Question on size increase

[Scott Violet]

Hello,

One of my patches is failing the fuschia-binary-size bot. The uncompressed size grows by 4k, and the compressed size by 16k. Is it normal to see a 4x *increase* in size between uncompressed and compressed? I suspect what is causing the increase is making a number of functions inline. This is being done to improve performance as this code is on the critical path for rendering pages.

Patch is here: https://chromium-review.googlesource.com/c/chromium/src/+/4020984

Thanks,

  -Scott

[Rohan Pavone]

This is a small growth for good performance improvements (which we might see on Fuchsia as well). Feel free to use the `Fuchsia-Binary-Size` footer and explain this is a performance improvement, so the size growth can be ignored.

As for the growth, the compression algorithm pushes the compressed code into pages of 8k. This means that if the compression without your change caused an 8k page to only be partially filled, and your change causes a small, but meaningful growth in the compression that pushes the size beyond that 8k page, you will see at least a +8k increase in size.

I am curious as to why you saw +16k (meaning a whole other page was filled, which shouldn't be the case for a 4k change). I will forward this case to the owners of the compression tool to investigate.

[Scott Violet]

Thanks for looking Rohan. I'll use the Fuchsia-Binary-Size footer.

  -Scott

On Thu, Nov 10, 2022 at 2:38 PM Rohan Pavone <rohp...@google.com> wrote:

Hey Scott,

This is a small growth for good performance improvements (which we might see on Fuchsia as well). Feel free to use the `Fuchsia-Binary-Size` footer and explain this is a performance improvement, so the size growth can be ignored.

As for the growth, the compression algorithm pushes the compressed code into pages of 8k. This means that if the compression without your change caused an 8k page to only be partially filled, and your change causes a small, but meaningful growth in the compression that pushes the size beyond that 8k page, you will see at least a +8k increase in size.

I am curious as to why you saw +16k (meaning a whole other page was filled, which shouldn't be the case for a 4k change). I will forward this case to the owners of the compression tool to investigate.

--
You received this message because you are subscribed to the Google Groups "fuchsia-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fuchsia-dev...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/fuchsia-dev/CAKARY_%3DRj_-bSry_ZRkcR0saGGtEiQN7vy%3DTkzGOKv1J1gm_tA%40mail.gmail.com.
For more options, visit https://groups.google.com/a/chromium.org/d/optout.