Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

chore: Use base docker image #132

Merged
merged 6 commits into from
Jul 18, 2024

Conversation

wojciechsromek
Copy link
Collaborator

No description provided.

@philsippl philsippl merged commit 362f63b into feat/INFRA-2749 Jul 18, 2024
6 of 7 checks passed
@philsippl philsippl deleted the feat/INFRA-2759/use-base-docker-image branch July 18, 2024 16:07
philsippl added a commit that referenced this pull request Aug 5, 2024
…stances for build (#119)

* feat: [INFRA-2749] change label to gpu for build on gpu-based instances

* feat: [INFRA-2749] change labels to match for gpu to use gpu-based instances for build

* feat: [INFRA-2749] change labels to match for gpu to use gpu-based instances for tests

* ci: try running E2E test on self-hosted gpu runner

* ci: install some missing packages

* ci: also apt update

* install cuda and nccl

* cuda 12.1

* new try

* more deps

* more

* more

* manual

* quiet

* deb

* quiet

* another try

* up

* remove existing

* up

* up

* up

* install all deps

* up

* up

* up

* up

* up

* up

* nccl

* don't install nccl from source

* up

* up

* up

* up

* cuda home

* export path and cache

* fix

* use 12.1

* cuda 12.2

* up

* up

* chore: Use base docker image (#132)

* Use base docker image

* add curl

* incorrect version fix

* do not use gpu on cis where not needed

* cleanup

---------

Co-authored-by: Philipp Sippl <[email protected]>

* dbg print cuda

* up

* .

* up

* new try

* dbg

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* test

* revert container options

* test

* Trigger Build

* remove fstab changes

* try NCCL_P2P_DIRECT_DISABLE

* disable shm

* also run normal tests

* remove tests again

* dbg

* dbg

* revert dbg

* limit db

* revert limit

* only random

* no insertions

* sync

* add compute-sanitizer test to ci

* fix broken shell command

* install compute-sanitizer as well

* test

* test

* hardcode compute-sanitizer path

* disable write

* dbg

* dbg

* test waiting for all work to finish  per job

* only 1 batch

* 2 batches

* 3 batches

* reducing complexity, db_sizes no longer need to be mutexed

* make e2e test deterministic, add super sync after each iteration

* only use 1 stream atm

* alloc on streams

* another sync after alloc

* sync after each op

* tracing

* try not freeing stuff

* Revert "try not freeing stuff"

This reverts commit 987fa80.

* Revert "tracing"

This reverts commit 7569f98.

* test replacing ptr casts with normal cuda types

* Revert "test replacing ptr casts with normal cuda types"

This reverts commit c480df7.

* try dirty hack to use default streams

* don't access null ptr streams

* Revert "don't access null ptr streams"

This reverts commit 4e04d49.

* Revert "try dirty hack to use default streams"

This reverts commit 7972c61.

* Revert "sync after each op"

This reverts commit 03aadd7.

* log mem addresses

* Revert "log mem addresses"

This reverts commit 331257b.

* remove bind thread

* dbg

* dbg

* dbg

* dbg

* dbg

* dbg

* up

* dbg

* dbg

* dbg

* dbg

* dbg

* up

* up

* up

* up

* up

* dbg: max for all

* up

* up

* up

* up

* up

* up

* up

* 2 byte aligned

* odd len in phase 2

* dbg

* up

* up

* cublas test

* up

* up

* up

* up

* add asserts

* cublasGetStatusString

* remove cuda test

* update batch size in server

* PR feedback

* Revert "alloc on streams"

This reverts commit 898eb13.

* fmt

---------

Co-authored-by: Daniel Kales <[email protected]>
Co-authored-by: philsippl <[email protected]>
Co-authored-by: wojciechsromek <[email protected]>
Co-authored-by: wojciechsromek <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants