Titan X Maxwell vs Pascal speed?
Slowmichael
Posts: 65
in The Commons
I am shure this has been asked before but... . I am considering buying a used Titan X. The Maxwell is a lot cheaper. So what would be the difference in speed between Pascal and Maxwell for rendering and also for dforce
Many thanks.
Michel
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
It's hard to quantify, particularly since you didn'y specify what to compare against besides just the whole microarchitecture. If you mean Titan to Titan then the difference in performance is substantial.
This should help you sort things out.
https://www.daz3d.com/forums/discussion/341041/daz-studio-iray-rendering-hardware-benchmarking/p1
For games Titan X Pascal is faster 60%, but in rendering not as much , for rendering for example 1:10 min with Titan X P, Titan X M will do that at 1:20 min, not as huge difference but the price is, it depends of the specifications , my Titan X M run at 1350Mhz , so look on the specifications as not all Titan X M run at the same speed..
Also bevare buying online used card, a lot of them have custom BIOS , make sure the BIOS are certified or original or you wasted your money for nothing and not able to use last drivers updates correctly for iray
Thanks for the answers. It's always arisk bying second hand certainly graphic cards in particular.
I had 2 purchases 6 years ago, and one with huge dispute , it was fake with custom BIOS promoted as new EVGA, but it was home made , not original EVGA, I won the dispute .Make sure your purchase is covered just in case at last 30 days . It will give you enough time to validate it at home . Maxwell cards are faster now as it was 5 years ago in iray, it improved dramatically now, I run 4 of them in iray and very happy about
I have a Titan X (Pascal) in my system and am well pleased with it. It's performane for IRAY rendering is very similar to the 1080Ti (you can find both the 1080Ti and Titan Xp, which has the full unlocked Pascal chip and which released 6 months after the 1080Ti, in the benchmarks thread referened by another responder). dForce perfromance is pretty good too--I usually use it for single figures at a time for dForce clothes and the simulation times are about 2 minutes or so. I bought mine directly from NVidia, so no worries about possible custom BIOS/proprietary driver issues--may want to look for that when buying used. I run mine for rendering with a slight overclock (+100 MHz on the boost clock, +450 MHz on the memory clock using MSI Afterburner and 120% power limit) and it runs perfectly stable and pulls 170-200 W during renders (boost clock hits around 1900 MHz with temps around 60-65 C on the stock cooler). The 12 GB VRAM is great for rendering--if I turn HD textures off and reduce texture maps to around 1024x1024 I can render scenes with up to 14 G3/G8 figures and a background/HDRI map and still keep the scene on my card. I would recommend you also consider RAM usage with this card--if you only have 16 GB system RAM you should consider upgrading to at least 32 GB for larger scenes. Gaming wise, my Titan X can handle pretty much everything out there at 1080p or 1440p resolution at high/max settings without breaking a sweat.
The only potential issue I would consider with the Maxwell cards is that it is an older generation and therefore one has to consider how soon support for them will be discontinued in D/S and IRAY.
If Titan X Maxwell loses it's support so will be Titan X Pascal , but I don't expect it very soon . Low capacity cards based on maxwell architecture may be soon as that what Nvidia was talking about and not Titan or Quadro , there is a difference between gaming and deep learning cards. I had to wait almost 3 years to see the full potential of Maxwell architecture in iray after purchase .. it is not going so fast because if they do that then 80% of users will have to switch to CPU rendering or buy new cards, it could kill everything .
I will give it at last 3 more years before starting to even thinking about but at that time I will be ready for snapping 3090 for half price and hopefully iray is fully optimized in D|S to handle it ..or maybe early because nothing last forever ..
I was pretty mislead once-in-a-lifetime , not going to do it again , wasted $6000 , had to wait 3 years to use it optimized in iray and now have to do it again ? no way, the cards just started to working the way they should 3 years ago. That's all I have to say on this subject.
Actually, the GTX Titan X and Titan X are not from the same GPU architecture. They will not end Iray support at the same time, unless Nvidia drops Pascal and Maxwell at the same time. BTW... When a GPU generation looses Iray support, it includes all Geforce, Quadro and Tesla cards. You do have at least a few more years on Maxwell, because Kepler will lose support first.
Realitsicaly at this point there are only a few Maxwell cards that are even worth using for Iray because of limited VRAM.
Geforce GTX 980ti 6GB
Geforce GTX 980M 8GB (Alienware version)
Quadro M5000M 8GB
Quadro M5000 8GB
GTX Titan X 12GB
Quadro M6000 12GB
Quadro M6000 24GB
it stated in the Nvidia post , they losing soon the support of low ends card with maxwell architecture in the upcoming builds and not all of them or all Geforce , so as I said before couple more years
What they mean by are probably cards with limited VRAM bellow 6GB and Kepler/Maxwell architecture
...I have a Maxwell 4 GB 750 Ti in the assembly system that works well for doing character proof renders. I can even render two fully clothed characters using the Fast Production Lights for Iray.
The simple down and dirty on Iray vs Nvidia GPU support has to do with the drivers. As soon as a GPU family drops to legacy status, Nvidia stops including the associated code into new driver packages. For example: "Driver Version: 391.35 - Release Date: Tue Mar 27, 2018" is the final driver version that supports the GTX 500 (Fermi) series of cards. That means that any version of Iray that's looking for a newer driver version will not see Fermi based GPUs for render jobs. The same thing will happen with Kepler based cards (GTX 600 and 700) when Nvidia eventualy tosses them to legacy status. To be fair, by the time a GPU family hits legacy status, all but the top teir cards will already be pretty useless for Iray weather it be for lack of VRAM or just being too slow.
I also forgot a couple of Maxwell cards on my list. :P
GTX 980 8GB (150W MXM Card)
Quadro M5500 8GB (150W MXM Card)
Quadro M4000 8GB
As far as the old Kepler cards... The few the proud, the slow cards with enough VRAM to be usefull for full scene renders.
Desktop:
GTX 780 6GB version
Geforce GTX Titan 6GB
Geforce GTX Titan Black 6GB
Geforce GTX Titan Z 12GB (This is 2x 6GB GPUs on one card)
Notebook:
Geforce GTX 880M 8GB version (This is a Kepler card despoite being marketed in the 800 series)
Workstation:
Quadro K5200 8GB
Quadro K6000 12GB
Mobile Workstarion:
Quadro K5100M 8GB
That's true, for that reason it is too early to make a call , I don't even consider update until I know 100% that the latest cards are 100% efficient in iray and Daz Studio
Only last year they got finally right and made cards with maxwell architecture efficient especially if you use more than one in your system and I am going to skip updates to RTX 3090 and by the time there will be already new model and RTX3090 price will drop in half . My workstation is faster than RTX3090 so not really hurry yet , plus I will need at least 2 of them to make the move forward. 4GB you have on this card is not much but enough for many tasks and simple scenes . Stonemason Streets of London don't even use 3GB , you can still render it on this card . Iray started as internal engine for rendering products, it is also moving towards commercial productions where VRAM is very crucial and that is the main focus , we are just in the middle here of it all, eating the crumbs ..
HUH?
Microarchitectures are distinct and separate. Kepler is not Maxwell is not Pascal. When each one is moved to legacy status it is no longer supported by iRay and when the next version of iRay is integrated into DS that microarchitecture no longer works. There is no distinction of high end or low end. People have had Fermi Tesla's not work just like they had Fermi GeForce cards not work. Nvidia did confuse things by having some 700 series cards be Fermi and some be Maxwell but that is just Nvidia being sloppy. You can look up which are which.
That's maybe what they meant by.. , there was not exactly a specification or list just , ending support for iray low end /Maxwell ... as JamesJAB stated that my be related to the driver support . Not going into dispute or arguments ...