Windows 10 and 20% of a missing VRAM

[coilbook]

Hi,
here is a good article about 20% of a missing vram. I think it's been years Microsoft and nvidia did nothing about it. So our 11Gb cards are only 8-9Gb instead. Anyway otoy can investigate with these companies to finally solve this problem. This guy bellow mentioned about starting 2 tasks so windows would release all of the vram. Maybe otoy can make a cheat code where it starts two tasks to grab all vram and then somehow leaves one task running with a all of the vram available.
Thanks

From the article:

Looks like I was able to utilize 10.8 GB out of my 1080ti's 11 GB. I started with my Matlab code that loaded the gpuArray up to what it saw as the max available, 9 GB. At the time there was also some system stuff using 0.6 GB. But then I opened a 3D rendering app which used about 2GB of GPU memory, and it loaded fine. In fact Task Manager showed 10.8 GB out of 11 GB of GPU memory "dedicated". So maybe what Microsoft is saying actually IS true. Maybe any one individual process can't access more than 9 GB, but if another process comes along and asks for VRAM it will get it. Now when I look in Task Manager under the Details tab, where you can see the Dedicated & Shared GPU Memory for each process, it shows Matlab taking 8.6 gB and the 3D rendering process taking 2.1 GB, which indeed totals 10.7 GB. BTW, both are also taking some "shared" GPU memory, which I believe is system RAM. There are also some other system processes taking GPU memory at the same time, amounting to maybe 0.2 GB or less. Interesting...so maybe that's the story. Maybe any one process can't access more than 9 GB on a 1080ti, but if someone else comes along it can grab whatever is remaining, so that the total utilization is just under 11 GB.


Article: https://www.mathworks.com/matlabcentral ... lable-vram

[haze]

Off the top of my head, windows imposes certain conditions on games that do this. For instance, at any moment windows is able to send a message to the app to force it to let go of its resources, mid-render. The sheer complexity of an infrastructure that allows this and/or having two processes to maximise VRAM capacity is so intractable we would have to rewrite the whole render device subsystem - this to solve an issue introduced only on 1 platform, and a restriction forced upon WDDM drivers by Microsoft.

That last statement is important, because NVidia also has another driver called TCC (tesla compute cluster), available on some devices, which is not a WDDM driver. Generally only available on Quadros etc, but it is supported on Titan X Pascal. You can use nvidia-smi to switch the device to TCC mode, and restart. This is what it looks like :

titanxpascal.png (9.65 KiB) Viewed 1379 times

[coilbook]

Off the top of my head, windows imposes certain conditions on games that do this. For instance, at any moment windows is able to send a message to the app to force it to let go of its resources, mid-render. The sheer complexity of an infrastructure that allows this and/or having two processes to maximise VRAM capacity is so intractable we would have to rewrite the whole render device subsystem - this to solve an issue introduced only on 1 platform, and a restriction forced upon WDDM drivers by Microsoft.

That last statement is important, because NVidia also has another driver called TCC (tesla compute cluster), available on some devices, which is not a WDDM driver. Generally only available on Quadros etc, but it is supported on Titan X Pascal. You can use nvidia-smi to switch the device to TCC mode, and restart. This is what it looks like :

titanxpascal.png

Thank you. I wish Microsoft could explain why they do this. When you report the problem to them a lot of people at microsoft have no clue.