You're right, I was confused... "-i" is the short form of "--input" but "-o"
is the short form of "--options" and not "--output".

There's still definitely something weird going on though, since alpha is a
little-endian architecture.  We run vortex on ALPHA_SE in our internal
regressions with "lendian.raw" and it works fine.

Steve

On Fri, Jun 5, 2009 at 5:12 PM, Alexander Neckar
<[email protected]>wrote:

> Hi, I've been working with the original poster. We eventually got vortex to
> run with the big-endian input data set. So we would have been calling:
> build/ALPHA_SE/m5.fast -d output configs/example/se.py -n 1 -d --caches
> --l2cache -c vortex00.peak.ev6 -o "bendian2.raw"
> *
> *
> As we understand it, the -o option is for input options to your
> application. I don't think we were ever 100% sure about this, but what
> seemed to work for us is that if a benchmark used a stream operator to take
> an input file, we used -i (input), and for anything else we used -o
> "(input)".
> *
> *
> On Fri, Jun 5, 2009 at 6:51 PM, Steve Reinhardt <[email protected]> wrote:
>
>> Did you solve this yet?  If not, do these benchmarks work without "-d
>> --caches --l2cache"?
>>
>> On Mon, Jun 1, 2009 at 7:41 AM, Anitha Mohan <
>> [email protected]> wrote:
>>
>>>
>>> *>> build/ALPHA_SE/m5.fast -d output configs/example/se.py -n 1 -d
>>> --caches --l2cache -c vortex00.peak.ev6 -o "lendian2.raw" -i lendian2.raw
>>> *
>>>
>>
>> The fact that you're using the same file for both the input and the output
>> seems suspicious to me...
>>
>> Steve
>>
>>
>> _______________________________________________
>> m5-users mailing list
>> [email protected]
>> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
>>
>
>
> _______________________________________________
> m5-users mailing list
> [email protected]
> http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
>
_______________________________________________
m5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/m5-users

Reply via email to