HAGS: Hardware Accelerated GPU Scheduling *Newt Thread*

iroboto said:
still not making a lot of sense to me, the GPU scheduler should have been in charge of this.
Click to expand...
This document might help.
As more hardware vendors move toward a hardware based scheduling model, where work is submitted to the GPU directly from user mode and where the GPU manages the various queue of work itself, it is necessary to eliminate the need for the video memory manager to inspect and patch every command buffer before submission to a GPU engine.

To achieve this, WDDM v2 supports GPU virtual addressing. In this model, each process gets assigned a unique GPU virtual address space in which every GPU context to execute in. An allocation, created or opened by a process, gets assigned a unique GPU virtual address within that process GPU virtual address space that remains constant and unique for the lifetime of the allocation. This allows the user mode driver to reference allocations through their GPU virtual address without having to worry about the underlying physical memory changing through its lifetime.
Click to expand...
https://docs.microsoft.com/en-us/windows-hardware/drivers/display/gpu-virtual-memory-in-wddm-2-0
 
Testing Hardware Scheduling in RDR2, Vulkan API.
Thanks to PhazDelta.

Edit: GTX 1080 8GB

New driver, Hardware Scheduling on:
piw3YbU.png



New driver, Hardware Scheduling off:
ZhJMNuR.png


Old driver:
Y7GqyrA.png
 
Last edited:
That's a huge increase. Looks like the new vulkan driver is quite a bit better, but turning on gpu scheduling makes an appreciable improvement.
 
Hardware-Accelerated GPU Scheduling Performance Snapshot With Turing
Today we have a simple goal, does it impact performance? I suggested we look at a card that might benefit more obviously from a memory management standpoint, but Hassan (I'm calling you out) suggested I use the RTX 2080Ti. I did, but I also used the GTX 1650 SUPER with it's 4GB of VRAM to see just what would shake out.
...
With only two cards and two games, I can't draw a hard conclusion across the board but we can see a tale of two different expectations. High-end enthusiast cards with plenty of VRAM to go around seem to not be impacted at this point, that could easily change over time as this is still a very new feature.

BUT, if you're a bit more down the rung and your card has less VRAM or is in the entry to mid-range when we see the GTX 1650 SUPER gain 7-8% performance in Gears Tactics and 8-9% in Forza Horizon 4 there's no denying that this update gives graphics card users in that range a welcome push upwards in performance.
Click to expand...
https://wccftech.com/hardware-accelerated-gpu-scheduling-performance-with-turing/
 
I imagine there will be upcoming HAG scheduling refinements in the future as the feature is still new, but at least they now have the software/hardware to conduct those tests.
 

From this vid on a 1080ti looks like 1% and 0.1% lows are much better even though the overall average is about the same. CPU temps look lower too. No idea how reliable this youtube channel is though.

Edit: Seems to vary by game, but there aren't any cases here where it looks like performance is noticeably worse, just some games where it's noticeably better. I'd really like to see it tested on a low end cpu, or an old cpu. Maybe a system that would typically be seen as imbalanced and cpu limited.
 
Scott_Arm said:

From this vid on a 1080ti looks like 1% and 0.1% lows are much better even though the overall average is about the same. CPU temps look lower too. No idea how reliable this youtube channel is though.

Edit: Seems to vary by game, but there aren't any cases here where it looks like performance is noticeably worse, just some games where it's noticeably better. I'd really like to see it tested on a low end cpu, or an old cpu. Maybe a system that would typically be seen as imbalanced and cpu limited.
Click to expand...

With a 1660Ti
 
You must log in or register to reply here.

Similar threads

D
Hardware implementation of threading models in contemporary GPUs
Replies
16
Views
3K
Lurkmass
L
P
GPU Utilities for tweaking and overclocking
Replies
6
Views
3K
pharma
P
Shortbread
Next-Generation NVMe SSD and I/O Technology [PC, PS5, XBSX|S]
105 106 107
Replies
2K
Views
189K
Karamazov
K
Kaotik
AMD Radeon Software Driver Announcement thread
2
Replies
27
Views
11K
Krteq
K
Ext3h
NVidia multi GPU (non-SLI) with PCIe switch
Replies
15
Views
5K
Davros
D
Back
Top