Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
Could it be a reverse marketing tactic? Like people are a bit underwhelmed but then it turns out that VRAM stacks on gamer cards too and are positively surprised?
I can't claim the credit for that. It was Outrider42
Perhaps they can write a negative review for *this* card, but will they end up being removed from Nvidia's "approved reviewers" list as a result? Nvidia has done that to other sites in the past. Will that knowledge cause current reviewers to pull their punches just so they can ensure they'll be able to review future products (which in turn drives page views).
The wierd thing about this, in my opinion, is that at the moment Nvidia seems to have the best-performing products anyway, so why go hardball like this and risk bad publicity?
...I'll wait, but sticking with my original call and not putting any money down that it will/won't happen (too poor to do so).
as to NVLink configuration here is the difference between the Quadro and GeForce series.
First image dual RTX 2080 Ti's with NVLink
Second image dual Quadros with NVLink
Notice that only one bridge is employed with the GeForce series (functioning much like an SLI bridge though with faster transfer rate) while two bridges are used on Quadro/Tesla cards (both needed to also support memory pooling). The article the first image is from only mentioned faster data transfer rates between dual 2080 Ti's with NVLink and nothing about memory stacking.
This is nothing new. NVIDIA or any company can decide not to send sample units to any reviewer if they don't like anything about that reviewer. That's always been the case. And not getting sample units to test means the reviewer needs to buy their own hardware, probably after the "approved" reviewers have already scored the big news with the early benchmarks. And how many reviewers would be willing to spend over $1,000 of their own money on a card that's old news? And if you're gonna review GPU's and CPU's and motherboards and water cooling and hard drives, etc., it gets real expensive real quick.
More on RTX: https://www.tomshardware.com/news/nvidia-tom-petersen-geforce-rtx,37727.html
Hmm, so will Iray be built around that or not. Sounds like not, at least for GeForce? What a confusing bunch of people.
...finally confirmation on memory stacking.
Maybe for the full Iray standalone accessible to pro grade software users, not sure about the version integrated into Daz. (though if someone can afford a subscription to 3DSMax or C4D, they probably can afford a pair of RTX5000s and the NVLink bridges to have 32 GB (about 5,200$ total).
If Otoy is in the position to build that function in for the 2080/2080Ti (maybe already working on it), then yeah, that would make a big difference.
looks like time to make a big bowl of popcorn and get a beer.
I thought so. The software must support it. The question then is which software will support it
With Nvlink Windows 10 may finally have an a advantage over Windows 7 because I think only wddm 2.x can stack memory and only directx 12 will have Microsoft Real Time raytracing API. However I'm sure a chunk of the stacked memory will still be unusable
Otherwise the two other choices would be using Nvidia SDK and may be Vulkan (not sure for this one)
My wish would be Vulkan, because that is cross platform and hardware agnostic, but there is a bigger probability that Nvidia API will be used in most implementations for rendering apps, including Iray
Now I'm rather curious about AMD's answer (Radeon Pro render is announced to get real time raytracing)
....hopefully that won't be the case or then for sure not going to bother as the W10 has other less than desirable issues in my book..
FYI:
Iray for C4D was discontinued in Nov, 2017. It was generally rejected as immature and not ready for production compared to the other 8 or 9 renderers available to C4D users. (Though the final straw was probably the adoption of AMD Pro-Render as the 3rd internal renderer.)
Times may have changed, but back in my early days contracting with small-medium sized businesses I worked at quite a few metal working and fab shops. They all balked at the cost of workstations and would always cheap out. If things were the same there today it's highly doubtful I could convince them to spend 5K on video cards alone for a CAD workstation.
This guy's answer is kind of ambiguous.
Sentence 1: Quadro + NVLink pooling = yes
Sentence 2: RTX + NVLink pooling = no
Sentence 3: NVLink pooling = maybe, but no GPU family defined.
So, how would The RTX series be capable of pooling if it has already be determined that the RTX family is a no-go? I would expect that this is a firmware switch that only nVidia is capable of state changes. Can the single RTX NVLink even support pooling? If so, why are there two bridges on Quadros?
It is suprising that this sort of information wasn't revealed at (or soon after) the announcement, but other important facts were ommitted as well. So, we get to keep on guessing.
BTW, it seems that the clearest performance picture can be found by examining the Titan V since the Titan is the closest relative on the family tree. I suspect that there was an algorithm change to the Tensor cores that allow for the ray tracing feature on the RTX. I wonder if the Titan V will get ray tracing?
Omen
I wonder if "developers" here emans third-party card developers, rather than software developers? It's a verys trange rider, it looks to me like one of the later sentences is left from an earlier draft of the reply.
What they talkin' 'bout, Willis?
This is the information Otoy developers shared for their estimate on Turing and NV Link support on Aug 31, 2018:
"1. We do have to recompile Octane for CUDA 10 for it to work on Turing, but unlike previous releases, NVIDIA has given us the CUDA 10 SDK and we have a V4 build on Turing already in house.
2. We should at minimum get V4 and possibly V3 updated in September before the cards launch with support. We are planning for V4 stable in September, assuming that RC3 fixes all known issues reported in RC1 and RC2 and no new issues are reported within a week or two following it's release.
3. NVLink on Turing GeForce RTX cards has not been tested by us yet, and we don't have all the details on how it works differently or is limited relative to Quadro/Tesla NV Link. We want to get our hands on the cards before we release this feature publicly, as we expect many Octane users will be looking to use it on 2080 Ti and 2080."
Source:
https://render.otoy.com/forum/ucp.php?i=pm&mode=compose&action=quotepost&p=345030
Note: RC = Release Candidate.
My understanding from watching the video interview is that RTX cards 2080 and 2080ti cards can share resources including RAM; it needs to be coded by the application. The caveat I noticed was that one shouldn't presume it will be a pool of RAM encompassing both cards that works seamlessly to the same levels of performance (for want of a better word).
Yeah, I would expect a performance hit with memory pooling.
Would be nice if we can get that in Iray with all the Turing cards, and be able to turn it on and off in the Iray render settings in Studio. Wishful thinking anyway.
...thank you for the update.
...from what I understand, the Titan V does not support NVLink. Even if it did one would be better off with two 2,300$ RTX5000s and a 39$ bridge (still not sure why the bridge for the RTX Quadro's costs less than the GeForce ones). If the Titan was able to make use of NVLink it you would need a pair of the Volta bridges, which cost 600$, making the total cost for 24 GB 6,600$ or 300$ more than a single RTX6000 and almost 2,000$ more than dual RX5000s that would give you 32 GB of pooled memory double the CUDA cores more Tensor (768) and likely more RTX cores (should they rededicate some of the Titan's Tensor cores to RTX).
All the cool youtubers get people to give them stuff free. Pretty much every video I see, the reviewer is thanking somebody for the hardware, whether it is a storefront, a manufacturer, or their own fans. And none of those are Nvidia, as 3rd party AIBs often do this. But some of these people actually do buy things with their own money. There are game reviewers who always buy their own games and outright refuse free copies (or donate the free copies to fans) on priciple. When you factor how many video games release, that is quite an expense to swallow.
The NDA only effects pre launch. Nvidia cannot stop anybody from posting reviews after launch when anybody can buy the cards off the shelf. They would have to have the most restrictive consumer EULA in history to so, where every single person buying would not be able to post a benchmark. Actually it is interesting, Intel had a really weird restrictive EULA recently, but backlash quickly changed that.
So while Nvidia can refuse to send certain reviewers units to test before launch, after the launch everything is fair game, and that is exactly what some reviwers are doing. This does mean that for some people their reviews will be later than others due to having to wait to get the cards. But for some of these people, they have enough respect from their audience that they will wait.
If Nvidia is doing anything dirty, it will come back on them one way or another. They are taking a big risk with Turing. The cost is very high, and the performance is just not that improved. You are absolutely paying for ray tracing here, but gamers are not too excited about the idea. NOBODY buys a $1200 GPU to play games at 1080p (with ray tracing). If it is doing ray tracing, they want it at 4K. Maybe 1440p at worst. But 1080p is like a joke. We might be happy, but gamers are not so excited. So Nvidia already has a problem. If they are doing things behind the scenes to restrict reviewers and whatever, that will only compound it. And then you have the simple fact that they have a monster chip, but less than half of that chip has CUDA cores on it. This opens the door for AMD and Intel to catch up, and if they can compete at all, gamers will be willing to buy from them before Nvidia because of the bridges Nvidia is burning.
Just think, AND or Intel could release a 7nm card, make it smaller than thus cheaper, and fill it with enough streaming processors to beat the 2080ti in raw performance. If they dedicate the full die to streaming processors they could do it. If Nvidia counters with 7nm, they have a problem. They cannot just abandon RT and Tensor so soon. They still have to dedicate half of the chip to these cores, and this cuts down how many CUDA cores they can place on any given die. Unless they build an expensive monster chip again, they would be faced with a disadvantage in "regular" games. In the blink of an eye, the whole GPU market can flip depending on how all things roll.
2019 and 2020 could get wild.
...well that's actually for two bridges as required for Volta cards.
NDAs (particularly the on from Nvidia) doesn't apply once what the NDA protects against, once the information is generally available. Either once the specific infor reaches the date the NDA is lifted, or if the infor is released early and becomes generally available.
Ok, still pretty expensive though.