Geforce TITAN tested for OpenGL, OpenCL & CUDA apps

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
  04 April 2013
Geforce TITAN tested for OpenGL, OpenCL & CUDA apps

http://www.tomshardware.com/reviews...ation,3474.html

A brief overview of what to expect with GF Titan in some 3d apps. Including Maya, Max, iray, octane, blender cycles, lightwave etc...
__________________
"Any intelligent fool can make things bigger, more complex & more violent..." Einstein
 
  04 April 2013
well.. that was disappointing..., but not surprising i suppose.
__________________
-- LinkedIn Profile --
-- Blog --
-- Portfolio --
 
  04 April 2013
It mostly presents some glaring driver issues, and they don't seem to have tested that many things.
I got better results from a 680 than they post, I wouldn't be surprised if the Titan was mis-represented.

It's been reported (reliably) by several people to be absolutely smoking in some relevant tests. Mind, TH doesn't even run some of the key tests, all they do is run blender benchmarks and things like that, not even the FFT suite of tests (which against a 680 and 580 is very important as it paints a stark picture of the 6xx double precision issues which the titan is exempt from, and that the 580 also didn't suffer from).

No professional cards in the comparison either. What a horrible article when they discuss professional use, basically using it in place of quadros, and don't include a single quadro either fermi or kepler in the tests.

Some of the comparisons are LITERALLY two lines saying "oh, it does better than both other cards", like the 5 graphs underneath didn't already point that out clearly enough.

It completely misses the opportunity of discussing spec availability (how many other options do you have in the current market to obtain that much VRAM in a consumer card?) and how it's a gift from above for things like Mari or Mudbox (or even octane. redshift and so on in fact) being an enabling factor, and not just a matter of performance (try loading a texture heavy scene and render it on the GPU with 1 or 2 GB of VRAM).

Mostly rubbish, sloshing a few benchmarks around and harping on about their numbers with some patently obvious observations.

Not saying the titan is a great card, or even good value for money, just saying their article is the usual incompetent rubbish probably run and written by people who never wrote a line of cuda or produced content with the apps in question in their life.
I would still most likely buy a Titan than I would a k5000 bang for buck wise.
__________________
Come, Join the Cult http://www.cultofrig.com - Rigging from First Principles
 
  04 April 2013
I wish the maya tests would also test viewport 2.0 performance

I see that likely being a totally different result
 
  04 April 2013
Yeah, Tom's is bottom of the barrel among review sites.

I haven't done comprehensive tests, but I ran some ArionBench.

1 Titan is faster than a K5000 and a K20X put together.
__________________
Fix it in pre.
 
  04 April 2013
Originally Posted by sentry66: I wish the maya tests would also test viewport 2.0 performance

I see that likely being a totally different result


Same for Max also. Max users have multiple choices regarding renderers, but everyone has to work in the viewport. Yet they opted to base their testing on Iray...
 
  04 April 2013
Keep in mind those tests are extremely old by todays standards - I remember the GTX 2xx series even having weird issues with those scenes when I benchmarked them years ago.

If you want some OpenGL tests and VP2 tests I probably have a few scenes at home I can toss at my Titan.
__________________
My opinions do not represent those of my employer.
 
  04 April 2013
Originally Posted by BoostAbuse: Keep in mind those tests are extremely old by todays standards - I remember the GTX 2xx series even having weird issues with those scenes when I benchmarked them years ago.

If you want some OpenGL tests and VP2 tests I probably have a few scenes at home I can toss at my Titan.


in

I'm going to build a new workstation soon and I've got to make the decision between a k4000 and a Titan. Any and all information would be much appreciated
 
  04 April 2013
Originally Posted by BoostAbuse: Keep in mind those tests are extremely old by todays standards - I remember the GTX 2xx series even having weird issues with those scenes when I benchmarked them years ago.

If you want some OpenGL tests and VP2 tests I probably have a few scenes at home I can toss at my Titan.


I'ld be very interested in something real tested... especially the DX11 features now in 2014.. considering i'm about to empty my bank account on the creation suite ultimate, i've been wanting to add either a K5000 or a Titan.

are there any examples of using the .NET API with dynamics?
__________________
-- LinkedIn Profile --
-- Blog --
-- Portfolio --
 
  04 April 2013
Originally Posted by NicholasG: in

I'm going to build a new workstation soon and I've got to make the decision between a k4000 and a Titan. Any and all information would be much appreciated


I'll grab the big Turtle Barbarian guy and throw him through some of the DX11 stuff. I've been using the card for a couple weeks now at home and haven't had a single issue with it inside of Maya doing a lot of rigging and animation work.

Originally Posted by tswalk: I'ld be very interested in something real tested... especially the DX11 features now in 2014.. considering i'm about to empty my bank account on the creation suite ultimate, i've been wanting to add either a K5000 or a Titan.

are there any examples of using the .NET API with dynamics?


I don't believe there are but is there something specific you were looking for? I can see about getting an example whipped up for you with regards to the .NET API and dynamics.
__________________
My opinions do not represent those of my employer.
 
  04 April 2013
Would the titan increase speed of v2.0? Would I be able to get better framerate when animating with it? I am guessing that is more the rig/deformation calculations than the gfx card though.
 
  04 April 2013
Originally Posted by SheepFactory: Would the titan increase speed of v2.0? Would I be able to get better framerate when animating with it? I am guessing that is more the rig/deformation calculations than the gfx card though.

Depends on what you animate and how (with what rig).
90% of the time latency in the graph is the single biggest fps killer in animation, and that's inevitably the rig.

Very rarely it's something GPU dependent, unless you animate with some insanely high detail, richly textured characters, in a full environment, with a GPU accelerated skinning op.
__________________
Come, Join the Cult http://www.cultofrig.com - Rigging from First Principles
 
  04 April 2013
Originally Posted by SheepFactory: Would the titan increase speed of v2.0? Would I be able to get better framerate when animating with it? I am guessing that is more the rig/deformation calculations than the gfx card though.


Pretty much all the playback evaluation is DG dependent though the Titan would certainly benefit the following scenarios:

1) Vertex Animation Cache uses the GPU to stream the geometry instead of evaluating the DG. There are some caveats in getting this to work efficiently (I think I've detailed these before) but the Titan's 6gb of VRAM will mean you can fit more frames or more characters into that stream.

2) GPU Cache also has the same setup where it streams using the GPU and again the more VRAM you have available the higher ceiling you have as far as what you can load into the scene.

As for VP2 the Titan performs quite well as again it's got a huge amount of VRAM which means you can throw quite a lot at it (SSAO, MSAA, Tesselation, higher res textures etc etc). Most rigs are still going to be bottlenecked by the DG complexity.
__________________
My opinions do not represent those of my employer.
 
  04 April 2013
Alright hopefully these two work as an example. This is done in the Maya 2014 DX11 version of VP2 using our Sven rig from the Hyperspace Madness Game we did last year.

The Rig (so you don't think I cheat and use just FK bones):

http://content.screencast.com/users...4d/sven_rig.png

Capture of animation playback performance w/ tumble. Also showing Vertex Animation Cache and how it streams off the GPU instead of playing off the DG:

http://screencast.com/t/QQPKd7aPoN (screen recording takes a bit of a hit - playback on VAC without screen recording is about 110-120fps for me locally and peaks at around 350fps once settled in).

I have another higher resolution version I'm grabbing right now that shows off more DX11 features and higher res textures as well (4k instead of 2k).
__________________
My opinions do not represent those of my employer.

Last edited by BoostAbuse : 04 April 2013 at 10:48 PM.
 
  04 April 2013
well, there was a definite boost after hardware caching for playback. I'm curious though how much the GPU is working though at that point.. perhaps GPU-Z could show.

Originally Posted by BoostAbuse: I don't believe there are but is there something specific you were looking for? I can see about getting an example whipped up for you with regards to the .NET API and dynamics.


I was just thinking a bit abstract in that perhaps using this would make it easier to gain access to CUDA resources for dynamic particle simulations/calculations while in Maya. There's a project at codeplex for C# developers that wraps the libraries.. this may be completely moot though. I actually have not seen the .NET API documentation for Maya 2014 to learn its' capabilities yet.
__________________
-- LinkedIn Profile --
-- Blog --
-- Portfolio --
 
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 08:20 AM.


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