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

support compress & decompress by stream(block by block) #384

Open
wants to merge 12 commits into
base: master
Choose a base branch
from

Conversation

sisong
Copy link

@sisong sisong commented Aug 7, 2024

for decompressor

Added several lower-level APIs with little as possible source code changes, to support decompress block by block;
And updated the gzip cmdline with new APIs, the memory usage when decompress will be greatly reduced, usually <1MB!
Also, because it needs to support any .gz format (not just libdeflate), the decompress buffer will automatically adjust the size to fit the input maximum block size.

Because decompressor reused a small memory, the CPU cache hit ratio will be greatly improved, resulting in an increase in decompress speed;
In my multiple test cases, it was able to get 20%--50% faster.

for compressor

Added several lower-level APIs with minor source code changes, to support compress block by block;
And updated the gzip cmdline by new APIs, the memory usage when compress will be greatly reduced;
The memory requires is related to the block size set and compress level, not related to input file size; when the block size is 2 MB and compress level is 12, the memory usage ~ 13 MB.
These new APIs can be used for multi-threaded parallel compression, and there is almost no loss of compress ratio.


( I used libdeflate for my multiple actual projects and make a lot of feature changes; Hopefully commit some useful changes back to libdeflate. )

@sisong sisong changed the title support decompress by stream(block by block) support compress & decompress by stream(block by block) Aug 8, 2024
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.

2 participants