In the past I have always got better VR performance using DX 11. This was both in FPS and Frametime consistency between frames (0 ms being ideal for ultimate smoothness). I use CapFrameX for my measurements. I use the same aircraft and flight route for my measurements.
Recently I tried using DX 12 again and to my surprise VR performance was significantly better in both FPS and Smoothness (near 0 ms difference between frametimes). I’m using DLSS Quality and DLSS version 3.7.20.0.
Anyone out there observed the same results?
I wasn’t too sure if this was a support question since I wasn’t really trying to solve a specific problem so I put it in the general discussion category.
Thanks for the clarification and moving the post to the more appropriate category.
Cheers,
Steve
2 Likes
Yes, using DX12 improve FPS and decrease latency
But unfortunately, Glass Cockpit become blurry
Back to DX11
Ryzen 7 5800X3D, RTX4070 TI Super 16Gb, Quest 3, 64 Gb DDR4 3200, VD Streamer
Does using DX12 require having HAGS on to boost performance?
I currently have HAGS ON. I’m using an RTX 4070ti Super and running 566.03 Game Ready Driver.
No, no direct dependency between DX12 and HAGS.
Interesting. I’ll have to try that. I mainly fly with steam gauges.
A little off topic but I see you are using the Ryzen 3D processor. I’m currently running an i9 11th gen. Curious what your previous processor was and if you noticed a difference?
Same for me. DX 12 actually works better.
And for glass cockpit as mentioned. TAA instead of DLSS
I normally use DX11 but try DX12 each time a new sim update comes out, including the latest one. Each time the result is the same - similar FPS between the two but more CTDs with DX12 - so I always go back to DX11.
I prefer the lighting in DX12. I fly almost exclusively at dusk and dawn. I am not impressed by the current state of DX12 with Vulkan waving it’s arms so clearly in the distance. I hope when 2024 comes out we get the hotrod version.
I tried 566.03 and DX12 and I agree that it seems that frametimes are more repeatable which really helps Virtual Desktop SSW for sidelooking views out of the cockpit. I also noticed that for my settings that VRAM using DX12 was finally, after all these months, quite stable. Before, with a 3080ti, I was very close to running out of VRAM since I only have 12gb
HAGS seems to be making a significant difference as well these days at least for me. The difference in FPS with HAGS ON/OFF was about 35%. I ran the benchmark twice to make sure. It didn’t make any difference in consecutive frametime difference (generally 0 or very close to 0ms). This is on a 4070ti Super.
Is that 35% increase with HAGS on or off?
1 Like
With my i913900k/rtx4090 using a QPro with Link, nvidia 560.81, and dlss 3.7.20 I don’t see any significant difference with HAGS on/off. Maybe still slightly better with it off. Likewise, I don’t see any significant differences between dx11 and dx12. Maybe slightly better performance with dx12, but unfortunately, still not as stable as dx11.
So for me, dx11/HAGS off (also game mode disabled) is still my overall favorite setup. I also find dlss/balanced with 1.3x Oculus Debug Tool pixel density (super sampling) gives me a TAA-like cockpit clarity, with a lot less distance shimmering.
When msfs2024 comes out it’s going to be dx12-only. It’ll be interesting to see if HAGS on/off makes any difference then. Time will tell I guess. Cheers.
35% increase with HAGS On. PiMax headset.
1 Like
I kinda find that pretty hard to believe because I get much better PCVR stability with HAGS = off, esp. with dx12 enabled. Probably depends a lot on your cpu/gpu I guess.
I was pretty shocked myself. That’s why I ran the test twice. Go figure🤔
Yeah I’m kinda surprised that you got that increase too. I’ll have to benchmark each in fs2020 to see. I am using quest 3 with Quest Link though so I imagine that alone changes how the HAGS interaction affects the fps while tackling the quest link encoder.
First test with HAGS on though seemed to be the same experience at first glance.
1 Like