Skip to content
TFHE Cuda Backend - 4090 benchmarks

feat(gpu): implement CUDA Graph to accelerate default classical and multibit PBS #1052

Sign in to view logs

feat(gpu): implement CUDA Graph to accelerate default classical and multibit PBS

feat(gpu): implement CUDA Graph to accelerate default classical and multibit PBS #1052

Triggered via pull request December 27, 2024 18:27
@pdroalvespdroalves
labeled #1907
pa/feat/cudagraph
Status Success
Total duration 8h 27m 6s
Artifacts 2

benchmark_gpu_4090.yml

on: pull_request
Cuda integer benchmarks (RTX 4090)
43m 50s
Cuda integer benchmarks (RTX 4090)
Cuda core crypto benchmarks (RTX 4090)
7h 36m
Cuda core crypto benchmarks (RTX 4090)
Remove 4090 bench label
2s
Remove 4090 bench label
Fit to window
Zoom out
Zoom in

Annotations

1 error and 2 warnings
Remove 4090 bench label
Error: failed to remove labels: 4090_bench
Remove 4090 bench label
ubuntu-latest pipelines will use ubuntu-24.04 soon. For more details, see https://github.com/actions/runner-images/issues/10636
Remove 4090 bench label
failed to remove label: 4090_bench: HttpError: Label does not exist

Artifacts

Produced during runtime
Name Size
dcf2cec97ea9e614f47c174192321fb420024edc_core_crypto
4.12 KB
dcf2cec97ea9e614f47c174192321fb420024edc_integer_multi_bit_gpu_default
4.09 KB