Skip to content

NIXLBench: Align memory for O_DIRECT - #557

Merged
aranadive merged 25 commits into
ai-dynamo:mainfrom
cheese-head:nixlbench/enable_directio
Jul 16, 2025
Merged

aranadive merged 25 commits into
ai-dynamo:mainfrom
cheese-head:nixlbench/enable_directio

Conversation

@cheese-head

Copy link
Copy Markdown
Contributor

What?

This PR provides support for O_DIRECT in NIXLBench.

Why?

To provide a more accurate representation of storage access performance when using the POSIX Plugin

How?

It aligns memory to the underlying page size using posix_memalign

@copy-pr-bot

copy-pr-bot Bot commented Jul 9, 2025

Copy link
Copy Markdown

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@github-actions

github-actions Bot commented Jul 9, 2025

Copy link
Copy Markdown

👋 Hi cheese-head! Thank you for contributing to ai-dynamo/nixl.

Your PR reviewers will review your contribution then trigger the CI to test your changes.

🚀

@cheese-head

Copy link
Copy Markdown
Contributor Author

cherry pick of 504

cheese-head and others added 2 commits July 9, 2025 14:09
Signed-off-by: Patrick Riel <priel@nvidia.com>
vvenkates27
vvenkates27 previously approved these changes Jul 10, 2025

@vvenkates27 vvenkates27 left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

mostly just sytle changes and includes posix_memalign. LGTM

Comment thread benchmark/nixlbench/src/worker/nixl/nixl_worker.cpp Outdated
ovidiusm
ovidiusm previously approved these changes Jul 10, 2025
Comment thread benchmark/nixlbench/src/worker/nixl/nixl_worker.cpp Outdated
Comment thread benchmark/nixlbench/src/worker/nixl/nixl_worker.cpp Outdated
Comment thread benchmark/nixlbench/src/worker/nixl/nixl_worker.cpp
Comment thread benchmark/nixlbench/src/worker/nixl/nixl_worker.cpp Outdated
Comment thread benchmark/nixlbench/src/worker/nixl/nixl_worker.cpp
Comment thread benchmark/nixlbench/src/worker/nixl/nixl_worker.cpp
@aranadive

Copy link
Copy Markdown
Contributor

/ok to test 4105513

@aranadive

Copy link
Copy Markdown
Contributor

/build

@aranadive

Copy link
Copy Markdown
Contributor

/ok to test 4044503

@aranadive

Copy link
Copy Markdown
Contributor

/build

@aranadive
aranadive merged commit fdf1d29 into ai-dynamo:main Jul 16, 2025
aranadive added a commit to aranadive/nixl that referenced this pull request Jul 16, 2025
* update nixlbench/kvbench docs

* updated readme

* update readme

* update tutorial for gds

* apply lint rules

* isort fix

* enable direct io support in nixlbench

* apply clang-format

Signed-off-by: Patrick Riel <priel@nvidia.com>

* fix clang-formatting

Signed-off-by: Patrick Riel <priel@nvidia.com>

* check for null pointer

Signed-off-by: Patrick Riel <priel@nvidia.com>

* merge main

* revert clang formatting

* calculate page size during init

* refactor transfer mem allocation

* add optional param for calloc allocations

Signed-off-by: Patrick Riel <priel@nvidia.com>

---------

Signed-off-by: Patrick Riel <priel@nvidia.com>
Co-authored-by: Adit Ranadive <aranadive@nvidia.com>
pvijayakrish pushed a commit that referenced this pull request Jul 17, 2025
* update nixlbench/kvbench docs

* updated readme

* update readme

* update tutorial for gds

* apply lint rules

* isort fix

* enable direct io support in nixlbench

* apply clang-format



* fix clang-formatting



* check for null pointer



* merge main

* revert clang formatting

* calculate page size during init

* refactor transfer mem allocation

* add optional param for calloc allocations



---------

Signed-off-by: Patrick Riel <priel@nvidia.com>
Co-authored-by: Patrick <71560045+cheese-head@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants