On Mon, Aug 9, 2010 at 12:23 PM, nathan binkert <[email protected]> wrote: >> Seems like an overoptimization to me, unless we plan on using readfile to >> bring in some really huge files. Also, there's no harm in doing this in >> non-x86 ISAs. I won't fight putting those changes in, I'm just questioning >> their necessity. > > Well, we don't know how it impacts performance. The concern is that > it just slows down simulations unnecessarily. I don't know what it > has to do with the size of files. It's sizeof(buf) which is > statically allocated, so we pay the same amount every time. I won't > fight either.
OK, yea, I was responding to the comment w/o looking at the code, sorry. Given that it's outside the loop I'd say the motivation for optimization is even lower. OTOH, I suppose if you add up the energy cost of doing 64*511 or so extra simulated stores every time you run m5 it might be worth optimizing... I'm pretty neutral overall, I suppose. _______________________________________________ m5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/m5-dev
