Hey folks
Is there any way to push GPU usage in games? Elite Dangerous seems to hover around the 75% mark? It’s not a game-breaker tbh as this is a decent 60ish fps in stations with decent quality. Running two 2080Tis albeit only one is used for ED
Does GPU catalyst actually do anything?
Thanks
As ive understood the catalyst setting it decreases the amount of overhead allocated to be unused to handle spikes in gpu usage.
You can try increasing it, the worst that will happen is that you get some flickering and then you just decrease it again, best case is that you get a little more performance out of your card.
With a 2080ti you are quite sure to have a beefy cpu as well and i hear Elite is quite CPU bound in many cases.
This is a bit of a red herring. What you should try to do is maximize the overall framerate, which requires balancing your CPU and GPU usage. Pushing one or the other can be done, but in practice, any game will probably have bottlenecks for both during various parts of the gameplay cycle. These bottlenecks will generally vary by game.
To answer your question, just keep increasing the super-sampling. Eventually, the GPU will be working as hard as it can, but your framerate will probably be awful.
I may try going back to 253 PT. I remember hitting 94% with the same setup. Now with the latest PT it doesn’t go above 76
What is your GPU latency statistic in FPSVR, and what is your framerate?
I’ll have a gander. I don’t see stutters or anything like that. And the CPU isn’t taxed either. It’s stuck at 76% and won’t go above. Weirdly maybe, I installed 253 again via fresh install. It was hitting 89-94% but the FPS was roughly the same as the latest. So maybe it’s a deeper issue.
For the future, if you can post FPSVR framerate and latency statistics, we can be more helpful. Also, even 76% GPU utilization is high enough that the GPU may be thermal throttling.
The GPUs are now under water so definitely not thermal throttling. They hit 2115MHz each and don’t go above 46C. How do I capture FPSVR stats as it won’t show up on the screen inside the Elite window?
Nice. I wish I did that.
How do I capture FPSVR stats as it won’t show up on the screen inside the Elite window?
FPSVR should display a panel visible in VR. All you need to do is tell us roughly what you see. For ED, this is mostly GPU, not CPU. So, give us a report like, 14ms GPU frame latency, occasional red spikes that line up with stuttering, green fill, no yellow fill, and CPU all green.
But hey, Elite Dangerous VR is ending, FDev was never great anyway, Parallel Projections threw away 56% of our performance and more visual clarity on top of that… so who cares. We need a better game now, not better performance for a game (oh yeah, it’s just a game anyway, not a sim) that plainly sucks.
(not to diminish your efforts in the slightest of course)
Ah I thought maybe I was missing something on the desktop. The 76% value quoted is the figure I see the GPU usage settle on. CPU usage is very low. So I think it’s actually PiTool itself
Haha on FDev. Have you seen the new trailer for “space legs”? I am hoping its ok albeit it won’t have initial VR support. EDIT: I realise maybe that’s what you’re talking about.
Try other versions. I think can confirm 1.0.1.254_alpha might be is what I am using at the moment.
Haha on FDev. Have you seen the new trailer for “space legs”? I am hoping its ok albeit it won’t have initial VR support. EDIT: I realise maybe that’s what you’re talking about.
No, no, not at all (friendly sarcasm). I am talking about how much FDev as a company sucks for totally not giving a … anything about … anything a large fraction of their customer base cares about, even if it is trivially easy for them to support.
Interesting!
Are you saying that using your pre production 8KX and that PiTool version you get similar GPU usage on the same apps as you would when using e.g. an Index or literally any other HMD?
Because when I still used my 5K+ before receiving my Index (so about a year ago) my single biggest issue was only getting a maximum GPU usage of 7X% fluctuating, regardless the application.
2080Ti under water, 2.1GHz stable. No CPU bottleneck.
Increasing SS or any other GPU stressor only reduced FPS.
This happened across all PiTool and fw versions available at the time and was most definitely down to PiTool overhead.
If this is solved I‘d be looking forward to my pre ordered 8KX (been waiting for 7months+ now) even more. If not, I‘d probably go for a G2 as an Index replacement and sell both the 8KX and the former to get an Ampere based Titan.
In VR, efficient and effective GPU usage is the single biggest contributor to performance and the sole reason I still hesitate to consider a Big Navi AMD GPU. If this advantage is negated by inefficient driver level manufacturer‘s sw, it‘s almost as bad as a hw fault to me personally.
SW can of course improve but it‘s been a long time now.
Yes, this is a software issue with Pitool versions.
They really must fix Pitool. 75% is unacceptable. The catalyst is not an option either, it doesn’t work.
Best use currently is, as mirage had stated, 254a. It hits 95% utilization without frame-jumps. But will it support the 8KX correctly?
Be honest. All other manufacturers can do it properly, why can’t Pimax?
And stop blaming the users. It’s an issue with the software Pitool. It’s agnostic to anything, except Pitool versions.
I went back to 253 with a fresh install of the tool. Weirdly, I found that
So I went back to the latest. Yeah I agree 76% isn’t ideal. There is a room available, i.e. maybe even +20% ish if we can hit 95% utilisation.
Definitely a PT issue IMHO
How are the black levels and FOV compared to your 5K+? Did you change the Index as your daily driver?
Black levels never bothered me on the 5K+ but I‘d say the Index is slightly better. Since both can adjust the brightness, an objective comparison would be challenging.
Yes I did change to the Index as a daily driver as its friction of use is a fraction of the 5K+.
The FOV is most notably smaller on the Index though.
IIRC, 9X% GPU utilization is typical with DCS World and Elite Dangerous under my workloads, with the 8kX, and the PiTool version I use.
This is why I put setup files for successful PiTool versions in a ‘cert’ directory, by themselves, along with everything I need to revert firmware in a hurry.
The scale/FOV seemed slightly different. If felt more zoomed out (yeah, I know how that sounds)
I guess software IPD may be wrong with that version.
I agree with your findings…
But for me there is a FPS hit,
As the later version cant hold 90fps at the same graphics settings.
I stepped back after pitool 260 locked me in at 76% utilization.
They have to fix the later versions…
That’s a perfect example of what is happening. It’s EXACTLY what I have found. So what version have you settled upon?
253 atm
…
