On Aug 24, 2021, at 9:52 AM, OpenSmalltalk-Bot ***@***.***> wrote:



...

On 2021-08-23, at 11:58 PM, Eliot Miranda ***@***.***> wrote:

Hi Tim,

...

On Raspberry Pi 4 v8 it compiles cleanly but appears to be ~10% slower on the tinybenchmarks. I know they're not exactly high-value tests but the difference is noticeable

I have reason not to believe this. In all the recent changes, no change has been made to the code generated for tinyBenchmarks. And I’m confident that the other changes made (save the now fixed regression) yield significant speed ups. Is there any other reason the timings could have changed?

I truly have no ideas here. Today that same VM is reporting numbers that are more in line with what I expect (ie 790&54 instead of 820&56) that could be trivial sampling error/aliasing. Yesterday it was wildly leaping between 720&54 to 320&26. No readily apparent reason; the temperature log wasn't fluctuating, the throttling wasn't being triggered, the cpu usage log wasn't suggesting anything stealing time, nothing like that.

Try changing the cog method alignment. It may make a difference. You’ll need to open up the code generator or hind the relevant define in the cogitARMv8.c file. Currently the scheme chooses the smallest alignment that allows the checked and unchecked entry points to be distinguished. But increasing the alignment should have no ill effects at a small decrease in code compactness.

tim

tim Rowledge; ***@***.***; http://www.rowledge.org/tim Useful random insult:- Several nuts over fruitcake minimum.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.