> models : move build_arch_graph() after graph() template specialization (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28934">#28934</a>) Move build_arch_graph()'s function definitions after the graph and gra…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-15 12:37

b10982

<details open=""> vulkan: support sparse Flash Attention (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28105">#28105</a>) <ul> <li> vulkan: add sparse Flash Attention support for DSV4/GLM </li> <li> tune implementation …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-15 12:03

b10981

<details open=""> OpenVINO: optimize stateful decode and GPU MoE inference (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28638">#28638</a>) <ul> <li> exclude GPU/NPU failing POOL_2D case </li> <li> Fix pool case </li> <…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-15 11:35

b10980

<details open=""> opencl: add generic ssm_scan (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28881">#28881</a>) <ul> <li> opencl: add generic ssm_scan </li> <li> opencl: fix whitespace </li> </ul> </details> …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-15 11:02

b10978

<details open=""> metal : add FA kernels for HSK=96, HSV=64 (MiniCPM3) (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28599">#28599</a>) <ul> <li>metal : add FA kernels for HSK=96, HSV=64 (MiniCPM3)</li> </ul> MiniCPM3 sets attenti…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-15 10:10

b10977

<details open=""> ci: Bump CUDA Windows x64 builds to 13.4.1 (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28930">#28930</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https://lla…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-15 07:26

b10976

<details open=""> ci : fix android release (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28936">#28936</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https://llama.app</a></li> </…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-14 20:56

b10970

<details open=""> HIP: fattn-mma: use fp32 accumulation on MFMA devices (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28576">#28576</a>) use fp32 accumulators in fattn-mma on CDNA </details> Website: <u…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-14 15:31

b10964

<details open=""> llama.cpp : bump version to 0.4.1 (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28900">#28900</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https://llama.app</a…
llama.cpp — Releases TIER_1 (CY) · ggerganov · 2026-09-14 14:01

b10964-b10964-b29c606

llama.cpp : bump version to 0.4.1 (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28900">#28900</a>)
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-14 14:00

b10956

<details open=""> sycl: rfc: Use radix select for top_k (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28670">#28670</a>) <ul> <li>sycl: GPU-resident TOP_K for large k, parallelised over the device</li> </ul> The SYCL backend refus…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-14 10:45

b10955

<details open=""> ggml-cpu : disable PCH and fix CACHE_LINE_SIZE ambiguity to fix heap corruption (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28882">#28882</a>) Disable the ggml-cpu precompiled header and remove the std::h…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-14 08:29

b10952

<details open=""> sycl : fix oneDNN scratchpad breaking the pool free order (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28704">#28704</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofoll…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-14 04:06

b10951

<details open=""> common : move llama_n_rs_seq to before llama_decode (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28749">#28749</a>) This commit moves the llama_n_rs_seq function call to before the llama_decode call and re…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-14 01:39

b10950

<details open=""> ggml-cuda: fallback to F32 on device without BF16 hardware acceleration (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28846">#28846</a>) <ul> <li> ggml-cuda: fallback to F32 on device without BF16 hardware accele…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-13 19:10

b10948

<details open=""> tests : exclude HY_V4 from WebGPU test-llama-archs tests (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28855">#28855</a>) Co-authored-by: Stanisław Szymczyk <a href="mailto:[email protected]">[email protected]<…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-13 18:45

b10947

<details open=""> models : guard the expert FFN size fallback in nemotron-h against a zero divisor (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28779">#28779</a>) The NextN/MTP tail loop derives the expert FFN size as n_ff/n_expe…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-13 18:21

b10946

<details open=""> ggml-cpu(s390x): guard VXE-only repack helpers (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28775">#28775</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https:/…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-13 16:14

b10944

<details open=""> sycl : Fix get mem error (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28227">#28227</a>) <ul> <li> fix for unsupport zes API </li> <li> optimize the code </li> <li> adjust the log level </li> <…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-13 10:36

b10941

<details open=""> tests : reduce FA test sizes (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28842">#28842</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https://llama.app</a></li…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-13 06:46

b10938

<details open=""> vulkan: workaround NV queuesubmit driver bug (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28830">#28830</a>) There is a driver bug where two queues on the same VkDevice simultaneously submitting can break …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-13 05:17

b10937

<details open=""> opencl: apply the noshuffle row-alignment rule to q4_K, q5_K and q8_0, not just q6_K (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28575">#28575</a>) </details> Website: <ul> <li><a href="htt…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-13 01:09

b10936

<details open=""> chat : improve parsing of complex types in qwen3-coder (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28742">#28742</a>) <ul> <li> chat : improve schema support in qwen3 parser </li> <li> cont : clean up gr…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-13 00:41

b10935

<details open=""> common: add LOG_JSON macro to log structured data (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28586">#28586</a>) <ul> <li> add LOG_JSON macro </li> <li> fit: add demo LOG_JSON </li> </ul> </details> …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-13 00:18

b10934

<details open=""> common : implement common_schema internal representation for JSON schemas (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28736">#28736</a>) <ul> <li> common : implement common_schema types </li> <li> common…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-12 23:37

b10933

<details open=""> jinja : support dot property integer literals (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28817">#28817</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https://…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-12 23:11

b10932

<details open=""> cmake: leave the timestamp out of precompiled headers on clang (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28816">#28816</a>) Clang stores the modification time of the precompiled header sources inside th…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-12 14:48

b10931

<details open=""> ui : add cache (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28802">#28802</a>) Signed-off-by: Adrien Gallouët <a href="mailto:[email protected]">[email protected]</a> </details> Website:</stro…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-12 13:03

b10930

<details open=""> server : allow model downloads at model limit fix issue <a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/issues/26809">#26809</a> (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28530">#2853…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-12 12:30

b10929

<details open=""> ggml-cuda: hip add specific config table for AMD GCN (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/27841">#27841</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">h…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-12 10:59

b10927

<details open=""> vendor : update cpp-httplib to 0.56.0 (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28787">#28787</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https://llama.ap…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-12 10:33

b10926

<details open=""> syscl : Handle (fail gracefully) unsupported tq1_0 quants (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28681">#28681</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofoll…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-12 09:20

b10924

<details open=""> server: frame the router child state command as a whole line (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28747">#28747</a>) The child writes its state commands on stdout while the logger writes on stderr,…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-12 08:58

b10923

<details open=""> opencl: fix several bugs where the backend aborts (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/27630">#27630</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">http…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-12 08:32

b10922

<details open=""> opencl: add bin kernel <code>kernel_gemm_noshuffle_q4_k_f32_32b_trans_ila_a8_bin</code> (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28677">#28677</a>) <ul> <li> opencl: add A8 Q4_K non-MoE binary kernel </l…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-12 08:06

b10921

<details open=""> webgpu: align tensor bindings to the type block size (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28382">#28382</a>) Walk the binding offset back until the distance to the tensor is a whole number of block…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-12 02:20

b10919

<details open=""> ggml-webgpu: Update to a recent version of Dawn (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28683">#28683</a>) <ul> <li> ggml-webgpu: Update to a recent version of Dawn </li> <li> No module scanning …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-11 20:32

b10917

<details open=""> cmake : skip PCH for llama-server PCH when using MSVC (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28763">#28763</a>) This commit fixes an issue that I introduced when adding PCH (precompiled headers) in C…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-11 12:15

b10909

<details open=""> metal : single-source fusion table + fusion debug rework (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28164">#28164</a>) <ul> <li>metal : rework fusion patterns into a single table</li> </ul> All fusable op patt…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-11 11:46

b10908

<details open=""> metal : fix idle threads in the remaining iq mul_mv kernels for ne00 < 1024 (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28692">#28692</a>) <ul> <li>metal : fix idle threads in the remaining iq mul_mv kernels fo…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-11 11:12

b10907

<details open=""> model : fix MTP context kv cache allocation for deepseek2, glm4moe, … (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28630">#28630</a>) <ul> <li> model : fix MTP context kv cache allocation for deepseek2, glm4moe,…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-11 10:41

b10906

<details open=""> server: fix speculation after an image (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28715">#28715</a>) <ul> <li>server: fix speculation after an image</li> </ul> Pass the actual position to the drafter after an …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-11 09:40

b10905

<details open=""> CUDA/HIP: Flash Attention tuning (gfx1201) (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28102">#28102</a>) <ul> <li>HIP: enable mma FA for head size 256 on RDNA4, tune configs</li> </ul> Assisted-by: Claude<br /…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-11 06:34

b10903

<details open=""> vulkan: fix data race and OOB access in argsort(large) (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28705">#28705</a>) argsort had a data race in the inner loop, which VVL caught. But I don't think this wa…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 22:42

b10901

<details open=""> vulkan: use CPU writes in ggml_backend_vk_cpy_tensor_async if the context is idle (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28618">#28618</a>) </details> Website: <ul> <li><a href="https:…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 19:02

b10900

<details open=""> vulkan: use add_alloc_dep to enable topk_moe fusion for prefill (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28422">#28422</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 18:35

b10899

<details open=""> vulkan: small M matrix optimizations for qwen (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28457">#28457</a>) <ul> <li> vulkan: optimize m=1 mul_mat by swapping A/B </li> <li> vulkan: Improve small M perf…
llama.cpp — Releases TIER_1 (SO) · shaofeiqi · 2026-09-10 18:25

b10902

opencl: add A8 Q4_0 mm binary kernel support (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28268">#28268</a>)
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 18:01

b10897

<details open=""> ci : Update WoA CUDA 13.4 release to use 13.4.1 GA redistributables (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28687">#28687</a>) <ul> <li>Move Windows ARM64 CUDA 13.4 builds from the Developer Preview archives t…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 15:45

b10896

<details open=""> spec: fix failed to decode mtmd chunk with DFlash (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28587">#28587</a>) <ul> <li>speculative: fix failed to decode mtmd chunk with DFlash</li> </ul> When using DFlash w/…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 13:54

b10894

<details open=""> models: clean up some dead switch branches in old models (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28669">#28669</a>) Some of these if statements were copypastaed in a former refactor and never cleaned …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 12:43

b10893

<details open=""> tests : increase tolerance for Add fusion tests (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28691">#28691</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https:…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 11:56

b10892

<details open=""> tests : drop SYCL special-casing in test-backend-ops.cpp (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28688">#28688</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollo…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 11:27

b10891

<details open=""> vulkan: fall back to shared-memory reduction for dmmv on PowerVR (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28341">#28341</a>) The Imagination proprietary Vulkan compiler returns VK_ERROR_UNKNOWN from vk…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 10:06

b10889

<details open=""> memory : avoid allocating V cache for indexer (it's not used) (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28330">#28330</a>) Co-authored-by: Stanisław Szymczyk <a href="mailto:[email protected]">sszymczy@gmail…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 09:41

b10888

<details open=""> vulkan : add command-buffer debug labels for GPU profilers (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28101">#28101</a>) <ul> <li>vulkan : add command-buffer debug labels for GPU profilers</li> </ul> Co-author…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 09:09

b10887

<details open=""> ggml-cpu(s390x): add repack support for q4_0 (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28667">#28667</a>) ggml-cpu: clean comments Signed-off-by: Aaron Teo <a href="mailto:[email protected]">aaron.teo…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 08:36

b10886

<details open=""> ggml-cpu(s390x): add Q1_0 vector intrinsic support (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28606">#28606</a>) <ul> <li>ggml-cpu: add <code>ggml_vec_dot_q1_0_q8_0</code> support</li> </ul> Signed-off-by: Aar…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-10 07:53

b10885

<details open=""> model: fix all granite family parameter counts (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28643">#28643</a>) <ul> <li>model: fix all granite family parameter counts</li> </ul> Signed-off-by: Aaron Teo <a href=…
llama.cpp — Releases TIER_1 (SO) · netrunnereve · 2026-09-09 19:46

b10884

ci: fix sanitizer tests (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28583">#28583</a>)
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-09 17:32

b10883

<details open=""> vulkan: use spec constant for matrix matrix multiplication A-type (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/25773">#25773</a>) <ul> <li>vulkan: use spec constant for mul mat type_a</li> </ul> vulkan: use map …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-09 15:34

b10881

<details open=""> vulkan: Convert FILL to distribute workgroups in 2D to avoid exceeding maxComputeWorkGroupCount (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28592">#28592</a>) <ul> <li>divide workload to 2D</li> </ul> This is t…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-09 13:25

b10878

<details open=""> llama : use int32_t for llama_sampler_chain_n return type (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28631">#28631</a>) Contributes to <a class="issue-link js-issue-link" href="https://github.com/ggml-org/llam…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-09 12:55

b10877

<details open=""> CUDA: size routed MoE MMQ N-tiles from typical expert width on RDNA3 (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28552">#28552</a>) Recreated from <a class="issue-link js-issue-link" href="https://github.com/gg…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-09 11:47

b10876

<details open=""> CUDA: replace GGML_FA_ALL_QUANTS with GGML_FA_QUANTS, more control over what is compiled (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28079">#28079</a>) <ul> <li>CUDA: add configurable FA quant combinations</li> </…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-09 11:17

b10875

<details open=""> args: officially deprecate --mmap|mlock|dio (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28334">#28334</a>) Signed-off-by: Aaron Teo <a href="mailto:[email protected]">[email protected]</a> </details> <…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-09 09:56

b10874

<details open=""> model: fix granite3 moe unknown parameter count (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28632">#28632</a>) Signed-off-by: Aaron Teo <a href="mailto:[email protected]">[email protected]</a> </details> …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-09 09:24

b10873

<details open=""> mtmd: propagate video ID to bitmap (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28601">#28601</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https://llama.app</…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-09 07:47

b10872

<details open=""> jinja: treat a null left operand of in as a plain lookup (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28620">#28620</a>) Templates that default an optional variable to none and then test its membership in …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-09 07:20

b10871

<details open=""> vulkan: add dedicated iq4_xs mat-vec shader (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28426">#28426</a>) <ul> <li>vulkan: add dedicated iq4_xs mat-vec shader</li> </ul> Dedicated mul_mat_vec_iq4_xs for the dm…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-09 06:40

b10870

<details open=""> vulkan: add f16 B-type matmul pipelines and warp tile size tuning for Intel coopmat1 (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/27471">#27471</a>) <ul> <li> vulkan: add f16 B-type matmul pipelines and warp til…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-09 05:27

b10869

<details open=""> tests : use less threads for data initialization (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28325">#28325</a>) <ul> <li> tests : use 1 thread for data initialization </li> <li> cont : scale threads with…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-09 02:03

b10868

<details open=""> Add IQ type handling for MoE (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28476">#28476</a>) Co-authored-by: cwriter <a href="mailto:cwriter@localhost">cwriter@localhost</a> </details> Website:</s…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-08 17:33

b10867

<details open=""> llama: disable lazy tensor loading by default on iGPUs (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28326">#28326</a>) <ul> <li> llama: add lazy mode auto, fix iGPU regression </li> <li> revert changes ex…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-08 15:14

b10865

<details open=""> Revert "ggml-cuda : restore prop.integrated on HIP builds (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/24233">#24233</a>)" (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28604">#2…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-08 14:33

b10864

<details open=""> server : apply checkpoint min-step eviction only when the checkpoint list is full (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28302">#28302</a>) The spacing eviction in create_checkpoint() keeps the oldest chec…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-08 14:01

b10863

<details open=""> metal : fix idle threads in mul_mv_iq3_xxs for ne00 < 1024 (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28086">#28086</a>) <ul> <li> metal : fix half-idle simdgroup in kernel_mul_mv_iq3_xxs_f32 for ne00 < …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-08 12:36

b10859

<details open=""> llama : add missing headers (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28566">#28566</a>) <ul> <li>fix compile-error: add missing header</li> </ul> Bug: <a class="issue-link js-issue-link" href="https://github…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-08 12:08

b10858

<details open=""> vulkan : fuse UNARY(GELU|SIGMOID|SILU|SOFTPLUS) + MUL (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/27220">#27220</a>) <ul> <li> vulkan : fuse UNARY(SIGMOID|SILU|SOFTPLUS) + MUL </li> <li> vulkan : fuse UN…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-08 11:38

b10857

<details open=""> Fix Vulkan-Hpp handle usage on 32-bit targets. (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/22892">#22892</a>) On 32-bit platforms, Vulkan non-dispatchable handles such as VkBuffer are represented as uint6…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-08 10:50

b10856

<details open=""> chat : split specialized parsers into common/parsers (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/27764">#27764</a>) <ul> <li>chat : split specialized parsers into common/parsers</li> </ul> Move the 14 dedicated…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-08 10:06

b10855

<details open=""> opencl: properly handle non-contiguous inputs to conv2d (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28503">#28503</a>) <ul> <li> opencl: fix conv2d non-contiguous strides </li> <li> opencl: format </…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-08 04:00

b10853

<details open=""> model : support Kimi-K3 recurrent-state rollback (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28466">#28466</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-08 00:42

b10852

<details open=""> hexagon: add RELU and LEAKY_RELU ops (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28585">#28585</a>) <ul> <li> hexagon: add RELU op </li> <li> hexagon: add LEAKY_RELU op too </li> </ul> </details> …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-07 20:05

b10850

<details open=""> tests : initialize the L2_NORM batch array (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28553">#28553</a>) <ul> <li>tests: bind the L2_NORM batch count to a local</li> </ul> GCC cannot prove the loop fills norms…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-07 19:30

b10844

<details open=""> vulkan: add DeepSeek-V4 hyper-connection fused ops (DSV4_HC_COMB/PRE/POST) (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/26578">#26578</a>) <ul> <li>vulkan: add DeepSeek-V4 hyper-connection fused ops (DSV4_HC_COMB/P…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-07 18:16

b10842

<details open=""> ggml: add gfx90c HIP support (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/26454">#26454</a>) <ul> <li> ggml: add gfx90c HIP support </li> <li> ggml: make gfx90c HIP support compliant with specifications</…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-07 14:33

b10840

<details open=""> CUDA: branchless Q4_K/Q5_K unpack to speed up mmvq, L2 prefetch on DGX Spark (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/26705">#26705</a>) <ul> <li> Update Q4_K and Q5_K to use branchless computation, which st…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-07 11:16

b10839

<details open=""> vulkan: support type-aligned GET_ROWS (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28253">#28253</a>) <ul> <li>vulkan: fall back to CPU for GET_ROWS with misaligned offsets</li> </ul> The Vulkan GET_ROWS shader …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-07 08:37

b10837

<details open=""> caps : recheck typed content if template checks for string (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28511">#28511</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofol…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-07 08:11

b10835

<details open=""> ggml-cuda: fix divergent barrier in f16 flash attention (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/27870">#27870</a>) <ul> <li> ggml-cuda: fix divergent barrier in f16 flash attention </li> <li> ggml-cu…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-07 07:20

b10834

<details open=""> ggml: allow backend inputs to not create another split (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28387">#28387</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow"…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-07 06:51

b10833

<details open=""> vulkan: rms_norm fusion opportunities (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28024">#28024</a>) Support RMS_NORM + MUL + ADD (+ MUL) and RMS_NORM + VIEW + SET_ROWS. Extend ROPE + VIEW + SET_ROWS to s…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-07 05:21

b10831

<details open=""> vulkan: add TQ1_0 support (mm, mat-vec, mat-vec-id, dequant, get_rows) (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/27765">#27765</a>) <ul> <li> vulkan: add TQ1_0 support (mm, mat-vec, dequant, get_rows) </l…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-06 23:30

b10830

<details open=""> convert : add <code>--fuse-qkv</code> flag to fuse Q/K/V into QKV during HF-to-GGUF conversion (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/22780">#22780</a>) </details> Website: <ul> <li><a…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-06 23:06

b10829

<details open=""> models : fix GDN normalization from <code>max</code> to <code>rsqrt</code> (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28068">#28068</a>) <ul> <li>models: use flash-linear-attention's l2norm for gated delta net q/…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-06 22:11

b10828

<details open=""> [Model] Support for Spark2_5ForCausalLM implementation (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/27868">#27868</a>) <ul> <li>Add Spark3 Model</li> <li>rename spark3 -> spark2_5</li> </ul> Co-authored-by: S…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-06 20:52

b10827

<details open=""> opencl: properly choose weights pack for q4_K, q5_K mul_mat (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28402">#28402</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofo…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-06 16:42

b10826

<details open=""> cuda: fixes races in mmid and mmf (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28475">#28475</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https://llama.app</a…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-06 13:24

b10825

<details open=""> grammar : fix max repetition threshold (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28469">#28469</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https://llama.a…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-06 11:59

b10823

<details open=""> common: add --log-jsonl (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28437">#28437</a>) <ul> <li> common: add --log-jsonl </li> <li> rename unknown to none </li> </ul> </details> Website:</…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-06 11:32

b10822

<details open=""> ui : embed assets directly with CMake (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28445">#28445</a>) Remove the build-time C++ helper and external gzip dependency, simplifying cross-compilation. Keep the …
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-06 10:01

b10821

<details open=""> metal : add remaining fa-vec tunings for M2 Max (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28458">#28458</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https:…
llama.cpp — Releases TIER_1 (SO) · JohannesGaessler · 2026-09-05 20:42

b10820

Github: limit blank issues to maintainers (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28435">#28435</a>)
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-05 10:38

b10819

<details open=""> metal : fix memory leak in early return (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28399">#28399</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https://llama.…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-05 09:28

b10817

<details open=""> sycl: attribute device allocations by site (GGML_SYCL_MEMTRACE) (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/27631">#27631</a>) define two new environment variables to better understand how much memory is …
llama.cpp — Releases TIER_1 (SO) · philip-jingxin · 2026-09-05 02:37

b10818

sycl : fix test-backend-ops CI break && restore Kronecker product FWH…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-04 20:00

b10816

<details open=""> metal : add remaining fa-vec tunings for M3 (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28396">#28396</a>) <ul> <li> addition of m3 in fa_vec_tuned_table </li> <li> adding q4_0,q4_1,q5_0,q5_1 in ggml-met…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-04 19:32

b10814

<details open=""> opencl: extend the elementwise and data‐movement op coverage (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/27633">#27633</a>) <ul> <li>opencl: add extended elementwise unary ops (sgn, step, elu, hardswish, hardsigmo…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-04 19:05

b10813

<details open=""> opencl: add Adreno xmem SDPA path (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/26331">#26331</a>) <ul> <li>opencl: add Adreno xmem SDPA path</li> </ul> Assisted-by: Codex <ul> <li> Removed the Adreno-spec…
llama.cpp — Releases TIER_1 (SO) · github-actions[bot] · 2026-09-04 17:57

b10809

<details open=""> llama.cpp : bump version to 0.4.0 (<a class="issue-link js-issue-link" href="https://github.com/ggml-org/llama.cpp/pull/28386">#28386</a>) </details> Website: <ul> <li><a href="https://llama.app" rel="nofollow">https://llama.app</a…

本文如何被排名

报道来源 [125]

相关实体

相关话题