OTOY Forums

I just bought a GTX 480, and for rendering in Octane it is great, as well as for Premiere Pro CS5. Thumbs up.

However! Unfortunately, I discovered that Blender, Cinema4d and other 3d apps now suffer from hiccups, delays and generally dismal viewport performance, to the point that it's unusable. The issue lies with the new 4xxx line of consumer cards: Nvidia crippled either the hardware, or the drivers, or both. They remain silent on this issue, which only proves in my opinion that this was done to save their Quadro profits.

Now, here's my problem: if I stick with the 480, I get great gaming (which I hardly do anymore), great Octane rendering, but I cannot bear using the 3d apps (where I spend most of my time). Animating is horrible now.

If I go back to my 280 I lose the render speed.

I can use my 280 as my main card, and the 480 as a rendering card. But it just does not feel right. So I am thinking about returning the 480, and switching to ATI (whose opengl performance runs circles around the current crop of Nvidia cards), and use my 280 as a rendering option for Octane for now. I have no idea whether the 580 runs into the same problems, or not.

My question to the Octane developer team: doesn't this development on the side of Nvidia concern you? Are you aware of this problem? Suddenly the prospect of buying these cards become much less attractive, in my opinion. I refuse to get a Quadro, especially now that Nvidia is forcing the matter in this unethical way. I would rather give up on Nvidia, and get an ATI in the future, even though I lose the render option of Octane. Or wait it out for ATI to get their drivers in line with CUDA, and wait for an ATI compatible version of Octane.

It seems Nvidia is not only screwing (pardon my French here, but that is exactly the word that covers my feelings atm) their loyal customer base, but even more important, they are doing developers like Refractive no favour at all either. This may have a big impact on your product in the long run. What are your thoughts? Is this a hardware issue, or is there a way to circumvent this through a patch, or something? Nvidia remains stubbornly silent (I did send them the same question to address this issue - no answers).

It's a no-win situation. Unless you have money to burn, of course

Hi Rogue,

The problem with mixing ATI and nVidia is, that you can't run Octane on that combo (see http://www.refractivesoftware.com/forum ... 999#p39999).

What are the problems with the GTX480 in the 3D apps? I haven't experienced any problems with a GTX460 or GTX470 in CINEMA 4D yet.

Unfortunately I can't give any recommendations what to do, as I don't have much experience with different hardware, but my gut feeling is, that you probably should keep the GTX480 and use it for rendering, while you use the GTX280 for display, until the issues with the GTX480 are resolved.

Cheers,
Marcus

I would never have bought a Nvidia card if it wasn't for Octane. I had a 'lousy' Radeon 3650 before I bought the GTX 460. Guess what; navigating in Blender was smoother under the ATI card, which is something like 4 years old and low-end. Now whenever I work with '2D' content (like nodes, or just tweaking materials), and return to the 3D viewport, I get a two second lag before it starts to navigate it smoothly. It's like the card gets in 'sleep mode', whenever there is no need to draw 3D stuff for some moments. I don't know if this is the same issue, but recently I hear more and more reports that the new series of GeForces are not that good in 3D apps.

Before discovering Octane I didn't even know there were specific graphic cards for 3D apps, like Quadros (graphic cards were always the least important component for me). Back then I was told that they existed because they offer better experience. But it turns out that the real Nvidia slogan should be: "GeForce - intentionally worse experience in 3D applications"

Cinema4d seems less affected by this issue, but working on a client project this week I noticed the pauses in C4d as well (working on a heavy xfrog scene). I ran Cinebench, and low and behold: the openGL performance for both cards is almost identical: about 44 fps (280), 45fps (480). That does not make sense at all. In the past I always enjoyed much improved performance when switching to the next gen video card. Cinema4d users are relatively lucky, though.

I bought the 480 to speed up openGL in the viewport of Blender - it runs slower now than a card from two generations back. Maya, Rhino and Max users have reported identical issues. Tests have shown the pixel fill rate is incredibly slow, as well as the bandwidth. Even Wikipedia mentions these problems, which are quite common now: http://en.wikipedia.org/wiki/GeForce_40 ... L_Problems

Nvidia has issued NO comments or assistance at all, bar stating that they focused on game performance and removed optimizations for 3d apps. Which makes me (and many other 4xx owners) think it is a deliberate driver 'degrade'. Thing is, I train people in 3d apps, and usually I would always suggest a good Nvidia consumer card. Last week I advised a student against this, running through all the options, advantages and drawbacks. So that's one less Nvidia customer, and one less potential Octane client. At least, for now.

And I am not at all convinced Nvidia is willing to fix this issue - since the new 4xx line came out, many 3d creation users have been complaining and begging Nvidia to make amends, or at least issue a statement regarding these problems. Business-wise and legally they can pull one over our heads in this manner, but ethically I am done with Nvidia for now - unless they prove me and all the others with the same issues wrong before the end of this year. I, however, am not holding my breath. I already contacted my vendor, and they seem to be willing to exchange the card.

I would never have bought a Nvidia card if it wasn't for Octane. I had a 'lousy' Radeon 3650 before I bought the GTX 460. Guess what; navigating in Blender was smoother under the ATI card, which is something like 4 years old and low-end. Now whenever I work with '2D' content (like nodes, or just tweaking materials), and return to the 3D viewport, I get a two second lag before it starts to navigate it smoothly. It's like the card gets in 'sleep mode', whenever there is no need to draw 3D stuff for some moments. I don't know if this is the same issue, but recently I hear more and more reports that the new series of GeForces are not that good in 3D apps.

That only strengthens my resolve to return to an ATI card. Experiencing the same problematic behaviour in Blender. It may mean a goodbye to Nvidia (and, sadly Octane) for now. Or use the 280 for my display.

There's a post here where the Quadro drivers were hacked to work with Geforce level cards. You might want to see if it improves your performance. Of course any use of the drivers is not officially recommended by Refractive Software and the users assume all risk of trying them.

http://www.xtremesystems.org/forums/sho ... p?t=256437

I had nothing but weird system problems with my 480, and sent it back. I'm on the verge of trying the 580, and hopefully it will work out better.

adrencg wrote:I had nothing but weird system problems with my 480, and sent it back. I'm on the verge of trying the 580, and hopefully it will work out better.

What OS and 3D package are you using?

Be sure to post your experience with 580, I'm interested if the 5xx series has the same problems as 4xx.

There's a post here where the Quadro drivers were hacked to work with Geforce level cards. You might want to see if it improves your performance. Of course any use of the drivers is not officially recommended by Refractive Software and the users assume all risk of trying them.

I tried that method with my 480, and it *seemed* to work during installation. But the performance was just as bad as before. Mind, I tried several times, because I was getting desperate. But it seems the Fermi have a hardware lock, or something. Although I sent Nvidia's support an email, I got this response:

Your case is being escalated to our Level 2 Tech Support group. The Level 2 agents will have to review your case notes. As this process may take some time we ask that you be patient and a Level 2 tech will contact you as soon they can to assist you with your issue.

Well, after four days of waiting, and struggling with dead slow opengl viewports, I had to make a decision: wait it out, or opt out. I only had two more days of rma-ing the card, so I was more or less forced to return it.

I had to RMA my 480 card, and exchanged it for a less expensive ATI 5870 - which I was able to soft-mod into a FirePro v8800. Blender runs silky smooth now (about 3 times faster than the old 280gtx), and animating is sheer bliss! Other apps, like Cinema4d run MUCH faster now as well.

Unfortunately, no Octane for me atm I truly believe Nvidia has shot themselves in their foot with this. (Reading some other articles it was mentioned that Nvidia regards the 480 line as a failed product, and their intention is to end it asap.) At this point I need to work in the viewports more than I require the Octane renderer, so I will see what the future brings.

So there is no way to get CUDA running with an ati card as your main card, and a 280gtx as your secondary card? I heard some people getting Physx to work with that combo?

there is a driver cleaner utility somewhere on the net. probably will help.

however i have no experience with new GTX cards.
and i was ready to buy one just for octane but after reading your post ... i will postpone it for a while.

a kind request to radiance = please start porting octane's code to ATI stream.
There is a whole bunch of potential customers out there

there is a driver cleaner utility somewhere on the net. probably will help.

however i have no experience with new GTX cards.
and i was ready to buy one just for octane but after reading your post ... i will postpone it for a while.

Thanks, I will have a look, and see if I can get the 280 to work again in other apps.

If you intend to use the 480 for Octane rendering only (secondary card), then it's okay, I guess. But I would probably go with the newer 570/580 version. I have no idea whether Nvidia fixed the opengl viewport issues in the new 5xxx line, though - I would wait a bit for the "reports" to come in.