PR #21160 Cross-backend profiler

Cross-backend profiler #21160

pwilkin wants to merge 17 commits into master from cross-profiler

github-actions added Nvidia GPU

github-actions added Vulkan

github-actions added examples

github-actions added python

github-actions added ggml

github-actions added SYCL

github-actions added Apple Metal

github-actions added Ascend NPU

github-actions added OpenCL

github-actions added IBM zDNN

github-actions added Hexagon

github-actions added OpenVINO

github-actions added WebGPU

github-actions added documentation

pwilkin force pushed from 532a8ebd to 1874fb9e 141 days ago

pwilkin force pushed from 1874fb9e to 0a20d658 139 days ago

github-actions added AMD ZenDNN

pwilkin force pushed from ec4489aa to d35b7668 139 days ago

github-actions added testing

pwilkin force pushed from d35b7668 to 65ab03a3 135 days ago

pwilkin force pushed from bd6f8008 to 09cd7887 134 days ago

pwilkin force pushed from 09cd7887 to 57e5c878 129 days ago

pwilkin force pushed from 57e5c878 to ac735691 127 days ago

pwilkin force pushed from ac735691 to 88d8c574 111 days ago

pwilkin force pushed from 88d8c574 to f0210cc4 103 days ago

0cc4m commented on 2026-05-13

pwilkin force pushed from d28472d3 to 7a3ed632 99 days ago

pwilkin force pushed from 2fc9362b to 3d909435 98 days ago

pwilkin force pushed from 5ce1c2a9 to e78392f4 94 days ago

pwilkin force pushed from e78392f4 to 78b55149 93 days ago

pwilkin force pushed from 9663aa8f to a8c5534d 86 days ago

pwilkin force pushed from a8c5534d to 2bfe4ff9 75 days ago

pwilkin force pushed from 22634e0e to 447144dc 51 days ago

github-actions added CUDA

pwilkin force pushed from 447144dc to 720846f2 50 days ago

pwilkin force pushed from 720846f2 to 6eac19fe 35 days ago

github-actions added server

ggml-cuda: avoid ROCm_Host compute on HIP integrated GPUs

fdc1260e

ggml-cuda: fix unused parameter in non-HIP builds

ce82541a

pwilkin force pushed from 6eac19fe to 180121b7 28 days ago

feat: cool profiler thingy

7705d453

add second dimension to reported tensors, fix Mac build, add missing …

153f8d92

Fix more missing backend stuff (and Python errors)

609c7867

fix builds, integrate vulkan profiler, fix copy events, fix export

2a2476e1

fix mul_mat_id stats, add throughput stat, add envvar trigger, add co…

7c22aecd

docs, pass copy details

abdf1bf2

Add missing op parameters to the profiler; add support for test-backe…

bfbf84a4

Converge implementation with export-graph-ops

c7d42e00

Fix FATTN profiling

49471724

Fix braces

f1c08eac

Revert accidental change.

695dd1ea

Add missing unrolls

2ade1b82

tentative Metal support

9ed6048d

Add tensor name to JSON output

df6daabc

pwilkin force pushed from 180121b7 to df6daabc 27 days ago

Merge branch 'fix-rocm' into cross-profiler

e7613f32

Reviewers

0cc4m

Assignees

No one assigned

Labels

documentation testing Nvidia GPU Vulkan examples python server ggml SYCL Apple Metal Ascend NPU OpenCL IBM zDNN Hexagon CUDA AMD ZenDNN OpenVINO WebGPU

Milestone

No milestone

llama.cpp Cross-backend profiler #21160 Open

Cross-backend profiler #21160

llama.cpp
Cross-backend profiler
#21160

Open