OTOY Forums

yes pathtracing is the full blown physically accurate renderer.
Directlighting is a simplified version and intended for setup / fast previewing.

As you can see here how good is ATi on OpenCL ... http://www.anandtech.com/show/2977/nvid ... he-wait-/6 also there is no OpenCL or Stream for OSX and the linux support is poor. The problem is: nvidia has the maturest software for GPU computing, even ATi has more compute power is a lot harder to optimize software on their vectorial architecture chip. The vectorial architecture is more complex and mush more harder for developers to make software so they chose nvidia's scalar architecture because is more easy to do things . Also there are some complains about stability on ATi drivers in Stream about video editing and image quality. http://www.pcper.com/article.php?aid=74 ... pert&pid=7 and http://forums.amd.com/amdlive/messagevi ... TMP=Linear , I guess this may affect octane too . I don't want to bring ATi vs nvidia war here (at least here) but i wanna answer some of your questions about this subject . BTW I'v tried again the benchmark with the right settings and I got only a few samples more on my hackintosh.

just tested benchmark in demo suite for 2.2, it rendered 8.84 Megasamples/Sec on GTX470

demo suite 2.2 and Octane 1.022 beta
470 GTX:
(pathtracing) = 2.65
(directlighting) =8.9

no overclocking. I am a little disappointed I thought I'd be around 3.8 - 3.9
Any idea why, and how to improve if possible?

Maryus3D wrote:As you can see here how good is ATi on OpenCL ... http://www.anandtech.com/show/2977/nvid ... he-wait-/6 also there is no OpenCL or Stream for OSX and the linux support is poor. The problem is: nvidia has the maturest software for GPU computing, even ATi has more compute power is a lot harder to optimize software on their vectorial architecture chip. The vectorial architecture is more complex and mush more harder for developers to make software so they chose nvidia's scalar architecture because is more easy to do things . Also there are some complains about stability on ATi drivers in Stream about video editing and image quality. http://www.pcper.com/article.php?aid=74 ... pert&pid=7 and http://forums.amd.com/amdlive/messagevi ... TMP=Linear , I guess this may affect octane too . I don't want to bring ATi vs nvidia war here (at least here) but i wanna answer some of your questions about this subject . BTW I'v tried again the benchmark with the right settings and I got only a few samples more on my hackintosh.

Still, there is something going on already http://www.luxrender.net/forum/viewforum.php?f=34 of course its far from usable but it works;)
im stuck between selling my ati gpu and buying nvidia but they are worse in price/power + drain a lot of electricity..

Are there any benchmarks ? with 9800GT or 250, 260,275 ?
cheers

gtx465 2.7 MS/sec reached 4000 samples in 14 min

Is that right for a GTX465? 2.7 path?
I am a little confused by the results reported.
What is a 470 producing with the latest beta and benchmark then? 3.3??
I notice the lighting direction has changed a little from earlier files.
A couple of posts up a 470 is reported as 2.65...and elsewhere it is mentioned at 3 or so...
I'd like to be clear about the relative performance gain before I buy a new card.
Can someone do a fresh 470 test to clarify it please.
Thanks

For reference my GTS 250 1gb gives:- 0.61 path and 2.50 direct from the .ocs which is similar to the another 250 report here.

in fact it's more a gtx470, unlocked cuda cores and bios flashed with gtx470 specs.

Ah! ok thanks so a 470 probably is about 2.65-2.7

That makes it roughly 4.4 times faster than my GTS250. Great!

And that reduces a 22 min render presently to 5 mins.

GTX 470
pathtracing: 2.69
directlighting: 8.84

no overclocking. I am a little disappointed I thought I'd be around 3.8 - 3.9
Any idea why, and how to improve if possible?

in Preview kernel change maxdepth from 8 to 4