Report: Nvidia GeForce Titan Might Outperform GTX 690

Page 4 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.
[citation][nom]photonboy[/nom]6GB of VRAM?That suggests a dual-GPU part to me (2x3GB). The GTX680 has "only" 2GB so jumping to 6GB, even for triple-monitor is a little overkill.Plus, don't Tesla cards have slightly DIFFERENT architecture to be optimized for their non-gaming task? Using the EXACT same GPU seems unlikely.I wouldn't put much stock in this info. All I take out of it is there will be a new card.[/citation]

With shared memory (at current memory speeds) graphics card memory is becoming less important.
 
[citation][nom]JonnyDough[/nom]With shared memory (at current memory speeds) graphics card memory is becoming less important.[/citation]

There is no shared memory in a modern computer for consumers that can get anything more than a small fraction of the performance offered by wide GDDR5 interfaces. Graphics card memory is getting more important, not less.
 
@JonnyDough; more like 800+ fps(if your cpu can push it).

@blazorthon:While bandwidth is certainly important to lower end graphics units like those in amd APUs, it is less of an issue on the high end. gtx660ti(with 1/2 the path to vram) has been known to perform similarly to hd7950.
 
[citation][nom]jtenorj[/nom]@JonnyDough; more like 800+ fps(if your cpu can push it).@blazorthon:While bandwidth is certainly important to lower end graphics units like those in amd APUs, it is less of an issue on the high end. gtx660ti(with 1/2 the path to vram) has been known to perform similarly to hd7950.[/citation]

Actually the Radeon 7870 is more of a parallel to the GTX 660 Ti than the Radeon 7950 and the GTX 660 Ti at least has a significant GPU performance and memory frequency advantage to partially make up for its disadvantage in memory interface width.

Furthermore, when you do something such as pump up the MSAA and/or overclock, the 660 Ti (even if it didn't have locked voltage) is left at a significant disadvantage to the 7950 due to its memory interface width disadvantage. Also, the 7950 Boost Edition is performs more on-par with the 670. The difference in memory interface width and ROP count is the only thing stopping the 660 Ti from performing exactly like a GTX 670 because it's literally a GTX 670 with a gimped memory interface and ROPs. Memory bandwidth is a very important factor in performance regardless of how high end the GPU is. It is not the only factor, but it is an important factor nonetheless.
 
[citation][nom]Kyuuketsuki[/nom]Correct me if I'm wrong, but doesn't 3D Mark 11 take advantage of GPU compute for the physics test? It also supports "compute shaders". It seems like the score could be getting some huge boosts on a couple of areas that take advantage of the compute performance, since Kepler sucks for compute.That's assuming this is even legitimate. Which is likely isn't, since rumors from anonymous sources cannot be verified.[/citation]
No. Just compare the GTX 680 to the Radeon 7980. It appears a couple of guys(or girls, I suppose) in Taiwan have the 3DMark11 record, running four GTX 680's in quad SLI.
http://www.3dmark.com/hall-of-fame/3dmark-11-top-extreme-preset/
You still have a good point though. Titan GPU should be good at brute-force code breaking attacks.
 
Yet we now know it only offers 90% of the performence of the GTX 690 at a cost of 1070 usd ! So this article was total bs from the Nvdia Marketing Department !
 
Status
Not open for further replies.