Andrei Stoian
87c0d646a4
fix(gpu): coprocessor bench
2025-09-18 13:56:55 +02:00
David Testé
f8684d1f67
chore(ci): add regression benchmark workflow
...
Regression benchmarks are meant to be run in pull-request. They
can be launched in two flavors:
* issue comment: using command like "/bench --backend cpu"
* adding a label: `bench-perfs-cpu` or `bench-perfs-gpu`
Benchmark definitions are written in TOML and located at
ci/regression.toml.
While not exhaustive, it can be easily modified by reading the
embbeded documentation.
"/bench" commands are parsed by a Python script located at
ci/perf_regression.py. This script produces output files that
contains cargo commands and a shell script generating custom
environment variables. The Python script and generated files are
meant to be used only by the workflow
benchmark_perf_regression.yml.
2025-09-16 13:33:49 +02:00
Agnes Leroy
0ed97cfba8
chore(gpu): update sxm5 cost
2025-09-10 10:49:25 +02:00
pgardratzama
11b540c456
chore(hpu): adds cost of hpu setups
2025-09-05 10:42:36 +02:00
Guillermo Oyarzun
b5a7199c15
chore(gpu): update cuda version in ci
2025-07-25 15:57:25 +02:00
Agnes Leroy
701411044b
chore(gpu): update SXM5 cost
2025-07-03 17:00:02 +01:00
Agnes Leroy
d3f36b1d86
chore(gpu): fallback to a6000 in case L40's are out of stock
2025-06-05 16:55:10 +02:00
David Testé
2749c1088c
chore(ci): handle multi directories for parameters records
2025-05-22 15:03:02 +02:00
David Testé
97b5973e4c
chore(bench): store object measurements results in tfhe-benchmark
2025-05-13 16:05:16 +02:00
Arthur Meyre
dc9afe1146
chore: bump to 1.1 and add V1_1 parameters
...
- add aliases for tests to avoid having to upgrade too many locations
2025-03-20 17:47:18 +01:00
David Testé
8287f59ebd
chore(ci): update aws ami for cpu
...
This is done to update python modules: pip, wheel and setuptools.
2025-03-20 15:47:17 +01:00
Agnes Leroy
d53db210de
chore(gpu): fix multi-gpu integer throughput bench
2025-03-12 12:16:00 +01:00
Agnes Leroy
3621d12c42
chore(ci): add hourly cost for sxm5 vms
2025-03-06 10:19:49 +01:00
David Testé
8565b79a28
chore(ci): switch environment and add fallback for gpu profiles
...
Switch n3-H100-SXM5x8 to US-1 as CANADA is out of stock on this
instance.
Also L40 instances fallback on n3-RTX-A6000x1 to mitigate
resource shortages issues.
2025-02-27 16:59:04 +01:00
David Testé
486ec9f053
chore(ci): update cpu aws ami and install git-lfs
...
Several network errors occurred while trying to install git-lfs
from within backward compatibility tests workflow. Having git-lfs
installed directly in the Amazon Machine Image fix this issue.
2025-02-25 12:45:47 +01:00
David Testé
7ee49387fe
chore(ci): deduplicate parameters set to send to lattice estimator
...
From SageMath point of view some tfhe-rs parameters set are
equivalent. We deduplicate those by storing their name in the tag
field. Grouping them that way we decrease analysis time
dramatically.
2025-02-13 17:10:45 +01:00
David Testé
3df08e9259
chore(ci): install github runner as ubuntu user for gpu workflows
2025-02-12 12:10:53 +01:00
Agnes Leroy
b22e369166
chore(ci): switch gpu tests on push to l40 and deactivate fast pks
2025-01-06 16:54:40 +01:00
Agnes Leroy
4388a3dc99
chore(gpu): add sxm5 vm target
2024-12-13 17:25:55 +01:00
David Testé
8b460072d3
chore(ci): check more parameters set on lattice estimator
2024-12-09 16:19:50 +01:00
Arthur Meyre
36deaec607
chore(ntt): bring concrete-ntt in the repo as tfhe-ntt
2024-11-21 10:47:48 +01:00
David Testé
b31fbf5f23
chore(bench): fix result parsing for object and key generation
2024-11-20 11:55:22 +01:00
David Testé
9059ddeacc
chore(bench): add throughput benchmarks to suite
...
All integer benchmarks make recipes can be run to ouput throughput
results.
Only CPU is supported for throughput benchmarks in GitHub CI.
2024-11-19 12:07:50 +01:00
Arthur Meyre
c9b4ee84ae
chore(ci): fix fft bench parser
2024-11-19 10:43:23 +01:00
Arthur Meyre
6a161fef0a
chore: bring concrete-fft as tfhe-fft in the repo
2024-11-18 13:17:58 +01:00
Nicolas Sarlin
a45b7b3974
chore(zk): add benches to tfhe-zk-pok
2024-11-18 13:17:28 +01:00
Agnes Leroy
a501285206
chore(gpu): change target for multi-gpu tests
2024-11-13 15:06:46 +01:00
Arthur Meyre
eac30027e9
chore(ci): run bench profile as ubuntu
2024-11-12 16:59:38 +01:00
David Testé
083e973fb2
chore(ci): update version of selenium to 4.26.0
...
This reduces console verbosity.
2024-10-31 09:30:41 +01:00
David Testé
f3a1b6b5b9
chore(ci): run aws workflows as ubuntu user
2024-10-18 14:05:36 +02:00
David Testé
1a5dfb3699
chore(ci): rename benchmark parsing option to object-sizes
...
Since not only key sizes are measured now, it makes more sense to
rename it to --object-sizes.
2024-10-16 16:28:11 +02:00
David Testé
483a4fecf1
chore(ci): fix firefox wasm benchmarks with new aws ami
...
Some libs were missing to be able to run Firefox out of the box.
Besides, action runner is now installed as ubuntu user since
Firefox is not able to run as root.
2024-10-16 10:50:27 +02:00
David Testé
ff0609fba3
chore(ci): force refresh page on firefox browser
...
A simple driver.refresh() wouldn't refresh script cache for web
workers. Thus leading to page not fully charged and having all
test and benchmarks buttons in disabled state.
That triggering timeout while browsing with Selenium.
Console log print has also been curated for Chrome browser.
2024-10-14 17:58:21 +02:00
David Testé
41b3edf48b
chore(ci): pass --no-sandbox option only to chrome browser
2024-10-14 17:58:21 +02:00
Agnes Leroy
e698d18242
chore(gpu): automatically generate rust bindings for cuda functions, except device.cu
2024-10-14 17:07:57 +02:00
David Testé
43e697ce51
chore(ci): fix browser identification on wasm benchmarks
...
Rust script wasn't able to parse results if the browser name was
suffixed after parameters name.
2024-10-09 15:26:42 +02:00
David Testé
cd36ac5092
chore(ci): add firefox support for wasm tests and benchmarks
2024-10-09 08:44:16 +02:00
David Testé
de7c7f209f
chore(ci): include snippets folder into tfhe npm package
2024-09-26 15:40:31 +02:00
Agnes Leroy
b365585c74
chore(gpu): add 2xH100 bench workflow
2024-09-26 12:42:05 +02:00
Agnes Leroy
2bf483c596
chore(gpu): add bench workflow on L40
2024-09-25 09:13:22 +02:00
Arthur Meyre
43d91f512f
chore(ci): use python as webdriver for wasm test and benchmarks
...
Swtich from Jest and Puppeteer to Python with Selenium. It relies
on browser and webdriver binaries from browser fabricators.
For now the Python script only supports Chrome browser.
2024-09-25 09:11:13 +02:00
David Testé
3ab7f49436
chore(ci): remove support for slab calls with issue comments
...
Now all workflows use Slab GitHub Action and thus can be launched
directly with a workflow_dispatch event.
2024-09-18 13:42:17 +02:00
Agnes Leroy
8299e1cb9a
chore(gpu): change multi-gpu tests to run on rtx so it's cheaper
2024-09-16 09:04:56 +02:00
Arthur Meyre
92be95c6b8
chore(ci): fix parsing for integer benchmarks
2024-08-14 13:42:14 +02:00
Agnes Leroy
5340859003
chore(ci): transfer all GPU CI to hyperstack
2024-08-07 17:08:13 +02:00
Agnes Leroy
eba4f6a89c
chore(gpu): use PCIe H100 for multi-gpu bench
2024-07-31 08:50:07 +02:00
Agnes Leroy
d9eca01631
fix(gpu): dispatch/gather inputs and outputs to the ks and pbs on all GPUs
2024-07-23 08:48:48 +02:00
David Testé
a34ddd7b54
chore(ci): split cuda-pcc and cuda-tests jobs to parallelize execution
2024-06-28 10:30:17 +02:00
David Testé
2ff3b75ef7
chore(ci): update gpu ec2 ami with rustup snap package removed
...
Rustup was installed using snap and could clash in the workflow
notably on cargo-clippy calls.
2024-06-20 14:42:02 +02:00
Ben
915eafac15
chore(ci): fix non-float print
2024-06-20 08:35:37 +02:00
