Adding to Cart…
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.You currently have no notifications.
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2025 Daz Productions Inc. All Rights Reserved.
Comments
You know what will be worthy Titan X in next 10 years ? nada... you will be again behind on the end of your system life with old graohic card lol
you will get better oerfomance for the money if you use the 970 for the DS Open GL and monitor and left the other for rendering only , that's for sure but still we need to wait for the iray benchmarks to see how well it render and if worth the money
Heat and volts is what kills with overclocking. Even with liquid nitrogen cooling there is a limit to how high you can clock a processor before it goes pop. That's because the faster cycles you go, the more volts you need to retain stability. And with high enough volts, even with lots of cooling, eventually the dialectric resistance of the basic substrate will be overcome. Indeed, you can easily kill a chip just by running it at high volts - no need to increase its cycles per second.
The levels of overclocking that folks like Cath and myself are doing are really quite low and unexciting in the greater scheme of things; not much different than manufacturers already do with their own hardware. At these levels the reduction in life expectency is theoretical. If a CPU or GPU lasts 25 years instead of 50 years, who cares apart from museum curators?
That being said, if you're uncomfortable with it - just don't do it. And if you do overclock and have software problems, retest the issue with the hardware at normal speed before bothering DAZ with it.
if you have the cash, a dual core Xeon (2 x 22 cores) would be awesome; the price is also awesome, as just the Xeons are about $4,000 each.
Although it looks like 10 series cards will be able to share memory, how soon, or if that will include rendering on the consumer versions i'm not going to speculate on.
Yeah exactly [ overclocking ] should be called Tuning today , you see I run my PC almost all day every day so no matter if I do or not it will never last more than 5 years as the Mobo will be the first to go on itself self before the Processor . My Titan SC don;t even have higher voltage and can run at 1400Mhz using 11.6 V where 12V is the max before you have to turn the volume up compared to much lower standard and still covered for 3 years . Beside this you still need to be lucky as things give up in the most not expected moment like my last 2 Mobos , one after 6 months and the second after 2 days .. just because it was faulty , but as long you dont run on Pentium 4 you should consider better speed , that why the new stuff is made for ;)
here is a nice article about overclocking myths ;)
http://www.lavalys.com/community/blog/2010/04/top-7-myths-about-overclocking/
Then wait. On the 10th is the release of the 1070 GTX cards. Within a month, we'll have 3rd Party non-Founders cards on the market (we're already seeing leaks of prototypes, like the ASUS strix linked earlier) and it will likely be AT LEAST 3 months before we see Iray compatibility in DS for the 1000-series cards (a month for nVidia to get it in Iray/drivers, and 2 months for DAZ to get it in an update.)
If you are wanting an upgrade/new system you can render with NOW, the 1000 series is NOT for you. If you can wait a few months, then it will be a good boost to go with the 1000 series.
Don't expect 1080Ti cards for at least 6-9 months. And a 'Titan-P' (p for pascal!) I wouldn't expect until at least the the end of the year, possibly early 2017.
@ performance when using different GPU for rendering
Thank you a lot I was not aware of that limitation of Iray and will make sure to adjust activated GPU accordingly for Iray.
I intend to use a mix of Iray and OctaneRender depending on projects.
Afaik for OctaneRender there is no render speed loss when using older cards along new cards for rendering.
The current rules are:
- VRAM aka graphic card memory does not add. The lowest available VRAM of any GPU on the system will be the maxium available VRAM for rendering.
BUT
- cuda cores do add in OR. Every single core that you have on any card will count towards faster speed in a linear way.
- - -
-> Some extremely motivated users are now even contemplating to build network rendering rigs with 20 GPU that combine new and old cards.
Basic idea: Do not throw away old cards if you can still use them to speed up your current render.
But unfortunately the external rigs for the additional cards are quite expensive. Prices for external card boxes are almost as expensive as an additional workstation. Therefore network rendering with external GPU is more of a long term strategy if you know you will have 2-4 "outdated" cards each year that you still want to use in some ways.
compare:
Best Practices For Building A Multiple GPU System
https://render.otoy.com/forum/viewtopic.php?f=40&t=43597
It is now a 63 pages thread but you can pick up quite some useful information.
- - -
@ Cuda 8
The release candidate of the Cuda toolkit 8 is now available.
https://developer.nvidia.com/cuda-toolkit
Simply put: Both Iray and OctaneRender developers had to wait for the Cuda 8 toolkit before they could start delivering support for the new Pascal GPUs.
...considering my current GPU has only 1 GB of VRAM (was one of the best available back when I built the system, then again we were all still using 3DL so VRAM only mattered for Viewport and desktop performance), 12 GB is a major upgrade. If Pascal pushes the price down for Maxwell GPUs It would be a bonus. While it so would sound nice, I don't need a render process to finish in 30 or 60 seconds as I'm not a major CG production or animation studio that has dozens of jobs to push through by "last week". If a large (meaning, say a 18 x 24" output for printing) render job takes a couple three hours, I'm good with that.
2 Titan X's should give me a decent screen response in IRay mode (like I saw in that demo you made a while back). That is far better than what I am dealing with now where it can take a minute or more for the viewport to refresh even for the tiniest change (which also keeps eating away at physical memory until the Daz application freezes up and crashes to the desktop).
My bigger concern right now is what OS to use when Win7 hits it's EOL in about three and a half years. 10 is out unless MS relaxes their update policy as well as offers a real "kill switch" for useless bloat like Cortana, and so far Linux and Daz (as well as Hexagon and Carrara) don't like to play very well with each other.
...I tend to overbuild my systems, particularly where components like the PSU is concerned. My current system has a 750w PSU and it rarely if ever breaks a sweat ("peak" outputs are in the 250 - 325 W range even during a "heavy" render job), of course, I am running no other processes in order to dedicate as much CPU and memory resources to the render job.
Yeah tech stuff tends outdate itself faster than one changes their socks so the trick is to build a system that works for as much of a long term life as possible. Trying to keep on top of the SOTA curve would be manic at best (and rather costly). That's why companies have dedicated IT departments I'm my own IT specialist, and modeller, and scene designer, and touch up artist. While yes, it's fun to design and build new systems, I'd rather be using most of my time to create art with the machine I build.
...yeah need to win the Megabucks lotto for that, however, most Xeons tend to clock slower than i7s (the E5‑2699V4 is rated at 2.2GHz, slower than my old Nehalem i7).
For just the price of those 22 core CPUs, I could build a pretty hefty i7 beast (now up to 10 cores at a base of 3.0 GHz), 128 GB of memory, x 4 Titan Xs (Maxwell) dual 1 TB SSDs with several 4 TB HDDs for storage as well as a 2,000 W PSU to drive it all.
As to memory sharing (using DX12), that will not facilitate memory stacking for rendering purposes and from what I have gathered is more to benefit to frame rate for the latest games.
Why not keep using Windows 7? I only dumped my WinME computer last year because the motherboard finally fried. Just keep it offline and download everything manually and it will last. If you want to get online, do what I do and build a cheap Linux box.
I'm not going to dump Win7 just because Microsoft quits supporting it, it works fine now, it will work fine in the future.
I would love to have kept Win7, but the manufacturer couldn't get parts for my laptop any more, so all they could offer was a replacement, and the only thing comparable doesn't support Win7 (and believe me, I tried).
Second that. My Win7 Pro system will stay Win7 until either the motherboard fails and cannot be repaired or DAZ Studio will no longer run on Win7. I already use a cheap Linux box for online access. :-)
With 4 x Titan X you need no more than 4 cores processor anything more will make no difference , the only difference is the 4 cores clock as I tested today from 1ghz to 4.2 ghz on my 6 core i75820K
since iray is more of a hybrid , that use both cuda and cpu in the workflow a faster card and slower processor will not improve anything . Each time you move object in the scene you qill have to wait for the CPU to update the scene before it render .
I am closer to my perfect build rig , I am never finished as before I do something new come out and I upgrade lol
LGA 2011 v3 Intel Core i75820K Haswell-E 6 cores at 3.30Ghz Boost 4.4Ghz Haswell-E water cooled $400
CPU water cooler $88
GA-X99P-SLI Mobo $270
128 GB memory $ 428
1TB SSD $320
1+2 TB Sata HD $160
3 xTitan X 12GB Superclocked Hybrid water cooled $3.600
1300W PSU $220
Case Fans +Cables $150
Case - $160
------Total ----- $5590 + my work an time as I build it myself for the last year , but you know, you never finish with building your CPU , I started 18 years ago and still doing it until today lol
now go and buy that by a computer dealer , you will pay at last $1000 or more for that just for the 60 min of work connecting cables .
If I ever want one more Titan X I will have to switch the processor to i75930K at last , same core count but more PCI-e lanes but I guess next year faster processor will be already much cheaper for upgrade
My Mobo and CPU can run now my cards at 16x 8x 8x but I had to disable some SATA ports in the Bios to free some lanes for that
It will work fine if you don't upgrade your system and keep it this way, but there will be no security updates in the future and as long the manufacturers support it with the drivers you good, I liked my win 8.1 Pro my favorite , I could change my Mobo or HD and still no issues now I have to reactivate each time I make changes , and I did a lot in the last 6 months like 2 motherboards and 2 processors and 2 HD , it is very invasive and updates screwed my SSD and locked it too so not best . So now I only update the system when I am ready .
I'm very curious to see how these handle Iray.
Meanwhile, here are more 3rd party varients of the 1080. Of note here, many of these use 8+6 pin power for more OC potential.
http://wccftech.com/nvidia-geforce-gtx-1080-custom-model-round-up/
Also, recall the power delivery during the presentation, and how wonderfully smooth Pascal is. Pascal regulates voltage better than any card before it. I believe this could be a sign the gpus will in general will have solid life spans. This is also part of why these cards OC so well. You do not get the less stable voltages that would come in older chips. But again, buy an extended warranty if it is that worrisome. The warranty can be transfered, which would increase resale value.
..what fixmypcmike said.
...the real trick is to find an MB that has four PCI 3.0 x16 slots capable of handling double width cards and also supports Quad Channel memory (LGA 2011-V3). The reason behind a 10 (3.0 gHz) core vs 4 core i7 is contingency in the event the render process exceeds VRAM and dumps to the CPU. 20 CPU threads is better than 8 in this case. Better for rendering in Carrara (which I also still use) as well.
...however the benchmarks for gaming mean little when it comes to CG prodution rendering.
Trick ? not really, the best that support everything you want is this one https://www.asus.com/us/Motherboards/X99E_WSUSB_31/
but I am not going into quad GPU not now not later for now I am done ..
you talking about cores or thread ? you will have to invest over $1000 for CPU to get 18 threads with Haswell-E, where 12 threads CPU cost $375-400
beside there is a minimal difference between 16x and 8x in games or rendering you will not even notice , that is another myth that cost a lot more for nothing .
do you need to spend $1000 for CPU so you have 1-2 sec faster rendering ? no, but if you raytrace then you shoukd go with different processor that offer more threads but they are not cheap
I tested 2 cards on 16x and 2 on 8 x too to see if that was true .. nada on perfomance ! then I research some benchnarks and the truth was naked ..
I predict that in near future raytraced rendering will be a history as I predcited unbiased GPU rendering back in the 90's in one inteview
if we go as fast as we do , in the next 5 years you will not find 1 product for 3DL lol
My Super clocked GPU's use 150-160W while rendering with all samplers and 175W when using standard rendering setting so it will be 125W for 1080
*super clocked without adjusting higher voltage
appears there's a problem with the new 1080s with the fans http://www.eteknix.com/gtx-1080-owners-worried-fan-revving/
is that only a minor problem that they'd be able to fix quickly easily?
...I still use 3DL for certain styles of output that Iray does not support. Why go through all the trouble of setting up PBR materials for a photoreal render only to resort to a lot of powtwork to make it look like it's not a photoreal render? I don't think 3DL or Renderman is going away anytime soon. 3DL is also pretty fast these days as long as you don't use UE (I use the AOA Advanced Lights and camera effects). Yeah you have to to a lot of "faking" for things like AO & such but I have produced some pretty nice results all with minimal postwork (and then, mostly tone mapping using filters).
I also work with Carrara and even Bryce from time to time, as they handle large environments and environment generation so much better. I have almost the entire library of Howie Farke's sets which are Carrara only, that is a fairly hefty investment. Both programmes employ ray trace render engines, the only unbiased engine available is for Carrara and that is LuxRender via Luxus for Carrara, though as I understand the plugin has not been updated in a while.
So yes, I do have a need for more than 4 cores/8 threads as well as a boatload of memory.
Well there is Octane for Carrara and faster than iray .. check it , the best thing is when you change the morph it is instant update no delay , with my cards it would be faster but not going to bother with Octane anymore .
Unbiased GPU rendering take over and in couple of years with very fast and affordable GPU it will be only the way , you know how much would cost 2GB memory in 1991 ? 8 million dollar . Future GPU card will cost no more than $40 or less or just build inside MoBo and you will render on your phone or laptop with 1TB of memory . I remember when people ask me what I would do with 2MB memory , waste of money lol it was not that long time ago.
But coming back to our time , you see when you don[t have the power to render fast, everything is fine , but once you taste the speed there is no return back , you will know what I mean when you get your dream rig .
I always love Carrara from the first time we started to use it over everything Daz Studio was my last choice for rendering as it was the worse .If I own both I would merge them together for one fantastic software .
If there was iray in Carrara I will be using it only for rendering as there is nothing you can't do but it is so underrated
and as we talking about the 8 cores , I just decided to upgrade and got the i75960x Haswell-E 8 cores with 16 threads , the i75820K was good for other things but did not perform optimal with my 3 Titans X as it has only 26 lanes and I run yesterday in trouble after I added extra HD there was not enough resource , with the new processor and 40 lanes I am good even with 4 xTitans X , but not planing to do that, not worth the investment at this moment . I have 30 days return policy so not big deal returning the other one .
Hi Cath
If you don't play games or if you are really not to gaming,then I would suggest something like is E5 2683 V3 ES,those ES chips cost lot less than brand spanking new Xeon E5 or i7-5820/5930/5960x etc and they just works,I'm running same ES(Engineering sample) on mine X99 ASRock X99 Extreme 6,this CPU have 14 cores/28 threads and speed 2.35GHz on mine,temps under rendering in 3DS MAX I've never seen higher than 60C with fans running just at 750RPM,just downside is if you are using single threaded application then you are in bit of trouble,have played and playing games on mine and I'm happy with mine,I7 5960X is great CPU,just over priced as well sadly
Regarding using slower GPU and Titan X,I've been using Titan X with GTX780 in IRAY or Octane for few months without the issues,GTX Titan X alone is not slow card,but still with GTX 780 renders are faster,I've done few tests and in some tests GTX780 is fast as GTX Titan X(I'm running Titan X SC with EVGA AIO kit with different fan as stock fan on their kits are just loud,I'm running BeQuiet Pure Wings fan) in some renders,this depends on scene and what counts for me with Titan X is 12GB VRAM although still is way over priced
And regarding GPU rendering as whole,this depends there,I'm still fan on proper old school CPU rendering with Corona and V-RAY which I use most of the time,in work we are have RedShift and Arnold with which I play a lot and would love to have them,but they're well over my budget right now
And if GTX1080 will dethrone Titan X in rendering,I'm loking forward if its worth to get or is worth to wait on new Pascal based Ti or Titan X
Hope this helps
Thanks,Jura
Thanks Jura, I have my system already and just upgraded to i75960x Hasweel-E and it is overprized but with my 3x Titan X SC it will be just perfect even if I decide to get 1080ti ,Titan Hydro water cooled later so just in case for the future . I have also i74820K and i75820K , on i75820K iray render genesis figure in 3 second with my 3 cards with. But there is larger delay since processor need to update the scene if morph or pose is changed to 3 cards the camera movement is super fast with no delay in real time, with 2 cards everything was faster regarding updates and delay , also Xeons tend to render slower with iray than i7 so definitely not my choice , my socket support both 2011 v3 & Xeon but I think the one you suggested use 2011 socket . I hope it make sense lol
Iray is kind of hybrid , a lot of stuff need good processor , cores don;t matter for GPU rendering as long you have 4 cores 4 cards can run smooth, but then you have another stuff that is processed by CPU like bloom filter, Tone mapping , Updating of the scene when you pose, morph or add another model so good speed core is important , I over clocked just 4 cores of 6 to test stuff out with my i75820K , the Ghz did matter a lot so between 1ghz and 2 ghz very slow , the moment it was on 3ghz it started to run smoother with 4.40 Ghz it was just perfect , I run my older LGA 1150 i74790K at 5ghz that was the best but my mother board got damaged when the card got unlocked and the PCI-e socket was pulled off the board by the card, glad it happen at night when PC was off . I was so mad it was 6 months old motherboard , my case use clips no screws for mounting cards.. well that was for me ..back to screws only as I will never trust this again . So after that I decided to get different MoBo with better processor and jump ahead . Added one more Titan SC Hydro so now 3 total and they going to rock with i7560X , also changed the memory back to 64GB [Corsair DOMINATOR Platinum Series 64GB] they are double expensive than Crucial but the best you can get for my system and they overclock and adjust itself automatic using XMP 2 profile build in so there is not need to do anything for optimal result and can go high . I always wanted but did compromisers , this time I said NO , I will get the rest of the memory when the budged allow. I want the best for me this time making myself the best B-Day gift rig ever lol My complete workstation reach almost 9K investment over 2/½ years without the parts I replaced but hey it is my business not just for rendering in Daz Studio as hobby or games and the only thing that stay me in the way to get better is my budget , not for nothing they say compromises are for losers, as each time I try to cheat with something it die or get broke so I am losing on it only and zero savings , not this time . sorry for the long reply but I am obsessed with gadgets and electronic stuff , I can't never get enough . lol
...so wait, when was Octane made available to Carrara? Not that it matters as the price is most likely just as steep as it is for Daz.
The issue I have with unbiased/PB rendering is there are not a lot of effects you can do "in render" (like can be done with 3DL) because they are generally set up for photo real quality. I don't do a lot of detailed postwork as I don't have a steady hand for 2D painting so I need to get the most out of the actual render process when I want to do something more stylised.
What iray benchmarks are out there?
one of many
http://www.migenius.com/products/nvidia-iray/iray-benchmarks-2015
What iray benchmarks are out there?