Thanks so much, Nithesh. Currently it just happens for mcf benchmark. I'll
test more. Thanks for kind help.


On Mon, Jul 7, 2014 at 3:56 PM, Nithesh Kurella <[email protected]> wrote:

> So, I checked the source file of mcf benchmark and can confirm  "read
> error, exit" is being called by the benchmark. If this is only benchmark
> which errors out, I wouldn't worry about changes that were made in
> simulator. If not, you might want to start with debug mode to dump trace of
> events which lead to error.
>
> -Nithesh
>
>
> On Mon, Jul 7, 2014 at 3:33 PM, Qi Jia <[email protected]> wrote:
>
>> Hi Nithesh,
>>
>> I added an additional tags which could be used to store the blocks
>> evicted from the original tags. When blocks are stored into the additional
>> tags I copied the status/tag/data from the original block to the new block
>> in additional tags.
>>
>>
>> On Mon, Jul 7, 2014 at 2:39 AM, Nithesh Kurella <[email protected]>
>> wrote:
>>
>>> Hi
>>>
>>> It would be helpful if you can let us know what changes you did to
>>> include the victim cache.
>>>
>>> Thanks
>>> Nithesh
>>> Hi Ali,
>>>
>>> I am not sure how to tell whether it comes from simulator or benchmarks.
>>> It just displayed "read error, exit". But if I skip the first 200 million
>>> instructions by fastforward, then it works well so far and I set the
>>> --maxinsts 1 billion. I also searched the codes of GEM5 but I do not find
>>> where it would print "read error. exit".
>>>
>>> Thanks
>>>
>>> _______________________________________________
>>> gem5-users mailing list
>>> [email protected]
>>> http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
>>>
>>
>>
>>
>> --
>> Qi Jia
>> Graduate Student
>> Department of Electrical and Computer Engineering
>> North Carolina State University, Raleigh
>>
>
>


-- 
Qi Jia
Graduate Student
Department of Electrical and Computer Engineering
North Carolina State University, Raleigh
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

Reply via email to