Graphics cards and Texture memory

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 07 July 2013   #1
Graphics cards and Texture memory

Hi guys,

I'm on the verge of buying a much wanted AMD firepro w5000.

Earlier this year I made the jump to my first pro graphics on a used quadro FX5600.

Now testing it from an old school consumer 7600 GT I had before, I have seen the increase in better handling and frame rate on my heaviest scene. But due to hard drive fail, I lost all my scenes with my heaviest textures. So I can't test quickly to make a comparison with out making a whole new scene with textures!

And I would like to test for this reason... In the future I can crossfire the amd w5000 and get double the specs? but not with the memory. I'll only have 2gb not matter how many cards are in there.

So I need to run a test with the 1.5GB that the FX5600 has to see how far that take's me to give me a clue if I'll be satisfied with the 2GB the W5000 has.

Is a way to test to just quickly making some simple cube geometry in maya and adding different 4k or 8k textures in the scene and push that method to see how many I can get away with with out hick up.

And also testing polygons, Can I get away with simple sub dividing a cube as many times until the system gets to bogged whilst reading the frame rate as indication..?

Lastly can anyone here give me a recommendation on the W5000, after months of research it's seem at least twice as good as the card I'm using now.

Thanks!
 
Old 07 July 2013   #2
Crossfire is only a valid option for games DCC applications do not get any advantage from this or SLI.
2 GB of VRAM is not adequate for a new card imo.
The FX5800 was released nearly five years ago, it is a completely outdated card and already was so at the beginning of this year, i hope you haven't spend to much on it. The Fire Pro W5000 is already a year old, you might be better of getting a more modern gaming card instead.
Workstation cards usualy only have an advantage over gaming cards when used witrh certain CAD applications and specific drivers.
__________________
- www.bonkers.de -
The views expressed on this post are my personal opinions and do not represent the views of my employer.
 
Old 07 July 2013   #3
Hi Srek

So are you saying I can get the same benefits or more from consumer graphics card.

I have read the forums and know people are using consumer cards and I would to.
before i bought the fx5600 i was leaning to a consumer card with great specs but I thought go professional because it is recommended.

Can you name a gamers card applicable to my budget. for the money I could get 6gb on a consumer card is it the same experience?

Thanks!

Thanks
 
Old 07 July 2013   #4
I've had great performance from mid-range gaming graphics cards. Certainly pro-range cards would be better, but much of my work is under a million polys (arch/viz) and it's rare that my GTX560 and 460 bog down much.

The current "flagship" was just released, the K6000. It's got 12GB of RAM on it, but obviously will cost an arm and a leg.
__________________
Commodore 64 @ 1MHz
64KB RAM
1541 Floppy Drive


"Like stone we battle the wind... Beat down and strangle the rains..."
 
Old 07 July 2013   #5
I just lately built a 4770K system with a GeForce 670 and the combination is right at the top in viewport speed (CB 74) of non overclocked systems. A faster card will score a bit higher, but you have to keep in mind that when it comes to viewport speed the CPU always has a huge influence as well.
I would realy only invest some serious money in GPU power if i were to use a GPU renderer.
__________________
- www.bonkers.de -
The views expressed on this post are my personal opinions and do not represent the views of my employer.
 
Old 07 July 2013   #6
An important note: GPU rendering engines aren't the only thing using the GPU, ESPECIALLY not the ram component.
Mari is a mofo of a VRam hungry beast. Mudbox for texture painting, same thing.
Nuke's CUDA nodes, same.

I'm also told, but this is NOT first hand experience, AFX's raytracer (which is used for a few bits and bobs in there) will also show that much.

Lastly, if you offload to GPU for caching or eval (which in commercially available DCC world is mostly Maya VP2 or Maya's GPU cache), then you can also kidney shot your ram quicker than you'd think at high resolutions.

If you animate and rig or polybypoly model, the CPU will most of the time be a bottleneck way sooner than the videocard will ever get a chance to be, and hopping to a ridiculously clocked Icy or Haswell will make a lot more difference than buying a grand worth of videocard.

But if you do any of the afore mentioned things, then stocking up on available fast VRAM can make a difference, in which case I'd say OCed 670 with loads of ram > 780 > Titan > k6000 in order of bang for buck.
Yes, I am leaving the k5k out of that ladder
__________________
Come, Join the Cult http://www.cultofrig.com - Rigging from First Principles
 
Old 07 July 2013   #7
Why the 670 over the 780?
__________________
Matthew O'Neill
www.3dfluff.com
 
Old 07 July 2013   #8
I thought the 6 series geforce was crippled in maya?

Also many test are showing the amd firepro line extremely superior to everything else relatively.

Can any one suggest a test for me to test how far 1.5gb on the FX will take me...

If I can get conclusion from this, it will let me know if i really need to spend extra to get 4gb cards instead of the 2gb w5000 which is in my price bracket.

Many thanks!~
 
Old 07 July 2013   #9
Originally Posted by imashination: Why the 670 over the 780?

Placing them in order of bang for buck.
The OCed 670s with the most ram they can have around here can be had for half the price of a 780. The 780 for being DP crippled and carrrying "only" 3GB doesn't strike me as terribly good value for money right now.
__________________
Come, Join the Cult http://www.cultofrig.com - Rigging from First Principles
 
Old 07 July 2013   #10
Originally Posted by pringleman: I thought the 6 series geforce was crippled in maya?

Also many test are showing the amd firepro line extremely superior to everything else relatively.

Can any one suggest a test for me to test how far 1.5gb on the FX will take me...

If I can get conclusion from this, it will let me know if i really need to spend extra to get 4gb cards instead of the 2gb w5000 which is in my price bracket.

Many thanks!~

The 6xx geforce are NOT crippled in Maya. They are Dual Precision crippled, as are the 7xx anyway.

The 5xx were nowhere as artificially crippled (only a fraction of the hit compared to 6xx), and the Titan isn't at all.

You know how often you need double precision computation in Maya on the GPU? Never
__________________
Come, Join the Cult http://www.cultofrig.com - Rigging from First Principles
 
Old 07 July 2013   #11
Ive not checked any benchmarks on them, but wouldnt the 770 also be a good contender? Theyre £150 less than the 780 and have an extra gig of memory.
__________________
Matthew O'Neill
www.3dfluff.com
 
Old 07 July 2013   #12
I'm not sure about viewport perf, but in gpu rendering the 770 4Gb seems like the best value for money - 2 of them will outperform a titan by around 25%.
__________________
www.matinai.com
 
Old 07 July 2013   #13
Guys, I've just a had a thought,

how about the best of both world approach?

AMD w5000 mainly for maya and other progrmas that shines with that.

And then a 4gb geforce card for everything else. Like editing, Motion capture (ipisoft) GPU rendering. It will have CUDA.

Can someone shed some light on this.. This would allow me to be comfortable in my budget and also wait untill AMD's new cards and future cards are out and prices lowered.
Thanks!
 
Old 07 July 2013   #14
Originally Posted by imashination: Ive not checked any benchmarks on them, but wouldnt the 770 also be a good contender? Theyre £150 less than the 780 and have an extra gig of memory.

Are they available now? If so my omission is just not being up to speed with offers, not a statement towards the 770
@Matt Two 770s though beat little, given SLI syncs the memory, doesn't extend it (so you still have 4gb "only" instead of six), and practically nothing truly uses it other than a couple rendering engines.
__________________
Come, Join the Cult http://www.cultofrig.com - Rigging from First Principles
 
Old 07 July 2013   #15
I did qualify that with a 'gpu rendering' though sorry just me banging on about redshift again.

When I said not sure about viewport perf, I was just talking about a single 770, could have made that clearer. Sure it does the job though.

Pringleman, sorry for the offshoot, I'd say listen to these guys, a gaming card will do you fine in the viewport these days. 2 cards in the machine means you have to make sure you have good cooling/power supply too.
__________________
www.matinai.com

Last edited by mattmos : 07 July 2013 at 09:50 PM.
 
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 10:53 AM.


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