feat(packages): add sglang-ssd-stream-test harness for the PLE decision #24

Open
hermes wants to merge 1 commit from hermes/ssd-stream-test-v2 into main AGit
Contributor

Test-only package: pinned fork of upstream sglang-ssd-stream's install
(wheel v0.1.0 sha256-pinned, SGLang 3df8e1e7, flashinfer 0.6.17) plus a
shared streaming tok/s bench, to measure the SSD-streamed PLE alternative
against the productive SGLang container before committing to the 128 GiB
RAM purchase for --ple-offload-embedding. Not wired into any host.

Co-authored-by: Hermes hermes@hoyer.xyz

Test-only package: pinned fork of upstream sglang-ssd-stream's install (wheel v0.1.0 sha256-pinned, SGLang 3df8e1e7, flashinfer 0.6.17) plus a shared streaming tok/s bench, to measure the SSD-streamed PLE alternative against the productive SGLang container before committing to the 128 GiB RAM purchase for --ple-offload-embedding. Not wired into any host. Co-authored-by: Hermes <hermes@hoyer.xyz>
feat(packages): add sglang-ssd-stream-test harness for the PLE decision
All checks were successful
nix flake check / flake-check (pull_request) Successful in 3m51s
31bd2f274d
Test-only package: pinned fork of upstream sglang-ssd-stream's install
(wheel v0.1.0 sha256-pinned, SGLang 3df8e1e7, flashinfer 0.6.17) plus a
shared streaming tok/s bench, to measure the SSD-streamed PLE alternative
against the productive SGLang container before committing to the 128 GiB
RAM purchase for --ple-offload-embedding. Not wired into any host.

Co-authored-by: Hermes <hermes@hoyer.xyz>
All checks were successful
nix flake check / flake-check (pull_request) Successful in 3m51s
This pull request can be merged automatically.
This branch is out-of-date with the base branch
You are not authorized to merge this pull request.
View command line instructions

Checkout

From your project repository, check out a new branch and test the changes.
git fetch -u origin +refs/pull/24/head:hermes/ssd-stream-test-v2
git switch hermes/ssd-stream-test-v2
Sign in to join this conversation.
No reviewers
No labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
harald/nixcfg!24
No description provided.