Search results

HIP: disable sync warp shuffel operators from clr amd_warp_sync_functions.h

ggmlchanges relating to the ggml tensor library for machine learningchanges relating to the ggml tensor library for machine learning

Nvidia GPUIssues specific to Nvidia GPUsIssues specific to Nvidia GPUs

Status: Open (in progress).

ggml-org/llama.cppnumber 15273

#15273 In ggml-org/llama.cpp;

· IMbackK opened on Aug 12, 2025

opencl: add initial mxfp4 support via mv

ggmlchanges relating to the ggml tensor library for machine learningchanges relating to the ggml tensor library for machine learning

OpenCLIssues specific to the OpenCL backendIssues specific to the OpenCL backend

Status: Open (in progress).

ggml-org/llama.cppnumber 15270

#15270 In ggml-org/llama.cpp;

· lhez opened on Aug 12, 2025

WIP: ggml-cuda: Add bf16 cuda support to fattn (Flash Attention)

examples

ggmlchanges relating to the ggml tensor library for machine learningchanges relating to the ggml tensor library for machine learning

Nvidia GPUIssues specific to Nvidia GPUsIssues specific to Nvidia GPUs

pythonpython script changespython script changes

Status: Open (in progress).

ggml-org/llama.cppnumber 15261

#15261 In ggml-org/llama.cpp;

· eous opened on Aug 12, 2025

musa: fix build warnings

ggmlchanges relating to the ggml tensor library for machine learningchanges relating to the ggml tensor library for machine learning

Nvidia GPUIssues specific to Nvidia GPUsIssues specific to Nvidia GPUs

Status: Open (in progress).

ggml-org/llama.cppnumber 15258

#15258 In ggml-org/llama.cpp;

· yeahdongcn opened on Aug 12, 2025

vulkan: fuse adds

ggmlchanges relating to the ggml tensor library for machine learningchanges relating to the ggml tensor library for machine learning

testingEverything test relatedEverything test related

VulkanIssues specific to the Vulkan backendIssues specific to the Vulkan backend

Status: Open (in progress).

ggml-org/llama.cppnumber 15252

#15252 In ggml-org/llama.cpp;

· jeffbolznv opened on Aug 11, 2025

vulkan: perf_logger improvements

ggmlchanges relating to the ggml tensor library for machine learningchanges relating to the ggml tensor library for machine learning

VulkanIssues specific to the Vulkan backendIssues specific to the Vulkan backend

Status: Open (in progress).

ggml-org/llama.cppnumber 15246

#15246 In ggml-org/llama.cpp;

· jeffbolznv opened on Aug 11, 2025

ggml-rpc: chunk send()/recv() to avoid EINVAL for very large tensors over RPC (macOS & others)

ggmlchanges relating to the ggml tensor library for machine learningchanges relating to the ggml tensor library for machine learning

Status: Open (in progress).

ggml-org/llama.cppnumber 15188

#15188 In ggml-org/llama.cpp;

· Tak-RS opened on Aug 9, 2025

ggml: add <code>conv3d</code> op

ggmlchanges relating to the ggml tensor library for machine learningchanges relating to the ggml tensor library for machine learning

testingEverything test relatedEverything test related

Status: Open (in progress).

ggml-org/llama.cppnumber 15182

#15182 In ggml-org/llama.cpp;

· rmatif opened on Aug 8, 2025

SVE support for exponential functions

ggmlchanges relating to the ggml tensor library for machine learningchanges relating to the ggml tensor library for machine learning

Status: Open (in progress).

ggml-org/llama.cppnumber 15145

#15145 In ggml-org/llama.cpp;

· s-goto-11 opened on Aug 7, 2025

CUDA: Optimize <code>reduce_rows_f32</code> kernel, leading up to 25x perf improvement on kernel-level and 10% perf increase for Gemma3n

ggmlchanges relating to the ggml tensor library for machine learningchanges relating to the ggml tensor library for machine learning

Nvidia GPUIssues specific to Nvidia GPUsIssues specific to Nvidia GPUs

testingEverything test relatedEverything test related

Status: Open (in progress).

ggml-org/llama.cppnumber 15132

#15132 In ggml-org/llama.cpp;

· ORippler opened on Aug 6, 2025

ggml: aarch64: Implement SVE F16 kernels for vector functions

ggmlchanges relating to the ggml tensor library for machine learningchanges relating to the ggml tensor library for machine learning

Status: Open (in progress).

ggml-org/llama.cppnumber 15115

#15115 In ggml-org/llama.cpp;

· Vithulep opened on Aug 6, 2025

CUDA/HIP: ssm-scan: switch from shared memory to reisters, fixes indexing problem on warp64 devices

ggmlchanges relating to the ggml tensor library for machine learningchanges relating to the ggml tensor library for machine learning

Nvidia GPUIssues specific to Nvidia GPUsIssues specific to Nvidia GPUs

Status: Open (in progress).

ggml-org/llama.cppnumber 15101

#15101 In ggml-org/llama.cpp;

· IMbackK opened on Aug 5, 2025