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
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Stoopid double post - apologies.
It's just that it sounds like nobody really knows if the Pascal technology or the new Cudas will even work. It sounds like they released a video card that hasn't even had it's Cudas or the Pascal arcitecture drivers built yet. Why else would they say that there's all sorts of testing that will have to be done after the drivers are built? Sure, they "Think" it should work, but what if they find that they just can't get it to work. What happens to all the owners of these cards then? Do they just say "oops"?
Judging by the blog post they put up earlier, it will work just fine. It seems however that iray support has been put on the back-burner behind Games/Vr support which is probably the lions share of their market, which is fair enough. If they are being sincere in their blog post, and they want to make sure the drivers/support are bullet proof when hit the public - I'm fine with that. I'd rather wait a bit longer and have something stable rather than running through several different builds until landing a stable release.
Whatever happens, migrating from maxwell at this stage is a risky move, and not one i''ll be in a rush to take. I might even wait for the next round of pascal cards next year before considering an upgrade, there should be solid support by then.
Yes, they say "we WILL support it". They don't say that the drivers are bullet proof now, they say that 2016.2 has been ANNOUNCED and should make it to their plugins SOON. My thoughts are that they have released a video card that ISN'T supported, but they expect to support eventually. I'm fine too with a card that has bullet proof drivers, but I'd prefer to see that stuff all sorted out BEFORE the cards hit the public.
I'm not migrating from Maxwell, this is my first GPU.
That's what I assume as well, that games especially are their top priority. They've been getting much better, so I imagine they'll be able to shift some focus soon if they haven't already.
For every one of us that renders with NVidia cards, there's probably 1000 or more gamers.
Still, I anxiously await - I think I'll keep my Maxwell cards for a dedicated render box, but it would be nice to use Iray preview!!!
Sorry? Where did you come up with those stats?
I read every days this conversation and i limit me to say only this.. if I knew of this problem would not have bought the videocard because i have buy it for Iray! Nvidia have put on the market a video card not complete, this is a very bad thing.
Complete guess - Nvidia and AMD sell enthusiast AIB (add in boards) by the millions - how many Daz users you think there are?
Daz users aren't the only ones using Cuda cores. Do you really think that Nvidea creates GPU's with Cuda cores and develops advanced Pascal architecture - just for Daz users :) We must be special! The problem that we're having is that that stuff isn't turned on yet - for anybody.
Me too. I'm in the same boat. It's my first GPU, so I have nothing to use in the meantime. It's like selling someone a car that's supposed to run like a charm and have all sort of power to spare, but they haven't turned on the technology that operates the fuel pump because that part of the car is still in R&D
Complete guess - Nvidia and AMD sell enthusiast AIB (add in boards) by the millions - how many Daz users you think there are?
No, but in terms of numbers, gamers are going to be where the numbers are at. There's millions of them, how many rendering hobbyists and professionals are there, irrespective of what software and render engine they are using?
I mistakenly said DAZ users as Iray is still fairly new - with my limited knowledge it doesn't seem to be nearly as popular as other render engines. I'd be curious if there were numbers out there that broke down Iray users.
It's not ridiculous to think that the numbers of active gamers utilizing Nvidia gpu's are far greater than hobbyist 3d artist. There are many GPU render engines out there, not limited to nividia, and the larger art-houses are still using cpu based render farms. I imagine the art-houses that are using GPU rendering are using TESLA or Quadro stacks. One could only asses that Hobbyist Iray users who have adopted pascal boards are probably quite few and far between - Thus, not very high in the pecking order.
I'm not trying to defend nividia here, Iray should have more resources allocated to it and it should have been supported from day one. However, it appears that Iray development has been shifted to a very small team with only a meagre online presence, so i have no doubt that those individuals are aware of the pressure they are currently under.
Nividia are well aware that this is an issue. I've seen the same topic brought up multiple times on the Allegorithmic forum, the 3ds max forum and even on nividia forums. We are not the only ones hit by this either - Octane is still having issues with pascal/cuda8, so we've all taken a hit. I'm pretty sure that if people keep the pressure on nividia to deliver, and make our voices heard, we will get results.
So let me see if I understand what you're saying. You are saying that there are "millions" of gamers using the GTX 1000 boards? And since the gamers don't require Cuda cores or Pascal architecture, that Nvidia incorporated that complicated, unfinished and very expensive technology into these cards just for the tiny segment of the market that renders 3d images? Don't you think that maybe since they advertise Pascal and Cuda technology in these cards so heavily that it must be because there are other applications for it besides gaming? Like scientists and statistical analysis and medical modelling and Geophysics and physics simulation and automotive design and parallel computing and so on and so on and so on? It is not just some tiny segment of their customer base who is waiting for them to get off their butts and finish releasing the technology.
Let's not turn this into a spat over an unresolvable side-issue please.
What is unfinished ? The card has nothing to do with your rendering application , the reason you can't render with it in iray is Iray issue not the card since it does not recognize the new architecture and the same for other programs that need to be adjusted to the new technology . It is the same with each new architecture release so not big deal , same thing was with Maxwell technology and the same now with Pascal so nothing new in this field that only waiting for all applications to be adjusted to it .
and btw scientists not use GTX cards for scientists and statistical analysis only people that can't afford Quadro cards ;) GTX are not build for this kind of work , it is for gaming in first place not rendering
New Pascal Quadro card will be released in October , until then iray will be ready for it on all platforms so professionals can profit from and that was Nvidia priority not GTX .
Look, I understand the reality is an answer you don't want to hear - I have a few Pascal Titans that are doing squat for rendering now.
It's undisputeable there are MILLIONS of PC gamers. How many rendering hobbyists and pros are there?
I never said millions of users of Pascal architecture cards. Simply a comparison of market sizes.
Squeaky wheel gets the grease.
The new Pascal Quadro P5000 and P6000 cards are coming very soon if not already availble from some resellers, so I imagine Iray Pascal support should be soon.
Pascal Iray drivers will be for the Quadro and $300/yearIray plugin customer base first, as they can be custom-tailored for each customer (part of the extra cost of Workstation cards, according to AMD at least. Surely Nvidia also follows that mantra.)
And the hardware has to be finalized before the drivers are. That's what Nvidia was saying earlier in the thread, not the other way around. You can't code final drivers or even beta drivers for alpha hardware, nor can you simply flip a switch in the Iray source for Pascal. If/when the hardware specs change, you have to rebuild the drivers. If you finalize the hardware and get it to market, then you can see what drivers your customers are going to need, as well as what features will be added to Iray that the Pascal architecture will make use of, becaue the Quadro crowd will have certain specific requests that are somewhat common across the user-base. Those features will be scaled down where applicable for the Pascal gaming cards. I'm sure there are already a few Quadro clients who have their driver requests in, and Nvidia's taking notes.
I can say for fact that DS Iray doesn't even tickle a Pascal Titan X. I'm using one as I type this and even though DS lets me select it for Rendering, the needles don't move. Meanwhile, my Titan Z and 980 are pegged. It's taking over 20 minutes to render a Dream Home (standard textures) scene which took maybe 5 minutes with the 980, Titan Z, and a 780 TI just a few days ago, and that was from a clean restart to flush the memory. Using only the Pascal, it renders a blank image in a half-second.
The new iray engine 2016.2 that support Pascal is already made by Nvidia and was ready on August 26 . If they released it already to DS programmers is a question , as DAZ stated as soon they get it from Nvidia they will work on and adding some new features to DS to support the new engine , but I am sure Nvidia want it to be ready for the new Quadro cards release first as usual, so you can forget until October to render anything in DS with Pascal cards , but I hope for you guys to get it early than that .
The SDK was not released in August, this is a common misapprehension arising from the way the nVidia post was phrased. Last I heard Daz was expecting to receive it in the middle of this month, after which they will need to do their share of the updating.
Some thoughts after reading trough the latest posts.
- It is not just Iray users who are using Nvidia cards for GPU rendering.
- Otoy has released experimental OctaneRender support for Pascal GPU based on the Cuda 8 Toolkit Release Candidate (RC) on July 15, 2016 (!).
- An inventive user now even figured out how to "update" Octanebench to detect Pascal GPU as well:
https://render.otoy.com/forum/viewtopic.php?f=9&t=56108
OctaneBench compares all GPU performance to the GTX 980. Score = 100
Based on that benchmark a stock, not overclocked, GTX 1080 scored 134.
This means if the GTX 1080 would be twice as fast as a GTX 980 it would have scored 200.
With the currently available drivers and kits we are looking at a very small render speed increase compared to the last generation.
Some people argue that with the final version of the cuda kit times could improve. Others say that there will not be a major improvement because so far the number of cuda cores is mainly responsible for the render speed.
-> Without a huge jump in cuda cores GPU rendering may not see much speed improvement...
- - -
[rant]
The underlying issue:
Is GPU rendering future proof?
The answer unfortunately currently is: No.
User should be able to purchase any Nvidia card, plug it in and use it to render with any software that uses cuda cores.
Cuda Kits and driver updates should be there to improve the performance a bit.
But the moment a new graphic card is released there should be drivers to make it detectable by all applications that rely on Cuda cores.
- - -
Imagine what the CPU rendering industry would say if Intel would release a new chip set and then it would take months until you can use it with a CPU render engine?
- - -
Yes, we are probably a small group who allready is "beta testing GPU rendering" and exactly because of that we need to bring the message across to Nvidia that we expect better in the future.
So if they really want to bring the first Volta GPU in 2017 they hopefully have it plug and render ready at launch.
If GPU rendering wants to be treated like a serious production ready competition to CPU rendering then the companies involved better start operating accordingly.
[/rant]
I have a GTX 1080 & feel your pain although its rather different for me since I bought it knowing that Iray was not yet supported & that it would put me back to CPU rendering for a while
The more important question for me is when Daz can move forward with the Pascal updates is there anything we as GTX owners can do to help test these updates & get them out quicker so we can close this thread and get on with some GPU rendering again.
Looking on the brightside I feel far less motivated to buy all the shiny new stuff that I really don't need when I can't render it in an acceptable time for me
Where do we sign on to be beta testers.
I UPGRAYYED a system to house my Maxwell cards so that I can continue to render. Rendering with CPU is way too slow. It's a bit of a PITA now that my primary system can't render or preview with Iray, but I do have a dedicated render box now I can render all the time with.
I can compare the Maxwell Titan to the Pascal Titan.
I totally agree with you , sadly it is not possible to create just a driver that will fix all application not even one in this case , iray update is not about only the engine but also Cuda and Optix that are 2 independent sections . The program need to be adjusted not just a driver , it would be too beautiful but each time the architecture change we have the same issue Octane included .
I wait for the final official benchmarks for iray to see the results , there is a huge community of people that render using GTX cards that nobody listen to since forever and nothing will change it now , Companies have own agendas and they care less if you render with GTX or not since it is not made for rendering in first place so not a priority to focus on , it never was .
I hope I see the day that we have 1 card for everything ...with super speed
I know everyone wants Pascal support in Iray but shouldn't we also want system memory assist in Daz Iray as well. I sometimes go over the limit in video memory and get bumped to cpu rendering and this sucks!!!! I have 64gigs of ram I can afford to give up some ram to the video cards to increase their abilty to hold textures. If this was a configurable option more people could use Iray. Please Daz System Memory augmentation for Daz Iray.
https://developer.blender.org/T48544
From what I'm seeing in the attached thread from blender.org, Blender is way ahead of Daz3d in getting these Pascal cuda's working :) (Read the more recent posts ie. - last week
Also,
https://lionrender.com/2016/06/16/top-gpu-cards-performance-comparison-in-blender-gtx1080-titan-x-and-gtx980ti/
And from these tests by Lion Render, the GTX 1080 leaves the Titan X in the dust! You need 4 suped up Titan X boards just to touch the coat tails of a single GTX 1000 board with Pascal working :) (exponential factoring if you consider multiple instances of the cards). Check out the "Fishy Cat" results - The GTX 1080 board took 12 minutes and the Titan X took 48 minutes! And this is with experimental none final release drivers! Even the GTX980ti is miles behind the GTX 1080. I pity those who invested their money into 3 or 4 of those lame cards when they could have gotten the real thing if they'd only waited. Now if Daz Studio's engineers could implement the necessary drivers to turn on the GTX 1000's too ... :)
Oh yeah, I'm using Windows 10 by the way, since that makes a difference here...
Blender rendered always slow with Titan X in first place so don't get excited . I am waiting for official Iray benchmarks .
And definitely there will be not 4 times the speed of Titan X in iray with 1080 as Titan X render 3 times faster in iray than in blender .
Those render times look a little suspect, something can't be right for a 980ti to outperform a Titan X (assume Maxwell) by that much. Should be really close to even.
Very fishy, i too would wait until iray benchmarks come down the line, Cycles and iray are two very different animals. The R9 result is what i find the most jarring - it's a budget gaming card. It's a very good budget gaming card, don't get me wrong, but this whole benchmark just feels backwards. I'm not married to any card or platform - If the iray benchmarks have the 1080's way ahead, i'll pick up a few, but a cycles benchmark doesn't translate to iray very well.
I've read benchmarks of scientists calculations, using CUDA 8 on titan pascal. I can't find the link. It was something like 30% faster than Titan maxwell. I did some linear estimation from those bench: From 980Ti to Pascal Titan, it gives 70% improvments. This was close to what I estimate at the beginning. So, when it comes to Iray, I think that at stock speed, Titan pascal will be 30% to 50% faster than Titan maxwell. The bonus is that you can OC way more on pascal than maxwell. So, to be short, a reasonable 50% gain is already a good progress for me (you can achive the same speed with 2 titan P than wth 3 titan M ;-) )