Re: [hatari-devel] Hatari data cache tests

[Eero Tamminen <oak@xxxxxxxxxxxxxx>]

Hi,

On sunnuntai 21 kesäkuu 2015, Nicolas Pomarède wrote:
> Le 20/06/2015 11:39, Douglas Little a écrit :
> > The tests I did were in CE/PF mode. It seemed like the i-cache was
> > affecting cycles but the d-cache was not (cycle sum unaffected), even
> > if the d-cache hit ratio would change.
> > 
> > However I only looked at that one sequence of code - I'll look closer
> > elsewhere and see if there is some additional context.
> 
> I don't use profiler that often, but maybe there's a bug in it when
> accumulating stats for all the loops ?
> 
> Is it possible to profile the block of instructions you posted in
> previous mail but by running it only once ? This way we can see if
> cycles reported by the profiler's stats are the same as those shown when
> running hatari with "--trace cpu_all".

Just do:

	profile on
	trace cpu_all
	c <instruction count>
	profile addresses       [1]

And compare the results.


	- Eero

[1] press Enter to get more lines disassembled,
    or give address from which to disassemble.