GTX 680 2mb vs. GTX 580 3mb for this software?

Become a member of the CGSociety

Connect, Share, and Learn with our Large Growing CG Art Community. It's Free!

THREAD CLOSED
 
Thread Tools Search this Thread Display Modes
Old 05 May 2012   #16
To play what games? Because at this point there is nothing in our industry that quite benefits from SLI in general, let alone narrowing and splitting VRAM access, even with brilliant hardware based synchronicity

Get a 580, deal with the poor Maya double shading issues, and call it a day for now, IMO. x90s, SLI, and other narrower but higher multiplier solutions are more pain than anything right now, when even used at all.
__________________
Come, Join the Cult http://www.cultofrig.com - Rigging from First Principles
 
Old 05 May 2012   #17
Originally Posted by Ian151: Well, you should be able to have 2 690 cards with 8GB - 4x2GB


Each core gets 2GB and everything has to be loaded into the memory for each one. The memory isn't distributed.
__________________
The Z-Axis
 
Old 05 May 2012   #18
Dude, you get 4GB total for the card, the 2 cores with 2GB each = 4GB, and if you have two 690's, you could have 8GB. Here's a review - http://www.youtube.com/watch?v=Wy36lRyy1pI&feature=plcp
 
Old 05 May 2012   #19
Originally Posted by Ian151: Dude, you get 4GB total for the card, the 2 cores with 2GB each = 4GB, and if you have two 690's, you could have 8GB. Here's a review - http://www.youtube.com/watch?v=Wy36lRyy1pI&feature=plcp

The 690 runs the two cards separately, so yes, there are 4GB of ram attached to one slot, but no, that doesn't mean you get 4GB of potential contiguous footprint, each of the two GPUs only has access to 2GB.

It's literally two 680 in SLI on a single slot, nothing more. Except they are also ever so slightly under clocked.

So, as far as we're concerned in DCC, the 690 is worse than a 680, and effectively gives you only 2GB of ram for normal use, and in any GPU rendering supporting SLI, all your textures will have to be loaded twice, with consequential synchronicity costs, race condition prevention, bus busyness etc.

Putting two of those together, would still only really give you 2GB of actual usable footprint. So all you would be doing would be wasting 4 times the money instead of double the waste

These are cards meant for large monitors that take dual link, stereoscopic game rendering, or immersive triple monitor setups, where the redundancy of data would be required anyway.
For anybody else, including gamers running games on a single monitor, it's a huge waste of money.
A dual 690 would be the apex of pointlessness, on par with a gold plated Ferrari.
__________________
Come, Join the Cult http://www.cultofrig.com - Rigging from First Principles

Last edited by ThE_JacO : 05 May 2012 at 05:48 AM.
 
Old 05 May 2012   #20
Well, that sucks....nice false advertising - http://www.newegg.com/Product/Produ...N82E16814130781 I assumed when you do SLI, you get double the greatness...oh well, spend the money on something else I guess, like a nice Canon T4i
 
Old 05 May 2012   #21
If you ever plan to use this card for computing or 3d rendering scenes with renderers like povray and the like - make sure you check some benchmarks first.

Nvidia recently started to "artificially" cripple their geforce cards. They want you to buy the quadro / tesla stuff. The AMD/ATI Cards beat the geforce cards in most of the tests sometimes even by 200 to 300 or more percent.

http://www.tomshardware.com/reviews...rk,3193-12.html

I've been quite loyal to geforce / nvidia so far but it's obviously time to switch.
 
Old 05 May 2012   #22
Originally Posted by lehmi: If you ever plan to use this card for computing or 3d rendering scenes with renderers like povray and the like - make sure you check some benchmarks first.

Nvidia recently started to "artificially" cripple their geforce cards. They want you to buy the quadro / tesla stuff. The AMD/ATI Cards beat the geforce cards in most of the tests sometimes even by 200 to 300 or more percent.

http://www.tomshardware.com/reviews...rk,3193-12.html

I've been quite loyal to geforce / nvidia so far but it's obviously time to switch.


Do those benchmarks reflect performance in GPU renderers? It seems it was just an OpenCL test where they were performing poorly.
__________________
The Z-Axis
 
Old 05 May 2012   #23
Also, as far i know, Modo and Maxwell dont render using the GPU, not even their preview renderers.
 
Old 05 May 2012   #24
Originally Posted by shokan: Right, the guy from The Foundry told me that if the specific card isn't on their list doesn't mean it wouldn't work... it only means they haven't put it through their testing procedures. It would be an expensive risk I'm not willing to take. I'll be getting the GTX 580 3GB probably; I can always wait until the GTX 680 4GB becomes easy to get and meanwhile read about its use with Mari from people who have tried it. Meaning, the 2GB version has been around for a while now and it still can't be kept in stock with the well-known N.American online retailers. It will probably be the same for the 4GB version. Some European companies have already released 4 GB ones (Gainward, Palit), and I believe EVGA will be out with theirs very soon.


Hi

We are testing the 680 at the moment. We haven't found any problems and should have a decision very soon about supporting it. If you can hold off a week or so, we should be able to say yes/no.

Jack
 
Old 05 May 2012   #25
Originally Posted by forelle: Hi

We are testing the 680 at the moment. We haven't found any problems and should have a decision very soon about supporting it. If you can hold off a week or so, we should be able to say yes/no.

Jack


Good to know. Thanks.

I am no big tech guy, but I keep reading about Nvidia having "crippled" capabilities in the GTX 680 that would make it a less attractive card for 3D content-creators, so that 3D CC folks will opt for their more expensive Nvidia Tesla cards. Foundry Yay or Nay would certainly help in the decision process.
 
Old 05 May 2012   #26
Originally Posted by shokan: Good to know. Thanks.

I am no big tech guy, but I keep reading about Nvidia having "crippled" capabilities in the GTX 680 that would make it a less attractive card for 3D content-creators, so that 3D CC folks will opt for their more expensive Nvidia Tesla cards. Foundry Yay or Nay would certainly help in the decision process.



I'll post back here as soon as we make the decision. It seems to be a pretty good card at the moment, but Mari is written in pure OpenGL without Cuda or OpenCL. If they have crippled the pure compute functionality we shouldn't see any problems with it.
 
Old 05 May 2012   #27
Quote: I am no big tech guy, but I keep reading about Nvidia having "crippled" capabilities in the GTX 680 that would make it a less attractive card for 3D content-creators, so that 3D CC folks will opt for their more expensive Nvidia Tesla cards. Foundry Yay or Nay would certainly help in the decision process.


from what i heard its not just the gtx 680 but i think its just the geforce cards in general
__________________
Heights of great men reached and kept were not attained by sudden flight, but they while their companions slept were toiling upwards in the night.
 
Old 05 May 2012   #28
Originally Posted by shokan: Good to know. Thanks.

I am no big tech guy, but I keep reading about Nvidia having "crippled" capabilities in the GTX 680 that would make it a less attractive card for 3D content-creators, so that 3D CC folks will opt for their more expensive Nvidia Tesla cards. Foundry Yay or Nay would certainly help in the decision process.


3D content creation as in modeling/working with faces/polys should be fine.
Its when you are going to use gpu computing (OpenCL, Cuda, Direct computing, ...), which is used for things like physics or gpu path tracing or decoding/encoding, that the cards are crippled. Well, not the cards themselves but the drivers are crippled that way (which makes it even worse/inacceptable imho).

Last edited by CHRiTTeR : 05 May 2012 at 04:16 PM.
 
Old 05 May 2012   #29
Originally Posted by CHRiTTeR: 3D content creation as in modeling/working with faces/polys should be fine.
Its when you are going to use gpu computing (OpenCL, Cuda, Direct computing, ...), which is used for things like physics or gpu path tracing or decoding/encoding, that the cards are crippled. Well, not the cards themselves but the drivers are crippled that way (which makes it even worse/inacceptable imho).


Would the GTX 680 performance in the areas you mention (OpenCL, Cuda, Direct Computing...) be less or more than GTX 580? This is for currently available GTX 680 2GB vs GTX 580 3GB. Same question for GTX 680 4GB vs GTX 580 3GB.

Thanks.
 
Old 05 May 2012   #30
Originally Posted by shokan: Would the GTX 680 performance in the areas you mention (OpenCL, Cuda, Direct Computing...) be less or more than GTX 580? This is for currently available GTX 680 2GB vs GTX 580 3GB. Same question for GTX 680 4GB vs GTX 580 3GB.

Thanks.


I dont know, but i also would like to know so.
 
Thread Closed share thread



Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
CGSociety
Society of Digital Artists
www.cgsociety.org

Powered by vBulletin
Copyright 2000 - 2006,
Jelsoft Enterprises Ltd.
Minimize Ads
Forum Jump
Miscellaneous

All times are GMT. The time now is 10:19 PM.


Powered by vBulletin
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.