View Full Version : go for CUDA or more memory?

10-14-2012, 06:00 AM
I've googled around for awhile now but couldn't find a solid answer to this question.

I do a lot of work in 3ds max (rendering with vray, 1080p stereoscopic, mainly animate and light, don't do any modeling), after effects and premiere.

I'm building a new pc since mines getting outdated:

CPU: Intel Core i7 3770K 3.5ghz quad core
HDD: OCZ Agility 4 256GB solid state drive
HDD: Seagate Barracuda 2TB 7200RPM SATA3 (getting two of them for a RAID0 setup)
MB: Gigabyte B75M-D3H

There's two potential gpu's i can get:

radeon 7970
nvidia 670 OC

Both cost the same.

So my main concern is, do i go with the nvidia 670 because it has CUDA support which is enabled by the "hack". Or, do i go with the 7970 that has 3GB of memory rather than the 2gb of memory the 670 only has.

That's my only concern, very simple thing I can't find a satisfying answer to.

Some people are saying get nvidia because cuda makes a huge difference, then I read somewhere else the difference is minor.

Any help would be appreciated, thanks!

10-14-2012, 12:29 PM
ive been doing some more research and i'm beginning to think getting a cheap entry-level workstation card may be a good solution.

I discovered these two cards at a fairly affordable price with good performance, both are roughly $190.

ATI Firepro V4900
Quadro 600

from my research, apparently the v4900 is better than the 600 since it's newer. I also noticed that the firepro cards run way better in softimage and maya. But the quadro cards runs better in 3ds max than the firepro.

I'm planning to do all my modeling, lighting and rendering using V-ray in 3ds max and using softimage for animation.

I'm leaning towards the quadro 600 because i'm going to be using 3ds max a lot, it also has CUDA support for fast renders, and v-ray supports gpu rendering.

So, a new question is, how does the quadro 600 stack against the gtx 670 or the radeon 7970 even though these two cards are gaming cards?

10-14-2012, 12:44 PM
If you're not doing GPU rendering, the number of cuda cores wont make much of a difference for you. The Quadro 600 is the lowest end Quadro. If you want a Quadro that performs on par with the 670, you'll have to bump up to the 4000 or 5000. Do a search, this has been discussed countless times.

10-14-2012, 09:17 PM
Why dont you get a 670 with 4Gb of ram and get the full package?

10-15-2012, 12:59 AM
yeah, that might be a good option. it'll be about $150ish more than the regular 670 though.

i've noticed more people in this forum prefer the amd radeon 7970 over the nvidia 670.

but in the editing world, ppl prefer the nvidia 'cuda' support over the 7970.

10-22-2012, 11:05 AM
sorry to bump in the thread but i guess it's OK if it's the same subject..

anyway, I'm looking to build a new rig and having hard time deciding on the GPU.
one thing for sure- it will be a mid or high-end gaming card and not a professional one (unless you can convince me that to take a low-end professional card will be a better option).

I'm Working mostly with Maya 2012- doing everything from modeling to animation and rendering. Also Mudbox and Adobe suite- AE and Premiere mostly- currently CS5.5, and Cs6 in the future.

In the past I use to use Nvidia gaming cards all the time, because they were known to work better the Radeon cards in Maya.

But from what I've read here in the forum it seems that things are different now and Nvidia has crippled their gaming cards, and Radeon gives much better performance in Maya.

On the Other hand, Nvidia gaming card along with the "Cuda hack" has an advantage over Radeon cards in Adobe Suite, specially in AE and PPRo- which I also use a lot..

So I guess I should decide which programs are more important to me and decide..
but i was wondering maybe you have new information that can make my decision easier..


10-25-2012, 03:06 PM
sorry.. and thanx

CGTalk Moderation
10-25-2012, 03:06 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.