What does benchmarking a Dell Inspiron Mini 9, a Radeon HD 4890 graphics card with Intel Core i7, and dual quad-core AMD Opterons look like? Well, if the systems are running Linux, BSD, OpenSolaris, or Mac OS X, it can look like this: Do you see any of those systems out in the lemon grove? Nope. As was shared earlier this month, with Phoronix Test Suite 2.0 (a.k.a...
Looks nice, hopefully you don't forget the issues i found with PTS lately.
a) vdrift preconfig - shader = false
in order to benchmark fglrx drivers which lockup if shaders are used. Still waiting to see a fixed fglrx driver as it sucks when you run a suite and then the system locks up. Nothing else you expect from ATI cards, do you
b) More depends:
b1) byte -> time package (check /usr/bin/time)
b2) build-imagemagick -> minimum libperl-dev
c) Some checks for invalid benchmarks (highly required!)
c1) check size of time files used for compression -> leads to wrong results for example with 7zip compression when hd is full.
c2) Check error level after make for compile benchmark - currently bad builds are of course faster than correct ones - just like that
build-imagemagick bench which could not link against perl and is of course faster that way.