Skip to content

common : throttle download progress output to reduce IO flush - #17427

Merged
ggerganov merged 1 commit into
ggml-org:masterfrom
angt:common-throttle-download-progress-output-to-reduce-io-flush
Nov 30, 2025
Merged

common : throttle download progress output to reduce IO flush#17427
ggerganov merged 1 commit into
ggml-org:masterfrom
angt:common-throttle-download-progress-output-to-reduce-io-flush

Conversation

@angt

@angt angt commented Nov 21, 2025

Copy link
Copy Markdown
Member

This change limits progress updates to approximately every 0.1% of the file size to minimize stdio overhead.

Also fixes compiler warnings regarding __func__ in lambdas.

@loci-dev loci-dev mentioned this pull request Nov 21, 2025
Open
@angt
angt requested a review from ngxson November 28, 2025 14:02
ngxson
ngxson approved these changes Nov 28, 2025
View reviewed changes
Comment thread common/download.cpp Outdated
@angt
common : throttle download progress output to reduce IO flush
3fc9f54 
This change limits progress updates to approximately every 0.1% of the
file size to minimize stdio overhead.

Also fixes compiler warnings regarding __func__ in lambdas.

Signed-off-by: Adrien Gallouët <angt@huggingface.co>
@angt
angt force-pushed the common-throttle-download-progress-output-to-reduce-io-flush branch from 5e49b30 to 3fc9f54 Compare November 28, 2025 14:43
@taronaeo

taronaeo commented Nov 29, 2025

Copy link
Copy Markdown
Member

By any chance, do you have merge permissions? Otherwise I can help to merge this PR

ggerganov
ggerganov reviewed Nov 29, 2025
View reviewed changes
Comment thread common/download.cpp

std::atomic<size_t> downloaded{existing_size};
const char * func = __func__; // avoid __func__ inside a lambda
size_t downloaded = existing_size;

@ggerganov ggerganov Nov 29, 2025

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

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

Any reason that this was std::atomic<size_t> before?

@angt angt Nov 29, 2025

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

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

I have no idea, maybe to be safer than needed. I noticed it was useless when adding progress_step.

@ngxson ngxson Nov 30, 2025

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

I think this is OK to merge if you confirm that the atomic was not important here.

We can re-implement it once we move the progress tracking to a higher scope though (i.e. shared among threads). This will be important once we implement the model download function in llama-server

@angt angt Nov 30, 2025

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

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

Yes absolutely useless

@angt

angt commented Nov 29, 2025 •
edited
Loading

Copy link
Copy Markdown
Member Author

By any chance, do you have merge permissions? Otherwise I can help to merge this PR

Thanks @taronaeo, I don't have merge permissions, but I’m not sure I've fully convinced @ggerganov about the atomic yet 😂

@angt angt mentioned this pull request Nov 30, 2025
Merged
@ggerganov
ggerganov merged commit beb1f0c into ggml-org:master Nov 30, 2025
72 of 74 checks passed
Anico2 added a commit to Anico2/llama.cpp that referenced this pull request Jan 15, 2026
@Anico2
common : throttle download progress output to reduce IO flush (ggml-o…
7a9c4fd 
…rg#17427)

This change limits progress updates to approximately every 0.1% of the
file size to minimize stdio overhead.

Also fixes compiler warnings regarding __func__ in lambdas.

Signed-off-by: Adrien Gallouët <angt@huggingface.co>
blime4 referenced this pull request in blime4/llama.cpp Feb 5, 2026
@angt shaobo.xie
common : throttle download progress output to reduce IO flush (#17427)
8022435 
This change limits progress updates to approximately every 0.1% of the
file size to minimize stdio overhead.

Also fixes compiler warnings regarding __func__ in lambdas.

Signed-off-by: Adrien Gallouët <angt@huggingface.co>
Seunghhon pushed a commit to Seunghhon/llama.cpp that referenced this pull request Apr 26, 2026
@angt
common : throttle download progress output to reduce IO flush (ggml-o…
5c8379a 
…rg#17427)

This change limits progress updates to approximately every 0.1% of the
file size to minimize stdio overhead.

Also fixes compiler warnings regarding __func__ in lambdas.

Signed-off-by: Adrien Gallouët <angt@huggingface.co>
ljubomirj pushed a commit to ljubomirj/llama.cpp that referenced this pull request May 6, 2026
@angt
common : throttle download progress output to reduce IO flush (ggml-o…
5cac407 
…rg#17427)

This change limits progress updates to approximately every 0.1% of the
file size to minimize stdio overhead.

Also fixes compiler warnings regarding __func__ in lambdas.

Signed-off-by: Adrien Gallouët <angt@huggingface.co>
my-other-github-account pushed a commit to my-other-github-account/llama.cpp that referenced this pull request May 15, 2026
@angt
common : throttle download progress output to reduce IO flush (ggml-o…
39c4b8c 
…rg#17427)

This change limits progress updates to approximately every 0.1% of the
file size to minimize stdio overhead.

Also fixes compiler warnings regarding __func__ in lambdas.

Signed-off-by: Adrien Gallouët <angt@huggingface.co>
my-other-github-account pushed a commit to my-other-github-account/llama.cpp that referenced this pull request May 15, 2026
@angt
common : throttle download progress output to reduce IO flush (ggml-o…
64bd022 
…rg#17427)

This change limits progress updates to approximately every 0.1% of the
file size to minimize stdio overhead.

Also fixes compiler warnings regarding __func__ in lambdas.

Signed-off-by: Adrien Gallouët <angt@huggingface.co>
phibya pushed a commit to ziee-ai/llama.cpp that referenced this pull request May 29, 2026
@angt
common : throttle download progress output to reduce IO flush (ggml-o…
7c62534 
…rg#17427)

This change limits progress updates to approximately every 0.1% of the
file size to minimize stdio overhead.

Also fixes compiler warnings regarding __func__ in lambdas.

Signed-off-by: Adrien Gallouët <angt@huggingface.co>
fewtarius pushed a commit to fewtarius/CachyLLama that referenced this pull request May 30, 2026
@angt
common : throttle download progress output to reduce IO flush (ggml-o…
8ffa3b5 
…rg#17427)

This change limits progress updates to approximately every 0.1% of the
file size to minimize stdio overhead.

Also fixes compiler warnings regarding __func__ in lambdas.

Signed-off-by: Adrien Gallouët <angt@huggingface.co>
fukuro-kun pushed a commit to fukuro-kun/fukuro-llama-cpp-turboquant that referenced this pull request Jul 5, 2026
@angt
common : throttle download progress output to reduce IO flush (ggml-o…
19683a8 
…rg#17427)

This change limits progress updates to approximately every 0.1% of the
file size to minimize stdio overhead.

Also fixes compiler warnings regarding __func__ in lambdas.

Signed-off-by: Adrien Gallouët <angt@huggingface.co>
MrLordCat referenced this pull request in MrLordCat/llama.cpp-with-GUI Jul 16, 2026
@angt
common : throttle download progress output to reduce IO flush (#17427)
a93f148 
This change limits progress updates to approximately every 0.1% of the
file size to minimize stdio overhead.

Also fixes compiler warnings regarding __func__ in lambdas.

Signed-off-by: Adrien Gallouët <angt@huggingface.co>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@ngxson ngxson ngxson approved these changes
@ggerganov ggerganov ggerganov approved these changes

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

4 participants

@angt @taronaeo @ggerganov @ngxson