GTX 560 TI 2gb, benckmark test

BlueBread · Fri Feb 25, 2011 10:01 pm

Sorry double Post

BlueBread · Fri Feb 25, 2011 10:05 pm

I got my new pc running with a gtx 560 2GB, i've downloaded the demo and the benchmark. However, it states that the benchmark scene i downloaded seems to be invalid and not accepted by octane. what should i do? Specifically it states invalid scene file VERSION...

radiance · Fri Feb 25, 2011 10:51 pm

Hey,

The benchmark scene should work fine with the demo, provided you downloaded both from our downloads page...

Radiance

BlueBread · Fri Feb 25, 2011 11:44 pm

It didn't work because I was downloading the bench scene from the forum ( Outdated Scene ).

Uploaded with ImageShack.us

Uploaded with ImageShack.us

Not all cores are being used

, however glad i could help. Also, promised I would produce these results anyway

.

Sat Feb 26, 2011 11:04 am

Ahhh.... it's 6.86 megasampels/sec for the 560Ti against 5.7-5.9 megasamples for the 460 in direct lighting.That sounds more like it.The extra speed is scaled correctly and is accounted for by the 48 extra cores and increased 100-200mhz GPU speed.

BlueBread · Sat Feb 26, 2011 1:54 pm

Hi, Jaberwocky dancer.

The problem is that it isn't utilizing all its 384 cores if i'm right, Since 256 are being put to use. Therefore I believe that when the full potential is utilized we can expect a raw 3.2 mega-samples on path-tracing during the execution of the bench test.

If anyone is interested I am currently running:

I5 2500 3.3 GHZ
Palit GTX 560 TI
P67 Gigabyte UD4
2 Caviar black 640GB Raid0
8GB Ripjawsx 1600mhz, latency 9.
In a Haf 922

Sat Feb 26, 2011 2:37 pm

I must did something wrong, because the bench march test in my machine doesn't render with any reflection. This could explain the high megasamples. I have to retry this again.

Sat Feb 26, 2011 2:49 pm

Ok this is embarrasing, but what I suspected was right, I accidently (for lack of knowledge) pressed the colored clay render (probably from fiddling with the control while testing it and before loading the benchmark scene). Sorry for the confusion

. Will post my screen grabs later.

Sat Feb 26, 2011 3:15 pm

here are my screen grabs. My question is, how come it's much slower than what bluebread is getting since all 384 cores are running? and when I am using both cards it only show 384 cores and not the additional cores of the second card but jumping a significant number in terms of megasamples.

Sat Feb 26, 2011 10:47 pm

I think the problem with the speed on the 560TI will be related to the speed problems on the 460.That is the GPU design Looks like the 560TI uses the same number of Warp schedulers as the 460 (EG 48). the GF114 GPU in the 560 is a mod of the GF104 GPU found in the 460, only with the missing block of Cuda cores now enabled.I think there is a Data Bottleneck.It my be interesting to compare it to the 570 which uses a different GPu design with More Warp Schedulers.