Quiet and desert ..

[YoonKyung]

Radiance, do you baking a T-bone stake with the GTX480s arrived?

[radiance]

i've been fighting to get it running,
finally have a build for compute model 2.0 and correct device manager report,
just the daylight system is broken, still looking for the issue.

this might delay our release a day or so...

Radiance

[n1k]

Can you tell us something about fermi performance now? Does new beta work faster with fermi then current version?

Cheers,
n1k

[radiance]

we're still working on it.
i can't give any propper benchmarks until the daylight is working on fermi,
we will fix that tomorrow.
once the new build is propperly finished/tested, i will post benchmark results.

Radiance

[n1k]

Cool tnx

[YoonKyung]

Can you tell us something about fermi performance now? Does new beta work faster with fermi then current version?

As far as I test with other softwares, it is 2~4 times faster than GTX280.

[pixym]

Any bench vs your gtx 295???

[radiance]

i'm getting about 2,8 megasamples/sec on the standardized benchmark scene. (the OCS one i posted)
this is with one GTX480.
however we haven't fully explored the optimizations for the fermi architecture, so this might go up.

The chessset demo scene navigates like a video game, super smooth with more than 30 FPS, it's really nice.
Tomorrow we will also report the results of 2x GTX480 with multi-gpu mode, which should be nearly double the speed.

Radiance

[andrian]

It must be really great to work that way nearly like video game ...

[pixym]

Only 2.8 Msample/s for a gtx 480 on the bench scene???
mmm not the score I was waiting for... If you take in mind one gtx 285 gives 1.9 Msample/s
That means a gtx 295 can give close to 3,7 Msample/s !

If YoonKyung can confirm this, I am going to buy two gtx 285 for my Mac Pro instead of a new PC with gtx 480!