> I was just working on something similar when I found your post. > Now I just have to adapt it to the Lgred datasets. I do have a few > questions though, that I hope you might be able to answer.
I fear the general answer is going to be "because I didn't and still don't know much about Python" but here goes... > 1. Why do you have a specific constructor calling the LiveProcess > base class? I thought this would be done automatically. So did I but without it I got errors. > 2. There is a lot of duplicate code, is there any reason that you haven't > made a class like Gzip that sets f.ex "executable" and let GzipSource > inherit that from Gzip? Yup, that would seem like a good idea (in hindsight). It was actually generated by a big regexp from a similar config file of mine for another simulator which didn't support classes. > Is there any way to set the working directory for the simulated binary? > Copying the files work, but it is a bit dirty... I chose to copy the files because that works best on the cluster of machines I'm using to run my simulations and it's also what the official runspec script does which is why I can take the command lines from the website and use them pretty much verbatim. Some of the benchmarks have implicit assumptions about where to find files relative to the cwd, so if one were to just use a single shared directory then you could end up with multiple output files clobbering each other. Hope this helps somewhat, James -- [EMAIL PROTECTED] http://www.cl.cam.ac.uk/~jrs53/ _______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
