On Mon, 29 Jun 2026 09:35:59 GMT, Egor Ushakov <[email protected]> wrote:

> Here are the results:
> It took 3ms - No stepping, just a breakpoint on the line with println and 
> resume
> It took 50045ms - jdk 25
> It took 15686ms - jdk 27 Build 28 (2026/6/25)
>
> So the stepping speedup looks significant!
> One thing I’m still wondering about with this fix, should we expect the 
> stepping case to get close to the 3 ms baseline, or is the remaining overhead 
> expected? In other words, do you have an idea why it is still ~15.7s rather 
> than near the no-stepping case?

Thank you for testing this. I'd expect less overhead. Will try to investigate 
it. Something is not right here.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/28407#issuecomment-4901129895

Reply via email to