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
I know that it is needed thing, the problem is that this issue is trimming SSD's more often than it should thus shortening the drives life faster than normal.. It also seems that this bug will try to use the trim on HDD's..
Sooo, since this is transferring the scheduler to the GPU, my question is, how does this play in to the 19% Windows 10 VRAM tax? Has anyone checked to see if more VRAM might be available for a render with the new hardware scheduler? The same amount? Less? Sure, the 'pro' cards have a 'headless' mode, which can bypass this, but the lesser cards don't have that option available, but since it says that Windows isn't doing the scheduling with this new option...
Well you can see why I'm wondering about this. My Daz build has been isolated from the internet for a long time now, hence it isn't getting any Windows, Nvidia, or Daz updates, so I'm not set up to test this. But when I do eventually tackle my next build... well inquiring minds want to know!
Yes, people have received an older version - though I don't have links to the posts to hand. Asking is actually one of the subject options for Technical Suport tickets.
WDDM does not claim a flat 19% of a GPU's VRAM. How these claims get started baffle me.
It's about that. 90% of 90% of the VRAM is the limit of the VRAM made available to any one application, UNLESS you have a 'pro' card in 'headless' mode. See this archived thread:
https://www.daz3d.com/forums/discussion/120541/vram-available-for-rendering-on-windows-10-systems-update-gtx-1080-ti/p1
Nvidia has acknowledged this behavior, and Nvidia's acknowledgement has prompted the Daz3D team to consider this 'normal Windows 10 behavior/solved'. See OP in linked thread above. For more info, there are several threads about this behavior over at the OTOY forums, Microsoft forums (although Microsoft dodges the issue), and a few other rendering-focused forums.
So I'm not looking to rehash the 'old behavior' in this post, just wondering if moving the scheduler over to the GPU has changed this behavior in any way (sets aside more VRAM for hardware scheduling, bypasses WDDM, etc). Has anyone noticed different VRAM behavior with the hardware scheduler enabled?
That's from 2018. Unless I missed one there have been 7 updates to WDDM since then, the version under discussion is WDDM 2.7.
I can test my system with my server monitoring SW. My 8 Gb card reports more than 7Gb of VRAM available and renders scenes that do take more than 7Gb on my 11Gb card. 81% of 8Gb is 6.5Gb. I'm sure you can see my problem.
With everything but MSI Afterburner shut down, I go down to 539 Mb used of 11 Gb. Now with just a few Chrome tabs open, it's at 729 Mb used.
Well, that's swell. What about those who didn't?
Also, this goes goes against this previous statement. Is this an actual change of policy? Will they really supply an older version to someone who asks now? What is the criteria for support providing it?
Huh? That literally makes no sense.
Well, that's literally what MSI Afterburner showed.
Sorry misread what you posted. My bad.
People have been reporting the same issue to Microsoft in 2020 btw. Still around 19-20% unavailable for rendering. Most people had simply given up on Microsoft ever doing anything about this, based on the more recent 'tech support' posts I've read via google.
HOWEVER, I did come across this Redshift Facebook post, that relates directly to this issue:
https://www.facebook.com/groups/RedshiftRender/permalink/959019687880575/
There was a similar post on the OTOY Facebook recently according to Google, but I wasn't able to spot the post due to it being buried by more recent posts. It's probably there somewhere, but there's a LOT of daily posts on the OTOY facebook that Google linked to...
Assuming this same behavior is also observed in Daz Studio, I would deem this an improvement vs pre WDDM 2.7, using the correct Nvidia companion driver along with 2.7 as mentioned in the quote above.
So again, I'm not asking about what Windows (10) reports as 'VRAM visible', but if people are actually able to RENDER larger scenes (over the @ 81% VRAM memory threshold for any one app) with the new drivers, or if they are still dropping to CPU only at that point.. The thread I previously linked explains how to read the Daz Studio logs for actual scene size and such.
I would appreciate it if someone OTHER than kenshawa would be willing to verify this 'new' behavior with a non-pro card.
Color me cautiously optimistic based on the Redshift info...