Re: [Gems-users] No user-mode misses


Date: Tue, 17 Aug 2010 10:39:40 +0930
From: tian@xxxxxxxxxxxxxxxxxxxxxxx
Subject: Re: [Gems-users] No user-mode misses
Hello Arka

   Thanks for the reply. You are right, this is not gems's fault. it turns
out to be no user mode request at all.

   But ... I havent change any setting in simics. Why would this happen?

   I've also randomly pause the simulation and manually call the
SIM_processor_privilege_level function, it always return a "1"  which
is supervisor_mode.

   The workload im running is facesim of PARSEC with input simsmall. Is it
possible that simics is not capable to differentiating user/kernel when
running PARSEC?

   Thank you very much

Best Regards

G.Tian

> Hello,
>
>       I would suggest first try to look at whether you are actually
> getting any user level requests or not (before looking at user level
> miss). In $GEMS/ruby/system/Sequencer.C there is a function called
> makeRequest. It receives the cache request message. First try to see
> if you get any request in user more. You can do that by checking
> (request.m_AccessMode !=  AccessModeType_SupervisorMode).
>
>
> Thanks
> Arka
>
>> Anyone have any idea about it?   I've tried many different protocols and
>> workload .... i haven't seen a single user_miss .  All the misses are
>> supervisor mode...
>>
>> Is there anything need to specify in simics to identify supervisor/user
>> mode ?
>>
>> Thank you very much.
>>
>> Best Regards
>> G.tian
>>
>>
>>
>>>
>>> Hi there:
>>>
>>> Anyone know why all the misses are supervisor mode. what i run is a
>>> parsec
>>> workload, it suppose to have at least a few user mode misses isn't it?
>>>
>>>
>>>
>>> Total_misses: 2273071
>>> total_misses: 2273071 [ 230701 53714 46071 42464 49511 47244 43489
>>> 42229
>>> 65465 71766 99267 58954 209682 1045153 58800 108561 ]
>>> user_misses: 0 [ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ]
>>> supervisor_misses: 2273071 [ 230701 53714 46071 42464 49511 47244 43489
>>> 42229 65465 71766 99267 58954 209682 1045153 58800 108561 ]
>>>
>>> Best Regards
>>>
>>> G.Tian
>>>
>>> _______________________________________________
>>> 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.
>>>
>>
>>
>> _______________________________________________
>> 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.
>>
>>
>
>
> _______________________________________________
> 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.
>


[← Prev in Thread] Current Thread [Next in Thread→]