Dan Kegel wrote:
On Sun, Jun 1, 2008 at 1:16 AM, Dan Kegel dank@kegel.com wrote:
http://kegel.com/new-computer-2008.html ).
Result: ... it takes me a bit longer today on this machine (11 minutes) to build as it did then on a dual cpu hot rod (8 minutes). I guess Moore's Law isn't quite keeping up with Wine's code growth :-)
Details are on buried that page, but it's "-j3" for the dual core / dual CPU systems.
I found a nice graph of Wine's codebase size over time: http://www.ohloh.net/projects/wine/analyses/latest Looks like it's just about exactly doubled since that last measurement in late 2002.
http://techgage.com/article/intel_core_2_duo_e7200_-_the_new_budget_supersta... has some recent wine build benchmarks:
Time to build wine-0.9.59: quad core, -j 5: QX9770: 182s QX9650: 193s QX6850: 200s Q9450: 216s
I have tested this on F9 with gcc-4.3.0 and Wine-1.0 and get for the Q9450 305 seconds. The compiler advances kill performance faster than CPU vendors can add it... Though to be fair gcc-4.3.0 does more checks than gcc-4.1.x by default with -Wall; the bounds check come to mind.
bye michael
Q6600: 251s
dual core, -j3: e8400: 363s e7200: 437s e6750: 422s e6600: 472s
Golly. Maybe I should have gotten a quad core :-)
BTW, I just compared the e7200 and e8400 on the 'valgrind wine test suite with -j2' task. Results: e8400 is 18% faster (83 minutes vs. 99 minutes). I'd love to try this on a quad core... maybe I'll find one at work and give it a shot.