Re: [hatari-devel] Hatari profiler updates and CPU cycle questions

[ Thread Index | Date Index | More Archives ]


On torstai 31 tammikuu 2013, Douglas Little wrote:
> That makes sense. I have been wary of the cache miss information because
> its harder to correlate with anything 'real' - there is no alternative
> source for it, and until the cycle counts look right I figured the cache
> miss stuff could be off too.

Could you somehow verify that it's not off-by-one, like the cycles
information was?

> However if you think it's working I'll start working with it - but as I
> say it's very difficult to check independently.

Somebody could check through WinUAE news (from about year ago when
Laurent imported the code) to see whether there are any comments about
bugs in cache handling code...

> I'll see how I get on with it.

Then there's the data cache thing, but as earlier discussed,
that would probably need separate statistics for reads and writes.

IMHO it's better to look into that only after i-cache is somehow verified
to be about correct.  Maybe after next week.

> > Do you think there's some problem with the Devpac symbols converter?
> The last test worked ok for me. I only inspected text symbols but I saw
> no pollution from data/bss in the text area... so it looks like it is
> working.

Good. :-)

	- Eero

Mail converted by MHonArc 2.6.19+