Search results

Vulkan: Fix host-pinned memory for large allocations

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 14249

#14249 In ggml-org/llama.cpp;

· 0cc4m opened on Jun 17, 2025

ggml : implement op fusion, starting with REGLU/GEGLU/SWIGLU

Apple Metalhttps://en.wikipedia.org/wiki/Metal_(API)https://en.wikipedia.org/wiki/Metal_(API)

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

help wantedExtra attention is neededExtra attention is needed

Nvidia GPUIssues specific to Nvidia GPUsIssues specific to Nvidia GPUs

SYCLhttps://en.wikipedia.org/wiki/SYCL - GPU programming languagehttps://en.wikipedia.org/wiki/SYCL - GPU programming language

testingEverything test relatedEverything test related

VulkanIssues specific to the Vulkan backendIssues specific to the Vulkan backend

Status: Draft (not ready).

ggml-org/llama.cppnumber 14158

#14158 In ggml-org/llama.cpp;

· CISC opened on Jun 12, 2025

Add support for VK_EXT_debug_utils to add labels to Vulkan objects.

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 13792

#13792 In ggml-org/llama.cpp;

· mtavenrath opened on May 26, 2025

Fix Vulkan glslc invocation command lines

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 13289

#13289 In ggml-org/llama.cpp;

· kangalio opened on May 3, 2025

llama : try loading tensors with pre-computed hashes

Apple Metalhttps://en.wikipedia.org/wiki/Metal_(API)https://en.wikipedia.org/wiki/Metal_(API)

examples

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

Komputehttps://github.com/KomputeProject/kompute/https://github.com/KomputeProject/kompute/

Nvidia GPUIssues specific to Nvidia GPUsIssues specific to Nvidia GPUs

SYCLhttps://en.wikipedia.org/wiki/SYCL - GPU programming languagehttps://en.wikipedia.org/wiki/SYCL - GPU programming language

VulkanIssues specific to the Vulkan backendIssues specific to the Vulkan backend

Status: Open (in progress).

ggml-org/llama.cppnumber 13106

#13106 In ggml-org/llama.cpp;

· rgerganov opened on Apr 25, 2025

Vulkan: Remove dedicated aligned matrix matrix multiplication shaders

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: Draft (not ready).

ggml-org/llama.cppnumber 12515

#12515 In ggml-org/llama.cpp;

· 0cc4m opened on Mar 22, 2025

Fixed Eval Bug: 12163 : Fallback to CPU when loading model: vk::PhysicalDevice::createDevice: ErrorExtensionNotPresent.

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 12329

#12329 In ggml-org/llama.cpp;

· ashwini778 opened on Mar 11, 2025

<code>tool-call</code>: Phi-4 support

androidIssues specific to AndroidIssues specific to Android

Apple Metalhttps://en.wikipedia.org/wiki/Metal_(API)https://en.wikipedia.org/wiki/Metal_(API)

devopsimprovements to build systems and github actionsimprovements to build systems and github actions

documentationImprovements or additions to documentationImprovements or additions to documentation

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

server

SYCLhttps://en.wikipedia.org/wiki/SYCL - GPU programming languagehttps://en.wikipedia.org/wiki/SYCL - GPU programming language

testingEverything test relatedEverything test related

VulkanIssues specific to the Vulkan backendIssues specific to the Vulkan backend

Status: Open (in progress).

ggml-org/llama.cppnumber 12288

#12288 In ggml-org/llama.cpp;

· jpohhhh opened on Mar 9, 2025

vulkan: optimization proposals for coopmat1 mul_mm

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: Draft (not ready).

ggml-org/llama.cppnumber 12260

#12260 In ggml-org/llama.cpp;

· remyoudompheng opened on Mar 7, 2025

vulkan : add GGML_VK_FORCE_HEAP_INDEX env var

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 9734

#9734 In ggml-org/llama.cpp;

· gyf304 opened on Oct 4, 2024

vulkan: fix storageBuffer16BitAccess detection on some adreno driver

bugfixfixes an issue or bugfixes an issue or bug

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

Review Complexity : LowTrivial changes to code that most beginner devs (or those who want a break) can tackle. e.g. UI fixTrivial changes to code that most beginner devs (or those who want a break) can tackle. e.g. UI fix

VulkanIssues specific to the Vulkan backendIssues specific to the Vulkan backend

Status: Open (in progress).

ggml-org/llama.cppnumber 8837

#8837 In ggml-org/llama.cpp;

· rhjdvsgsgks opened on Aug 2, 2024