Page 1 of 3
GTX 580 benchmark
Posted: Sun Dec 19, 2010 11:53 am
by t0m4sk0
Could someone put here 580 bench. ?
Re: GTX 580 benchmark
Posted: Sun Dec 19, 2010 12:11 pm
by t0m4sk0
In every review I read that quadro cards has better performance in GPGPU applications (DP perf.). What about octane? Maybe would be helpfull if somebody could make some perf. summary results with at least 450, 460, 470, 480, 570, 580 and some quadro or tesla cards.
Re: GTX 580 benchmark
Posted: Tue Dec 21, 2010 9:27 pm
by ifusion
Hey t0m4sk0,
I just ran the benchmark file in the demo and reached 16000/16000 at 13:20
My basic specs are:
(1) GeForce GTX 580 w/ 1.5gb RAM, 512 CUDA cores, 772mhz graphics clock, 1544mhz processor clock, 2ghz memory clock
Intel 980x (6-core, 3.33ghz) OC'd to 4ghz
12gb DDR3 1800 RAM
Hope that helps. Let me know if there are any other tests you would like.
Re: GTX 580 benchmark
Posted: Thu Dec 23, 2010 2:05 am
by BlueBread
Hello, ifusion
I was wondering if you tested your GTX 580 with 3d applications such as 3DS Max? This is because I would really appreciate if you could inform me if it can handle the view-port with sufficient fps whilst modelling high polygon meshes and executing general video-editing. I am aware that the previous models GTX 400 are performing poorly within 3DS Max's view-port. Therefore, I am stuck in deciding which GPU to purchase for my next modelling rig.
My last option would be a Quadro 4000 which is expensive considering my current income.
I am also planning on investing into a copy of Octane Renderer my self once I have successfully purchased and assembled all my computer components.
Thnx a lot, Jan.
Re: GTX 580 benchmark
Posted: Thu Dec 23, 2010 4:56 am
by radiance
I think there won't be a difference between a GTX580 and GTX480, as the modeling handicap is probably implemented in the geforce driver, not in hardware.
Therefore i think only a quadro card will do the job nicely.
Radiance
Re: GTX 580 benchmark
Posted: Thu Dec 23, 2010 9:27 am
by t0m4sk0
Thnx for replays
I will make comparison to GTX 470, if number of CUDA cores and their frequency is only one thing which make performance.
Re: GTX 580 benchmark
Posted: Thu Dec 23, 2010 12:53 pm
by BlueBread
I see so it was ment intentionally, now i know what i wanted to know thanks Radiance

Re: GTX 580 benchmark
Posted: Thu Dec 23, 2010 8:22 pm
by colorlabs
ifusion wrote:Hey t0m4sk0,
I just ran the benchmark file in the demo and reached 16000/16000 at 13:20
My basic specs are:
(1) GeForce GTX 580 w/ 1.5gb RAM, 512 CUDA cores, 772mhz graphics clock, 1544mhz processor clock, 2ghz memory clock
Intel 980x (6-core, 3.33ghz) OC'd to 4ghz
12gb DDR3 1800 RAM
Hope that helps. Let me know if there are any other tests you would like.
ifusion, could you post how many Megasamples/sec you're getting?
Re: GTX 580 benchmark
Posted: Sat Dec 25, 2010 5:10 am
by BlueBread
Hello Radiance, again. :~)
Do you happen to know if the GTX 285/295 are also intentionally crippled for professional use?
Re: GTX 580 benchmark
Posted: Sat Dec 25, 2010 11:06 pm
by metalliandy
Hey guys,
I have been going back and forth with nvidia about the speed issues of the Fermi based cards over the past month or so and its an opengl speed issue with the 4xx architecture only, which may or may not be fixed in with a new driver.
Apparently its an issue with double sided lighting.
Here is a snippet of the chat
Support: so some bad news, i just went through some reported issues in our database and it looks like the low performance for apps such as Blender, Cinema 4D and Maya all have the same issues due to it being a hardware issue on the card itself, and something to do with 2 sided lighting
Andy Davies: ahh
Andy Davies: is there anything to be done?
Support: so if they are able to find a way to resolve this on a driver side which im not sure if they will or not, so for now there isnt a way to resolve this, unless you were to use a Quadro based graphics card
Andy Davies: hmm
Andy Davies: isnt quadro the same architecture?
Andy Davies: just with better drivers etc.
Support: it is yes, but for some reason it doesn't have this same hardware limitation that the 400 series card has, not sure why specifically, its a little over my head
Andy Davies: ah ok
Andy Davies: so i just wait for a driver update then?
Support: this is only present on our Fermi 400 series cards and was not present on previous series cards too
Support: Pretty much all i can suggest, that or use a different card in the mean time
Andy Davies: is it present on the 500 series?
Support: I didn't see anything mentioning the 500 series in our DB but i can only assume it is since based off of the same reference design as far as i know
Support: since blender is an open source application it is something that could be addressed on their side to from what i read
Andy Davies: do you have anything that i could have that explains the issue on a more technical basis?
Andy Davies: problem with 2 sided lighting isnt very clear for a dev.
Support: all i can find is references to 2 sided lighting nothing as to way this is the problem technically or specifically
Hope that helps