|
Post by Rune on Jan 14, 2023 11:22:59 GMT
The sources was just a joke ;-) I think your Aero with the RTX 4090 has reached a performance level where the hmd may be a lot more interesting to many. A level where it really starts to shine. I'd love to see your performance with the Aero after getting the 4090. Testing is a way to get rid of subjective impressions and get some objective results. Seeing Actual FPS and frametime numbers in a game that he plays is a better form of testing than running a benchmark that has consistency issues in a game (CyubeVR) that he doesn't and never will play imo. For him, yes, but not for the rest of us not into the same sim. I don't mind subjective experiences. There are a lot of Index users saying they got great performance with their 1070 cards, and while that's certainly nice, I'd appreciate more objective results. Can of course be from other VR benchmarks too, but we don't got a lot to choose from, and some do not use engines from real VR games. 2c. Btw, one dude tested several hmds (Pimax, Pico, Meta), but not the Aero or Index, and got about 50 - 70% increase for all in the OpenVR Benchmark going from 3090 to 4090 - so I'd expect 70% too for DBurne, but would be interesting to see the actual results.
|
|
|
Post by Pyroth309 on Jan 14, 2023 11:27:33 GMT
Seeing Actual FPS and frametime numbers in a game that he plays is a better form of testing than running a benchmark that has consistency issues in a game (CyubeVR) that he doesn't and never will play imo. For him, yes, but not for the rest of us not into the same sim. I don't mind subjective experiences. There are a lot of Index users saying they got great performance with their 1070 cards, and while that's certainly nice, I'd appreciate more objective results. Can of course be from other VR benchmarks too, but we don't got a lot to choose from, and some do not use engines from real VR games. 2c. Well until someone builds a better VR Benchmark, subjective is pretty much all we have. CyubeVR's benchmark has too many issues to be looked at as definitive or objective proof of anything. It's a benchmark made by a solo developer...with minimal testing and minimal options and practically no updates. Hell it doesn't even work in December because of a bug he's too lazy to fix LOL. It's a decent test if you plan to play CyubeVR. But we'll just have to disagree on it being an objective marker of much.
|
|
|
Post by Rune on Jan 14, 2023 11:36:27 GMT
For him, yes, but not for the rest of us not into the same sim. I don't mind subjective experiences. There are a lot of Index users saying they got great performance with their 1070 cards, and while that's certainly nice, I'd appreciate more objective results. Can of course be from other VR benchmarks too, but we don't got a lot to choose from, and some do not use engines from real VR games. 2c. Well until someone builds a better VR Benchmark, subjective is pretty much all we have. CyubeVR's benchmark has too many issues to be looked at as definitive or objective proof of anything. It's a benchmark made by a solo developer...with minimal testing and minimal options. It's a decent test if you plan to play CyubeVR. But we'll just have to disagree on it. Subjective I found my results in the OpenVR Benchmark to very closely predict my performance in the majority of my demanding VR games. I went from 35.5 fps to about 80 fps in the OpenVR Benchmark going from 1080 to 3090. And nearly all my games went from 45 fps to 90 fps, but could not push ss a lot more than previously, before I went back to 45 fps. Oculus Tray Tool confirmed about 100% and little more improvements - similar to the OpenVR Benchmark. BabelTechReviews got quite similar results from 10 VR games in average and comparing to OpenVR Benchmark - about 80% performance increase from 3090 to 4090. So OpenVR matched common VR game performance well. Sure, testing 20 games before and after would be better, but I doubt DBurne would be interested in that :-)
|
|
|
Post by Pyroth309 on Jan 14, 2023 11:45:54 GMT
Well until someone builds a better VR Benchmark, subjective is pretty much all we have. CyubeVR's benchmark has too many issues to be looked at as definitive or objective proof of anything. It's a benchmark made by a solo developer...with minimal testing and minimal options. It's a decent test if you plan to play CyubeVR. But we'll just have to disagree on it. Subjective I found my results in the OpenVR Benchmark to very closely predict my performance in the majority of my demanding VR games. I went from 35.5 fps to about 80 fps in the OpenVR Benchmark going from 1080 to 3090. And nearly all my games went from 45 fps to 90 fps, but could not push ss a lot more than previously, before I went back to 45 fps. Oculus Tray Tool confirmed about 100% and little more improvements - similar to the OpenVR Benchmark. BabelTechReviews got quite similar results from 10 VR games in average and comparing to OpenVR Benchmark - about 80% performance increase from 3090 to 4090. So OpenVR matched common VR game performance well. Sure, testing 20 games before and after would be better, but I doubt DBurne would be interested in that :-) Except they likely went higher than 90...which you can't see. That's not a ringing endorsement of accuracy and is still subjective imo. But, with the tools we have is a fair way to get a picture of what's going on. The problem I have with that benchmark is that when I run the G2 in it, it gives me some super low number like 30 fps....I never got lower than 70 in the games that I played with it unless I cranked Supersampling way up. If I really wanted it, 90 was easy to obtain just coming down on SS. It's a subjective meaningless number it spits out. What matters to me is what FPS can you get while still maintaining a quality image. High resolution panels assist with that. Subjective tweaking is the only way to achieve a good experience for yourself imo. What I would like to see is a benchmark that uses an in game render and also locks your CPU and GPU or at least reports them to a certain speed AND locks everyone to the exact same resolution at a minimum for like HMD's IE 2 indexes need to be at the same resolution always. Ideally we could have some presets like 3 to account for peoples machines. Low res, Medium and High. Then we might have something that could be considered Objective. Until then... it's just a facade. PS sorry Rune accidently hit edit on your post instead of Quote. I didn't actually change your post.
|
|
|
Post by Pyroth309 on Jan 14, 2023 12:03:30 GMT
Here is a link to babel's test. babeltechreviews.com/rtx-4090-performance-45-games-vr-pro-apps-benchmarked/4/I would wager the reason they didn't use the CyubeVR benchmark is the same reason I don't like it. There's way too much room for error in the way it calculates. After quickly browsing the results... I love that he's able to provide Unconstrained FPS for his testing. That gives a great picture of what's going on. 192 in F1 2022.... That said, I agree with him that we need frametime plots for it to have meaning because VR FPS is so volatile. Which is what I meant above about you likely being higher than 90. Because even a short interval of lower FPS will drop you back to 45. I need to stop looking at this or I'm going to end up with a 4090 in my cart. I'm holding out for a 5090 lol....or at least a 5080. Don't have the VR playtime to justify it right now. Especially with PSVR 2 coming next month.
|
|
|
Post by Pyroth309 on Jan 14, 2023 12:30:29 GMT
And Rune, sorry if I come across as rude or condescending. Not trying to be. I know I'm blunt and like to get my point across sharply. I respect your opinions and enjoy engaging with you to both find flaws in your arguments...and in mine. I find our various discourse on performance enjoyable and enlightening.
|
|
|
Post by Pyroth309 on Jan 14, 2023 12:50:00 GMT
Back to that Babel test-
Now this is how a review should be done. A proper Review. He gives you the HMD, what resolution it's set at, frames, frame times, A controlled repeatable testing environment and the in game settings he obtained the numbers at, the specs of his machine and card and even goes beyond that. Very well done.
So far his results match a lot of my own findings with Reverb G2. Assetto Corsa competizione is a bear to run on Ultra lol. My G2 had no chance on it and I actually preferred Index for the short time I played that game because I couldn't get the balance acceptable. I opted for more smoothness and higher settings on Index.
But he's showing why G2 won for me with Elite Dangerous, No man's sky, Project Car's 2 and many other games. There's enough headroom to run it at 100% or higher on some games and especially if you are ok with either synthesized frames or lower than 90 fps. I personally don't mind running with no motion smoothing until it gets down to under 60. Then it's too noticeably choppy. But above 60 I'm good. If it dips lower... At that point I either want Motion Smoothing or to lower settings to get FPS up. But this number varies for everyone. It also varies for me on the game type. Faster motion games need more FPS etc. Flight sims that have a slower motion...I'm good at 60 all day.
Anyways, just some more commentary.
|
|
|
Post by Rune on Jan 15, 2023 16:21:48 GMT
And Rune, sorry if I come across as rude or condescending. Not trying to be. I know I'm blunt and like to get my point across sharply. I respect your opinions and enjoy engaging with you to both find flaws in your arguments...and in mine. I find our various discourse on performance enjoyable and enlightening. No worries, lol - just happy to be able to discuss some subjects that I do care about. Maybe also better to illustrate - BabelTech found: www.reddit.com/r/ValveIndex/comments/y1xf8u/using_index_the_rtx_4090_is_average_75_faster/Not a lot of games, and 4090 was average 75% faster than 3090 in VR games. - and the usual OpenVR Benchmark 4090 Index res 100% results are about 85% better than mine: www.reddit.com/r/ValveIndex/comments/y38541/openvr_benchmark_with_the_rtx_4090_and_index_not/Best would be to get 20+ different VR games tested, but for now - and with little else to choose from - the OpenVR Benchmark has been a great tool for me. I'd like to see 200+ fps using Index res 100% with my next gpu. I used the OpenVR Benchmark to choose the 3090 instead of a 3080 - so far I've had no reason to regret that :-)
|
|
|
Post by Rune on Jan 15, 2023 16:30:41 GMT
Back to that Babel test- Now this is how a review should be done. A proper Review. He gives you the HMD, what resolution it's set at, frames, frame times, A controlled repeatable testing environment and the in game settings he obtained the numbers at, the specs of his machine and card and even goes beyond that. Very well done. So far his results match a lot of my own findings with Reverb G2. Assetto Corsa competizione is a bear to run on Ultra lol. My G2 had no chance on it and I actually preferred Index for the short time I played that game because I couldn't get the balance acceptable. I opted for more smoothness and higher settings on Index. But he's showing why G2 won for me with Elite Dangerous, No man's sky, Project Car's 2 and many other games. There's enough headroom to run it at 100% or higher on some games and especially if you are ok with either synthesized frames or lower than 90 fps. I personally don't mind running with no motion smoothing until it gets down to under 60. Then it's too noticeably choppy. But above 60 I'm good. If it dips lower... At that point I either want Motion Smoothing or to lower settings to get FPS up. But this number varies for everyone. It also varies for me on the game type. Faster motion games need more FPS etc. Flight sims that have a slower motion...I'm good at 60 all day. Anyways, just some more commentary. Seems that many observed reprojections in Call of the Mountain, interesting if we can get solid 90 fps - or if Sony used 120 Hz defaulting to 2 x 60 fps. I'm using 2 x 60 fps in Squadrons, because it needs 60 Hz to be smooth (probably caused by the engine's console origin). I'm not happy about that, but it works ok-ish. For new PCVR hmds, I'm not going to accept less than 90 fps in the majority of games. Btw, Levitation has been fixed, I just checked the game, it's awesome, I just got some serios spanking, lol. And it works in solid 90 fps with Ultra - using Index. I'll be back later this evening.
|
|
|
Post by Pyroth309 on Jan 15, 2023 16:56:16 GMT
Back to that Babel test- Now this is how a review should be done. A proper Review. He gives you the HMD, what resolution it's set at, frames, frame times, A controlled repeatable testing environment and the in game settings he obtained the numbers at, the specs of his machine and card and even goes beyond that. Very well done. So far his results match a lot of my own findings with Reverb G2. Assetto Corsa competizione is a bear to run on Ultra lol. My G2 had no chance on it and I actually preferred Index for the short time I played that game because I couldn't get the balance acceptable. I opted for more smoothness and higher settings on Index. But he's showing why G2 won for me with Elite Dangerous, No man's sky, Project Car's 2 and many other games. There's enough headroom to run it at 100% or higher on some games and especially if you are ok with either synthesized frames or lower than 90 fps. I personally don't mind running with no motion smoothing until it gets down to under 60. Then it's too noticeably choppy. But above 60 I'm good. If it dips lower... At that point I either want Motion Smoothing or to lower settings to get FPS up. But this number varies for everyone. It also varies for me on the game type. Faster motion games need more FPS etc. Flight sims that have a slower motion...I'm good at 60 all day. Anyways, just some more commentary. Seems that many observed reprojections in Call of the Mountain, interesting if we can get solid 90 fps - or if Sony used 120 Hz defaulting to 2 x 60 fps. I'm using 2 x 60 fps in Squadrons, because it needs 60 Hz to be smooth (probably caused by the engine's console origin). I'm not happy about that, but it works ok-ish. For new PCVR hmds, I'm not going to accept less than 90 fps in the majority of games. Btw, Levitation has been fixed, I just checked the game, it's awesome, I just got some serios spanking, lol. And it works in solid 90 fps with Ultra - using Index. I'll be back later this evening. For me, 90 fps is not worth the graphics and clarity I have to give up for a tiny amount of smoothness unless I'm playing something that requires extreme precision at high speed...like an online shooter or something. There's almost no benefit to it in the vast majority of games. And believe me, I've used my Index on 144, 120, 90, and 80 many times. 95% of the time I'm at 80 hz and wish I could go down lower than 80. Give me the graphics but to each their own.
|
|
|
Post by Pyroth309 on Jan 15, 2023 16:56:25 GMT
Back to that Babel test- Now this is how a review should be done. A proper Review. He gives you the HMD, what resolution it's set at, frames, frame times, A controlled repeatable testing environment and the in game settings he obtained the numbers at, the specs of his machine and card and even goes beyond that. Very well done. So far his results match a lot of my own findings with Reverb G2. Assetto Corsa competizione is a bear to run on Ultra lol. My G2 had no chance on it and I actually preferred Index for the short time I played that game because I couldn't get the balance acceptable. I opted for more smoothness and higher settings on Index. But he's showing why G2 won for me with Elite Dangerous, No man's sky, Project Car's 2 and many other games. There's enough headroom to run it at 100% or higher on some games and especially if you are ok with either synthesized frames or lower than 90 fps. I personally don't mind running with no motion smoothing until it gets down to under 60. Then it's too noticeably choppy. But above 60 I'm good. If it dips lower... At that point I either want Motion Smoothing or to lower settings to get FPS up. But this number varies for everyone. It also varies for me on the game type. Faster motion games need more FPS etc. Flight sims that have a slower motion...I'm good at 60 all day. Anyways, just some more commentary. Seems that many observed reprojections in Call of the Mountain, interesting if we can get solid 90 fps - or if Sony used 120 Hz defaulting to 2 x 60 fps. I'm using 2 x 60 fps in Squadrons, because it needs 60 Hz to be smooth (probably caused by the engine's console origin). I'm not happy about that, but it works ok-ish. For new PCVR hmds, I'm not going to accept less than 90 fps in the majority of games. Btw, Levitation has been fixed, I just checked the game, it's awesome, I just got some serios spanking, lol. And it works in solid 90 fps with Ultra - using Index. I'll be back later this evening. For me, 90 fps is not worth the graphics and clarity I have to give up for a tiny amount of smoothness unless I'm playing something that requires extreme precision at high speed...like an online shooter or something. There's almost no benefit to it in the vast majority of games. And believe me, I've used my Index on 144, 120, 90, and 80 many times. 95% of the time I'm at 80 hz and wish I could go down lower than 80. Give me the graphics but to each their own.
|
|
|
Post by Rune on Jan 15, 2023 18:20:34 GMT
|
|
|
Post by dburne on Jan 15, 2023 23:32:13 GMT
I still find it hard to believe the big leap in performance the 4090 has over the 3090 cards. Absolutely amazing and also combined with much better air cooling as well. My 3090 ran at 70c and my 4090 barely hits 60c on occasion, and that is with a temp controlled variable speed on the fans. The 3090 I would game with it at 100% fixed on all three fans. Nvidia must have worked some serious majic with that cooling design on the 4090 card so and also runs considerably quieter than my 3090 did.
IMHO Nvidia hit a home run with this design. Only thing I did not care for was the power connecters on the side of the card rather than the rear. Can make for a pretty tight fit in some not so roomy cases. Thankfully I have a huge case so not an issue on my end. I wonder if the power placement on the GPU had anything to do with the better cooling? Not sure it would really but it is curious.
|
|
|
Post by Pyroth309 on Jan 15, 2023 23:43:46 GMT
Ah ok I stand corrected. I never saw that chart when I went through the article so it either didn't load or I didn't see it somehow. I agree with that. I've never had a contention with OpenVR's correspondence of power within the same HMD. It's going to go up relatively as anything else dependant upon how much GPU power you put to it. My issues with it has always been trying to use it to compare HMD's across different systems and configurations. And also I hate the low useless numbers it provides that people try to justify against HMD's as being unusable. (not talking about you, I see it on forums). It's only unusable in that specific scenario at those settings. The 90 FPS setpoint in VR was put in place on the CV1 because of the thought that it reduced latency and thus reduced motion sickness. But that was a flawed theory because there's other factors to motion sickness and frametime. For example if your GPU is running wide open and struggling to hit 90 your frame time is going to be a lot higher than someone who could really achieve 150 fps and is cruising while hitting 90. This is why they brought the Quest down to 72 base fps and people are able to use it fine without getting sick. (Of course not everyone, some people just can't handle VR in general). I personally do not like a frametime above 17ms so I will back down my settings even if my FPS is good enough. I've never gotten motion sickness from my quest or my G2 dipping down under 60. But.. From 60 and under..it starts to feel choppy, especially in my hands in games and that's really my bottom number personally. I actually aim for 75fps and crank the graphics until I hit it then go play. Gives me some buffer for dips. Dips to 60 are fine...staying at 60 full time is not unless it's a slow moving game like a flight sim. So all that being said, these benchmarks are a good indicator of potential power but really do not determine which games are playable in certain HMD's. As you can see from the benchmarks, he gets wildly different results between games. For you Rune, I know you like max settings and max resolution and high fps...and are in that 0.0001% of gamer. Yea fine an Index makes more sense. But for those of us who like everything to be clear and no visible SDE, 4k HMD's are great with a 3090+ and should be incredible with a 4090+ regardless of what OpenVR says.
|
|
|
Post by Pyroth309 on Jan 15, 2023 23:56:41 GMT
I still find it hard to believe the big leap in performance the 4090 has over the 3090 cards. Absolutely amazing and also combined with much better air cooling as well. My 3090 ran at 70c and my 4090 barely hits 60c on occasion, and that is with a temp controlled variable speed on the fans. The 3090 I would game with it at 100% fixed on all three fans. Nvidia must have worked some serious majic with that cooling design on the 4090 card so and also runs considerably quieter than my 3090 did. IMHO Nvidia hit a home run with this design. Only thing I did not care for was the power connecters on the side of the card rather than the rear. Can make for a pretty tight fit in some not so roomy cases. Thankfully I have a huge case so not an issue on my end. I wonder if the power placement on the GPU had anything to do with the better cooling? Not sure it would really but it is curious. That's weird because my 3090 runs cool as ice. Fans never even hit 30% and are off most of the time. But I know you had an overclocked card from EVGA. They must not have done a good job on the cooling solution.
|
|