blender benchmarks

[Fab]

Or in other words -- when data no longer fits in L2 cache in G4 then 1682M is winner?

This is why i'd be rather curious to see the result of the benchmark i suggested:
http://amigaworld.net/modules/newbb/viewtopic.php?mode=viewtopic&topic_id=35053&forum=33&start=240&viewmode=flat&order=0#650877

But this thread suddenly got very silent.

[Fab]

@fab:

btw heres your result, just seen it posted over at awnet:



so its again confirming the previous tests.

sigh, i hoped the ban will keep me off there, but it doesnt seem to work that way..

Yes. Though the time spent in I/O seems a bit too high (probably some other factor). But in the end, the X1000 is just marginally faster than the Mac mini to decode the stream (82.8s against 84.6s). I actually expected better there. I'll just wait a bit to be sure this test was also done with AltiVec (even if it wouldn't be drastically different with or without it, unlike LAME).