Re: benchmarks (ashes or otherwise)

[Bruno Dufour <bruno.dufour@mail.mcgill.ca>]

Hi,

	Unfortunately, all of the ashes benchmarks have a fixed input. You can 
either change the input of benchmarks (when possible) to make them run 
longer, or you can wrap their execution inside a main method to make them 
perform the same task multiple times. If you know some basic shell scripting, 
then either of these solution should be rather simple to implement, although 
not convenient if you have to do it for a large number of benchmarks.

	Such problems will be fixed in Ashes2, the next generation of the Ashes 
benchmarking framework. There is a pre-release of Ashes2 available at:
http://www.sable.mcgill.ca/~bdufou1/ashes2/
but I still have to convert most benchmarks to use the new format. 
Suggestions, comments, feature requests or other forms of contribution are 
always welcome. If you tell me which particular benchmarks you are interested 
in, I can probably manage to to package them within a reasonable time frame.

Best regards,

	Bruno Dufour

On October 7, 2003 12:43 pm, Will Benton wrote:
> Hello,
>
> I'm interested in using a sampling profiler to time some idiomatic OO
> programs and to get a better idea of how to characterize their
> workloads.  It seems that many of the ashes benchmarks fit the bill
> of "idiomatic OO programs", but they don't run long enough to get
> reliable/repeatable results from the sampler.  Is there some way to
> run some of the ashes benchmarks with larger workloads without too
> much script-munging?
>
>
>
> thanks,
> wb