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
That is impressive, and I'd be keen to look into that myself.
Currently I use Octane, and I can use my main PC (32GB) for texture RAM, and the only thing that needs to be in the GPU is the geometry. (I think.)
I'm currently using 2xGTX 1660TIs with 6GB of RAM, and my biggest render was this with 11.5GB of RAM. Using out of GPU RAM doesn't really impact performance. (10%?)
https://www.daz3d.com/gallery/images/988816/
All I can say is that all this is so much better than using Reality and 20 hour CPU renders. Or 3Delight with dedicated CPU license.
Another concern is that I have heard from multiple sources that (for Daz Studio Iray anyway) your PC should have at least 2x the ram as your card has, preferrably more. I have not confirmed this myself, or seen any actual data to say why this is so. I guess this makes sense, particularly if you might also run the render on CPU.
I am concerned that newer versions of Daz Studio are reported to use more resources in older cards than previously, and this may make these Maxwell-based cards less of a bargain.
I will update when I have more results.
Addition to above data: Power consumption was about 170-180 watts at >99% GPU utilization. Was able to power from an EPS 12v 8-pin CPU plug, or using the adapter for two 8-pin PCIE plugs (I used two separate cables).
...issues with non RTX cards and the newest version of Daz? Which version? I'm still on the 4.12.0.47 Public Build Beta and if there's any instability older GPUs and the newest version, I will probably stay where I am at. I don't have the resources (or the recommended memory as the MB and BIOS on the render system tops out at 24 GB) for an RTX Titan.
....I checked out the new Octane4 and saw that the 2 GPU subscription version requires being online while working. That doesn't sit very well as being online eats up extra system resources and you are also at the mercy with of connectivity quality/speed. Definitely cannot afford the perpetual licence which at least lets you have a USB key (at an extra 45$) to work offline. If I had a SOTA system and Gigabit fibre optic that would be one thing, but my hardware is ancient and my connection is at best a 100 megabit with which I get buffering pauses even while trying to watch a live 15 min newscast on YouTube let alone live stream 2 hour film.
Another point to consider is that the drivers for the Tesla M40 (24GB) for WIn 7 were last updated in Nov., 1019, while the drivers for Win10 were updated in March 2020.
Wondering if NVIDIA has stopped updating drivers for Win7 for this.
They probably stopped Win 7 drivers when EoL was reached.
Yep, that's what I figure. Fortunately, they claim that because the Teslas run in TCC (Tesla Compute Cluster) mode, Win 10 can't steal the 15% of VRAM it demands to reserve with "normal" GPUs. That is one reason I got these.
Win10, Server 2019 at least, doesn't.
...yeah may just stay with what I have.
...but that is a pretty expensive option. Even W10 Enterprise LSTB (which doesn't have all the useless "fluffware" or dumps pointless feature updates on your system every 6 months) requires volume licencing with a minimum of 5 nodes.
Windows Server 2019 for my 24-core server (in the picture) would be $1100.00. Ack!
TCC being separate is reported to be in all versions of Win10. The only one I can test with Quadro's or Tesla's is Server 2019. That doesn't mean the others don't have it.
...the Titan series does as well (even Maxwell) but as I understand at a slight cost in fan speed, not sure why. Using TCC mode so also requires a second GPU to run the displays. For now I'm content with W7 pro as this is older hardware and the OS is more than sufficient plus I like having full VRAM for Iray mode.
I haven't checked what my license is.. its perpetual and I paid like $300US years ago. I can use as many GPUs as I like in Daz, and I can export to 1 other computer with has many GPUs as it can handle (the limit is 8 GPUs per PC I think). I rarely export to another PC unless I'm desperate for performance, so meh. I heard the free license was 1 GPU, and all the latest updates. That means 1 OK GPU with RAM for geometry, and you should be good to go. Online doesn't use resources, but it has dropped out once for about half an hour this year. The Plugin and Daz died violently. :-)
The downside to Octane is the work I have to put into textures\shaders. I'm kinda used to it since 3DL wasn't exactly fully supported when I first got into Daz.
If you're going through hoops to get performance, perhaps you should take some time, to save and move to a better system? There are so many other things driving performance for Daz. I'm running 2 2TB Intel NVMEs one is for my primary drive, (Windows, Daz, Games, etc), and the other drive is my Daz library. Finding files for Daz, loading, etc, is all way faster. (My old PC was running 4SSDs as a 1TB RAID 10, it was no slouch, but these cheap NVMEs blow it right out of the water.)
...unfortunately I am on a fixed pension now so a more "SOTA" system is out of the question. It would also require I move to W10 and I still have some reservations about that (I'd only go with Enterprise LSTB but that is more expensive as well as you have to deal with volume licencing and I'm not sure I even would qualify as I don't own a business). So, I'd have to stay away from anything after Haswell (Intel) or first generation Zen (AMD), the latter which really restricts the maximum amount of memory to 32 GB.
Yeah, the Bios of the P6T doesn' support NVME only SATA III Though the upgrade from an HDD to a SATA III SSD was remarkable enough for me.
One thing I did discover doing a bit of research is that a P6T X58 MB can actually support 48 GB of memory I've been reading reports of some doing so and the key seems to be running an X56xx Xeon (I have an X5660 6 core). Given that,it makes putting in a 24 GB GPU more attractive now, however I think I'll hold off and see what happens to Turing prices when Ampere is released. Even with 7 fans (2 which directly blow on the GPU) and a tonne of airflow, not sure about running an M40 "bare".
OToy's licnencing seems a bit confusing. They mention perpetual licences but only seem to have is what they call an Enterprise Box Licence that includes one year of their Maintenance extension for 699€ (about 764 USD). It still sounds like an online situation as for an extra 49€ (53 USD) you can get a USB dongle that allows you to work offline (though that needs to be updated every four months) There is also a Box licence that includes two years of the maintenance extension which is 899€ (983 USD). It appears the least expensive way to qualify for the offline dongle is going with a one year Enterprise Subscription which is 299€ (327 USD). seems much simpler when it was 299 USD for the engine and 124 USD for each plugin. and you got a break on upgrading to the next newest version. the Studio Licence does not support the offline Dongle (even for the annual one). Personally I like working with software that is on my system,out out there in the cloud. so it looks ins spite of all the improvements (including support of NVLInk) that Octane 4 will be a "no go" as they just made it too complicated and unwieldy. Sad because I love the concept of out of core rendering since I tend to do epic level scenes..
As to translating shaders that is nothing new because I used to work with Reality/Lux long before Iray was adopted. For 3DL I also have Wowie's AweShader kit and RSSY's Iray to 3DL converter.
I would not invest any cash at all in the short term on server/workstation hardware as old as X58. Once the current crisis passes the datacenter world will get back to dumping a lot of newer HW for EPYC Rome. We wanted to dump all but a couple of systems this year, obviously not going to happen now but we might still get rid of most of them. The big players will maybe not go quite so fast but there will still be a lot of reasonably new server chassis in the used market soon.
Also while there is a big theoretical speed difference between a SATA SSD and NVME ones, and between Gen 3 and gen 4 PCIE, the reality is not as large as those raw numbers. The bandwidth is not the sole consideration and you will almost never saturate a SATA 3 connection so the performance advantage of an NVME might never even be measurable
I'm not liking W10. In fact talking about it just winds me up. This PC is multi user.. but Cortana isn't? Whiskey Tango Foxtrot! And really I don't want Cortana scaning my drive to create advertising options for me. Oh and its trashed the other account twice now. (W10 works... its just annoying.)
....the situation I built the system back when the P6T X58 was still "State Of The Art". Built that system on a "shoestring" the way it was. Since retiring (physical disability) haven't had the income to embrace newer technologies (the cryptomining craze didn't help either as it came at just the wrong time, causing GPU prices to skyrocket). On the pension I have now, the best I can do is upgrade the old system as far as I can push it. Also as I mentioned have no interest in W10 which the latest generation CPUs require.so even if I could afford to build a new system, I'd still be using "legacy" components (pre Kaby Lake/Zen architecture), most likely going to dual Xeon workstation setups to get high core counts should the process dump to he CPU as well as for 3DL/Carrara. One benefit with Intel, is their CPUs supported 4 memory channels years ago when DDR4 was introduced. whereas AMD didn't until Threadripper (that does make a difference).
The good thing is newer GPUs also work in older PCIe slots (for now) and most DDR4 MBs (even some DDR3 ones) have PCIe 3.0 expansion slots. The two two rubs I see is Nvidia possibly pulling the plug on W7 drivers (even though CUDA ver 11 still supports the OS) and older GPU technologies as well as what Greymom alludes to with newer versions of Daz being harder on the resources of pre RTX GPUs, than the RTX ones.
...yeah, Cortana is just one of my beefs with W10 as it is integrated into the OS so if you totally remove it (like doing so with IE in older Windows versions), it hamstrings several basic OS functions. Yeah you can "disable" it but it still runs in background and returns like a bad penny after feature updates. Crikey, I don't need some goofy anthropomorphised digital assistant nagging me. Did they not learn their lesson from "Bob" and "Clippy" that serious users don't want such annoying rubbish?
...and adverts, ugh. It's bad enough browsers try to tailor them to your search habits and net sites are rife with them to the point, what I am trying to read or research is almost unbearable because of all the popups, scrolling adverts and embedded videos vying for my attention (I've actually closed tabs after only a few moments sometimes because it becomes so bothersome while more and more sites are using Anti Advert Blocker software to force them on you), now we have an OS pushes them on you as well.
Were are you seeing advertising? I have W10 pro, have disabled Cortana and turned off all the permissions for it and I don't see any ads in the OS.
Hey @uezi and @Greymom - fancy contributing some benchmarking runs to the updated Iray benchmarking thread we currently have going here? Full directions here. We have yet to get any numbers for Tesla cards, and the M40s are obviously of particular interest to lots of folk here.
As soon as I get DS installed on the test machine, I plan to run those benchmarks. Have to test the M40s while they are still in the 30-day return period first. Hope to run direct PCIE connection vs. 16x riser vs. 4x riser vs 1X USB eCard. I am using LuxMark 3.1 first as it does not change or update very often, and is very quick to set up for each card. It also has a stress test mode, and you can run either the OpenCL engine or the C++ engine.
Today, brand new M40-2 (24GB) were running as low as $318 on EBAY.! I have seen the used ones for as low as $225. That is an incredible price for a 24GB rendering card!
I don't either. The only place you see adverts is if you download free applications.
Good morning
Would it be possible to mount a Tesla m40 24gb on a Razer core x to connect to a Surface pro 8?
Thanks