Re: [Gems-users] High IFETCH ratio

[Edward Lee <edwl202@xxxxxxxxx>]

Thanks Dan for the prompt answer although you make me more worried now :)

Pretty much rest of the kernels have ~1-2% IFETCH ratio out of all
events for the parallel section. Would that be because I only captured
the parallel section trace? Also, just to make sure ruby0.tracer
captures all the events that are issued from the processors right?

Regards,

Ed

On Thu, Apr 15, 2010 at 9:12 PM, Dan Gibson <degibson@xxxxxxxx> wrote:
> I would be worried if IFETCH wasn't quite high... seeing as every LD or ST
> requires an IFETCH... as do all the non-memory instructions.
>
> Regards,
> Dan
>
> On Thu, Apr 15, 2010 at 6:12 PM, Edward Lee <edwl202@xxxxxxxxx> wrote:
>>
>> Hi,
>>
>> I collected traces from (parallel sections only) my SPLASH2 runs.
>> Couple of the benchmarks (namely Barnes-16k and raytrace) show 40%
>> IFETCH count and pretty much ~1% STORE count and the rest is LOADs. Is
>> having this large IFETCH counts normal, as the rest of the suite has
>> less than 1% ratio for IFETCH events?
>>
>> Regards,
>>
>> Ed
>> _______________________________________________
>> Gems-users mailing list
>> Gems-users@xxxxxxxxxxx
>> https://lists.cs.wisc.edu/mailman/listinfo/gems-users
>> Use Google to search the GEMS Users mailing list by adding
>> "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.
>>
>
>
>
> --
> http://www.cs.wisc.edu/~gibson [esc]:wq!
>
> _______________________________________________
> Gems-users mailing list
> Gems-users@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> Use Google to search the GEMS Users mailing list by adding
> "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.
>
>
>