It’s actually the first thread I participate in on this forum so I’d really hate to see this turn into a Nvidia hate/fan discussuion
I think OP intention (which is also mine interest) was to bring some light on an issue that would concern a specific scenario - that is rendering scenes that take vram load in those questionable 3.5-4GB
I was like few days away from buying a new GPU primary for getting into learning CG and blender (I don’t really do games) and I’m holding it since this came out.
It’s expected for most of review sites to do some benchmarks that would also test that issue, but since there are members of this forum that already own the cards it would be great if they could do a Cycles bench and share them with the others here
maybe I’m wrong since I’m not really familiar with Blender but I think we need a comparison of rendering times below 3.5gb vram and 3.5 to 4gb vram scenes. Maybe same bench on 980s so we could compare them too.
Isn’t that exactly what you said in your last post?
That’s true, but as far as I can tell the specs that were incorrect (L2 cache and number of ROPs) were never listed on their website so there isn’t really anything to change. The review sites are the only places you ever see those reported.
I wouldn’t be surprised to see some official press release from Nvidia in the next few days, but if you really do have a Senior VP on the phone with tech sites doing damage control on a Sunday night I’m also not surprised that the rest of the system hasn’t caught up yet.
They clearly list the memory bandwidth on there. It says 4GB at 224GB/s, not 3.5GB at 224GB/s + 0.5GB you don’t want to touch. The fact that there’s apparently also a couple of ROP less here is insignificant.
A Blender novice with a GTX 970 here. I was wondering how much real world impact this will have? Has anyone done any benchmarking with this for cycles? As an additional question: What does it take to get to 3.5GB of VRAM? Is it based on the complexity of the scene or tile size when rendering or a whole bunch of stuff that I don’t know about?
They don’t. Advanced Microchip Devices, a California-based company with quite a long history, also makes
graphic cards (and microprocessors also) and competes with Nvidia in that market segment. You might have
heard of Radeon GPUs, the’re quite the brandname, often striking the sweet-spot
in price-performance ratio, which means they don’t sell that bad at all.
here goes another Green VS Red battle… I thought something more useful could come out from this discussion…
you know all review sites are doing game benchmarks, and they are underway to start doing tests on the issue here’s one
but yet those are game benchmarks, specific variables are in play there (drivers etc)
But from a purely practical standpoint, this doesn’t really change anything for the end user. The GeForce GTX 970 remains one of the best graphics card buys on the market. It performs the same way it did at launch — which is really good. As such, we will continue to recommend it until there is a better-performing option for the price.We can empathize with buyers who feel betrayed, though. Nvidia definitely has some mind-share to earn back. But to us the price/performance ratio trumps everything else, and that is no different today than it has been since the GeForce GTX 970 was released.
But Nvidia and Intel are better so much more superior 400 dollars Intel CPUs have better price to performance ratios than AMDs 150 dollar CPUs and Nvidias cards have so much proprietary software there is little reason to use AMD cards Nvidia cards also support OpenCL better than AMD cards and are even faster since the 9xx cards. I actually felt bad for AMD when Nvidia was releasing the 9xx cards my mouth kept hitting the floor but then I was thinking how far behind AMD was a few tears shed for them
Also I own 2 970s I am rendering scenes the 780 cant all because of that little bit of memory xD
For the $ it is still the best deal currently. The card runs well renders fast and is silent like a lam.
However stating wrong facts about VRAM amount,speed, L2 cache, and also ROPs. To excuse this by a misunderstanding between marketing and engineering is rather silly. So the engineers delivered by accident the wrong data to the PR team?
So either each team is incompetent or that mistake was signed off on.
I am curious if this also impacts the CUDA cores now and if the number here is wrong.
Well, the FTC only very recently started clearly enforcing that if a phone plan says “unlimited” you aren’t allowed to throttle it. So I guess if you make a big deal out of it, they might get around to cracking down on Nvidia in 4 or 5 years. (don’t hold your breath)
It’s still a decent deal, but if they are honest they should be up-front about this.
That said I’m still on a 480 so if anyone is upset and wants to sell theirs to me at a discount, PM me plz
nsone the 780ti is faster the even the 980 with cycles. if its cheaper as well you should go for it unless you are going to use it for gaming as well. i’m not sure the 7XX series is going to get the dx12 update when windows 10 releases.
thank you all!
I need a gpu for rendering purposes only, don’t really do games, so it’s primary for AE, Premiere and getting into Blender
Been saving for a while and when the 970 were released I was really excited on the price/performance they (seem to) offer, but than all this came out and got me really confused…
I’m sorry if I bother you with few more questions, but I’m totally new and eager to get into this and don’t want to mess this up since it’s like ~300€ I’m willing to spend and in my case it kind of a major investment
Since all benchmarks/reviews we see around are for games, the only GPGPU benchmark I found so far that included Blender is this one from sweclockers
My primary concern is whether this VRAM/bandwidth issue with the 970 would effect real-work scenarios
from the benches listed here, those that are done on single GPUs seem to place the 970 ~1min above the 780 which looks nice but I was wandering if the “depends on scene” part refers to scenes using +3.5GB vram and how much concern do I need to put into it
used GPUs are kind of hard to find right now, since it looks 7xx still perform quite well, and most of all cause it’s a small market here.
so @rdo3 that’s why I probably won’t be able to find 780ti anytime soon, or at least not before new AMD 3xx | GM200/titan comes out
my options right now are
970 ~300€
760 ~150€
750ti ~ 120€
or wait
what would you do?
