[swift-dev] Mysterious memory improvement and regression
gparker at apple.com
Mon Apr 24 18:28:53 CDT 2017
> On Apr 21, 2017, at 3:57 AM, Pavol Vaskovic via swift-dev <swift-dev at swift.org> wrote:
> Hi swift-dev!
> I was working over the Easter holiday weekend on script that reports the benchmark results comparison. I wasn’t always up-to-date with the master, but I have noticed a mysterious change in MAX_RSS value reported from benchmarks. The change went away again in later builds.
> For example the AngryPhonebook benchmark usually reports MAX_RSS of 10 543 104 - so around 10 MB. But my local builds from 2017-04-15 had all similar tests drop to around 3 MB (AngryPhonebook reported 3 506 176).
> That seems like a significant and positive change, but later builds returned to previous values. I have tried to look through the commits around that time, but nothing there strikes me as obvious cause.
> Forgive my cluelessness, I lack proper visibility into the projects current development, but is there some work going on in this area, that got merged and reverted?
The value of MAX_RSS depends on OS behavior. Other activity on the same machine may change MAX_RSS of the benchmark. The changes you saw might not be "real".
If you re-run the 2017-04-15 builds today, do you still see 3 MB instead of 10 MB?
Greg Parker gparker at apple.com Runtime Wrangler
More information about the swift-dev