so i heard it is a single 6gb chip card.
i have a gtx 580 3gb single chip for now..
would it be a good idea to get it for houdini, will it help?
if you have any other knowledge about the card it would be welcome.
thanks
GTX TITAN 6GB
8411 11 1- sami.tawil
- Member
- 172 posts
- Joined: March 2012
- Offline
- malexander
- Staff
- 5218 posts
- Joined: July 2005
- Offline
The Titan is a good graphics and compute card, however it is also a limited edition card (10,000) so it might be hard to get a hold of one. It'd be a better card than the 680 or 690 because of its solid compute, loads of memory, and the fact that Houdini doesn't take advantage of dual GPUs well.
With 2688 shaders and capable of 1/3 rate FP64, the Titan is the successor to the 580. I was expecting this GPU to become the new Quadro K6000, actually.
However, the card will require a recent driver, and currently any GEForce driver later than the 306 series will cause Houdini to hang. This is going to be resolved in the 319 series, due sometime in the spring. Also, it seems that a fair number of OpenCL benchmarks being run on it by reviewers around the ‘net aren’t working on the Titan yet. Coupled with the fact that it's not available yet, it'll be awhile before you can use this card with Houdini.
Edit: I should note that the GEForce driver 310+ hang only occurs on Windows 8. Linux and Windows 7 are unaffected.
With 2688 shaders and capable of 1/3 rate FP64, the Titan is the successor to the 580. I was expecting this GPU to become the new Quadro K6000, actually.
However, the card will require a recent driver, and currently any GEForce driver later than the 306 series will cause Houdini to hang. This is going to be resolved in the 319 series, due sometime in the spring. Also, it seems that a fair number of OpenCL benchmarks being run on it by reviewers around the ‘net aren’t working on the Titan yet. Coupled with the fact that it's not available yet, it'll be awhile before you can use this card with Houdini.
Edit: I should note that the GEForce driver 310+ hang only occurs on Windows 8. Linux and Windows 7 are unaffected.
- sami.tawil
- Member
- 172 posts
- Joined: March 2012
- Offline
thanks for answering
i am on windows 7 and wont change for a while
if i can get one i think i will buy it anyway and wait for the drivers and houdini update to mount it..
i can afford that card but not a quadro k5000 which looks amazing too
also i hope it will fit in in my computer as i have a server mother board micro
i had to force the gtx580 into the slot…i hope i wont have this problem with the titan.
thank you for the info
i am on windows 7 and wont change for a while
if i can get one i think i will buy it anyway and wait for the drivers and houdini update to mount it..
i can afford that card but not a quadro k5000 which looks amazing too
also i hope it will fit in in my computer as i have a server mother board micro
i had to force the gtx580 into the slot…i hope i wont have this problem with the titan.
thank you for the info
- JuanDos
- Member
- 35 posts
- Joined: April 2012
- Offline
- sami.tawil
- Member
- 172 posts
- Joined: March 2012
- Offline
- sami.tawil
- Member
- 172 posts
- Joined: March 2012
- Offline
- malexander
- Staff
- 5218 posts
- Joined: July 2005
- Offline
It's cooler than the 580 reference design. It draws more power at load but less at idle than the 580. Here's a page from a review that details power, noise and thermals:
http://www.anandtech.com/show/6774/nvidias-geforce-gtx-titan-part-2-titans-performance-unveiled/15 [anandtech.com]
Edit: Yes, it should work with win7.
http://www.anandtech.com/show/6774/nvidias-geforce-gtx-titan-part-2-titans-performance-unveiled/15 [anandtech.com]
Edit: Yes, it should work with win7.
- igou
- Member
- 102 posts
- Joined: March 2013
- Offline
It'd be a better card than the 680 or 690 because of its solid compute, loads of memory, and the fact that Houdini doesn't take advantage of dual GPUs well.
sorry to dredge up an old post. just wonder what is “solid compute”?
I noticed that if use openCL to sim pyroFX the result will some weird. for ex, dissipation can't work good.
if card has solid compute like TITAN or GTX780, then can get the same result that CPU?
- malexander
- Staff
- 5218 posts
- Joined: July 2005
- Offline
just wonder what is “solid compute”?
What I meant was that the 480/580 cards are much better at compute than the 600 series. For the new generation, Nvidia optimized all its 600 series for graphics, removing some compute features out in order to scale up the number of ALUs (1536 from 512; they had to make the ALUs less complex). When executing code with a lot of branches, the 600 series has a bit of a handicap compared to the 580/480. And compute workloads tend to have more branching than graphics workloads.
All OpenCL programs should be executed the same way, regardless of the hardware (though with minor precision differences). So there should not be any difference in the output, just performance, between different hardware.
The Titan is closer in architecture to the 580 than the 680, so many people (myself included) expected it to perform a lot better in compute than it does.
- igou
- Member
- 102 posts
- Joined: March 2013
- Offline
- malexander
- Staff
- 5218 posts
- Joined: July 2005
- Offline
- buckhorn
- Member
- 1 posts
- Joined: May 2013
- Offline
twod
What I meant was that the 480/580 cards are much better at compute than the 600 series. For the new generation, Nvidia optimized all its 600 series for graphics, removing some compute features out in order to scale up the number of ALUs (1536 from 512; they had to make the ALUs less complex). When executing code with a lot of branches, the 600 series has a bit of a handicap compared to the 580/480. And compute workloads tend to have more branching than graphics workloads.
Interesting.
What compute features have been removed? And how are the ALU's less complex? If branches diverge within a thread group (warp), each path will be run serially on fermi or kepler architecture, and good gpu code avoids this situation whenever possible.
twod
The Titan is closer in architecture to the 580 than the 680, so many people (myself included) expected it to perform a lot better in compute than it does.
I'd argue that the Titan is much closer to the 680.
The main reason why people see reduced performance on kepler IMO is because most performant code for previous architectures reduces global memory accesses through shared memory (local memory in opencl speak), and on kepler the amount of shared memory didn't scale up with the increased core count per multiprocessor. Writing high performance code on kepler will often involve register swapping to compensate for less shared memory per core, which isn't available via opencl.
You can find plenty of examples of kepler cards outperforming fermi many times over.
-
- Quick Links