Re: [hatari-devel] Adding cache support for the MegaSTE

[Nicolas Pomarède <npomarede@xxxxxxxxxxxx>]

Le 13/07/2024 à 09:45, Christian Zietz a écrit :
> Nicolas Pomarède schrieb:
>
> > that's what I thought too, that it was not a very efficient cache but
> > being able to cache data and not code would be better than nothing :)
>
> As the CoreMark values (https://github.com/czietz/coremark/wiki/Results) 
> prove, it is actually a very effective cache: The 16 MHz result is 
> almost double the 8 MHz result, meaning that you have a high cache hit 
> rate overall.
>
> > Yes, there was a typo. Attached version fixes "R words, R words" case
>
> Result attached.
>
> I don't understand the "cache flush, R words" number, though, as it is 
> slower than the "no cache" number, which cannot possibly be. Could you 
> check this test, as well?

yes, I forgot to reset the counter before this test, so it adds time 
with the previous one :(

This is fixed in attached prg.

Attachment: mste_cac.prg
Description: Binary data