DeltaMike

Not an issue for me, but then I prefer a light touch on the throttle. Friction is just barely tight enough to keep it where I put it, and I just have fingertips on the throttle. For some reason the throttle base action isn't quite as buttery smooth as the joystick base, which is exquisite. Doubt this would be obvious if you prefer a full fisted approach, but note, the ergonomics are off for that kind of grip.

Ive done two builds in the last month. It's hard to find GPUs these days 6900xt is on the market, for a price. 6800xt, if you can find one, is fetching the same price as 6900xt. And you literally have to win a lottery for the opportunity to buy a 3000 series GPU. Speaking of which you should get the Newegg app and start entering the lottery every day. I finally won in the chance to buy a 3060 for my son in law but it took several weeks and still haven't scored anything I'd want. Finally got my 6900xt off eBay. The scalper erred by ending his auction at like 4am eastern time. ~$1350 shipped. Mobos are in short supply also. I wound up getting a GB x570 cuz it was the same cost as a b550. Asus mobos are nuts. As for storage, DCS will take up 500GB easy. If you start downloading content for msfs or xplane you're gonna wish you had 2 or more TB. Although, a mechanical hard drive works well enuf for terrains and might be an inexpensive option for ya Did they ever fix the spotting problems in 4k?

Note you don't have to use simapp, except for calibration, flashing firmware, and deciding whether you want your sliders to act as axes or buttons. Once you've done all that, I'd close simapp (and deactivate in startup folder, you can get at that using task manager) DCS recognizes your hotas, you can program it in game. Wags and grim reapers have YouTube vids on how to do this, and chucks guides also addresses binding Best to get used to this, you'll be doing it a lot. Modules are like pokemon, gotta catch em all. You'll quickly see the benefit of coming up with a consistent scheme as much as possible. So, you'll want to keep basics like comms, brakes, flaps, gear the same from module to module. Same goes for weapons select and sensor control. Muscle memory counts here.

Hard to squeeze out much in the way of single core performance with an overclock. Multicore is another matter bit is irrelevant to DCs. How much ram ya got? 32g minimum for MP. As another poster mentioned make sure xmp is on. If you have to buy RAM don't go nuts, 3600mhz is plenty (and even 3200Mhz should work) Most "graphics" settings affect CPU more than GPU. You may have to turn off shadows, turn down clouds and water, and might need to put vis range as low as medium and even then you may have to get off the deck before things run smooth. In terms of coolers, I put in a Mugen 5 on a build the other day. That things a beast.

Undersampling does seem to affect spotting distance. At 46% (approx 2200x2200, what many mistake for native resolution) I can just barely see a tanker at 16nm. 10 to be able to join up visually ie without a radar lock. Doesn't seem to matter whether that's set in global or per app settings. At 100% I can see the thing at 20, and can join visually from 16. Which is fine, I can run full res without msaa and by taking it easy in the cloud department. Speaking of which, I see the clouds don't dance and twirl around like they used to. Not gonna miss that, except to the extent it reminds one of a goa gil concert. (Believe me, a little of that goes a long way)

So far so good, you guys are the bomb. Couple of additional observations -- -- Seems like there are a couple of different artifacts we are going after. "Judder" is a big one I think, which near as I can tell has two definitions, both relevant: The strobe effect that comes from panning a camera too quickly relative to FPS. In a flight sim, that would be the stuttering when you look out to the side, for example. Or when you're doing a snap roll. Mismatch between FPS and refresh rate. Interesting discussion here from a film-geek's perspective. As for the first, I'm drawn to the conclusion that frame rate does matter. Not so much for motion sickness (which is my primary interest) but for judder. Suspect the second is also relevant to us. Between the two, there's a strong argument I think for using motion reprojection, and perhaps for keeping it locked on. -- Interesting, to me anyway, how undersampling (in the per-app settings) doesn't seem to degrade the image very much. I was initially against undersampling, figuring it would have been easier and a little cheaper to buy an Oculus. But, I'm surprised how good things look, and how well things work at 70%. Flabbergasted actually. It's weird. Interested to try out the FSR mod. -- Looking back at the MSFS 6900 vs 3090 vid, the 6900xt could only keep up when undersampled. At native resolution, the stronger card pulled out ahead, as we would expect. I wonder if the additional "smoothness" of the 3090, at the beginning of the video, wasn't the fact it was running quite a bit closer to 30fps, which I guess would reduce the second kind of judder, no? ----- Here, I diverge mmmm somewhat from Thud's guide. I would not go so far as to say framerate doesn't matter. Oculus taught me to pick a framerate, and tune to that. Probably still good advice. That's part of what held me back from buying a new GPU for my Rift S, with the question being, what would it take to drive that thing at 80hz? Nothing that was on the market at the time. Likewise, that was part of my theory in going with the 6900XT to match up with the G2. I wouldn't mind having a 3080ti, if I could get my paws on one. But in reality, there's really no advantage in being able to run say 65fps vs 47fps. For all practical purposes, it doesn't change game play. (Unless you can stand to run your headset at 60hz, and even then, good luck in MP.) By the same token, there's no particular reason to favor the 6900XT vs say a 6800XT or a 3080, just a matter of what you can get your hands on. Good news is, it seems the 6900XT can hold 45fps easy as pie. So, for me anyway -- Tune the card (and keep it cool) Turn on SAM; otherwise, there really isn't anything in bios or Adrenaline that matters PD 1.0, MSAA no more than 2x; otherwise it's kinda hard to screw up DCS settings (in SP anyway) Turn motion vector on Start with SS Global 100% and SS Per App say 80%, enjoy your first flight and then tweak from there Thanks, Ted! You saved me $300! Sort of! (As I'm not sure if Newegg really had a 3080ti to sell, or if they were just jiving me)

CPU and GPU frame times are in the teens. If fpsVR gives you a total frame time, I didn't see it. Down low to the ground I know for a fact I wasn't holding 50fps tho.

OK well I got impatient and tried it in 2.5. Three trials: Radeon chill locked at 45 (min and max), motion vector off "Frame rate target control" set at 48 (hard to get it right on 45), motion vector off (Settings... Graphics... Advanced) Adrenaline set at defaults, motion vector locked on Regardless, my frames were in the range of 40-55 (with my precious shadows) Observations: Looking out to the side, zipping along close to the ground, it's stuttery no matter what you do. I've noticed the same thing at the movie theater watching at 30fps, so I think you're right about that. 30-40 fps isn't quite enough to keep up with that kinda motion. The worst option is to just let the G2 runs how it wants to, it's stuttering all over the place. I can see why people get frustrated with it. Between the three options above, they are roughly equivalent to me. Each is associated with its own weaknesses. In general, looking outside, motion vector works the best. Reading cockpit instruments, Chill worked the best for me. ASW has its issues -- propellers are just comically bad -- but motion vector is even worse, even just looking around the cockpit makes things jump and shimmer. That part needs some work. I guess we just need to pick our poison. Appreciate your thoughts re: 60Hz. I'll give that a try. Gotta go throw up first tho.

Yeah but the question is, what's the active ingredient? The frame rate cap, or the synthetic images? A lot of DCS'ers hate the synthetic images and turn em off. Which works with Oculus, at least in the Rift/S era. Back in the day, Oculus only gave you two choices, basically. 45 or 90fps (or 40/80 with the Rift S iirc). Challenge was to maintain 45fps on the ground, and there was basically no way to maintain 90 anywhere in DCS so with a nice 1080ti and the right settings you could make it work. (We drove the developers nuts. "No matter what you do, my framerate is still only 45!") Not so with the Reverb, evidently. Which near as I can tell, runs at whatever frame rate it wants. If you go back and look at that tuning series I linked above, you can see the problem with the 6900xt, which is "spikes" in your render time. The average render time ain't bad; in MSFS for example, the 6900XT can keep up with a 3090 on average. Question is, what to do with the spikes. How do you smooth it out. That's the problem Ted is trying to solve, I think. ----- Reverb is up and running. I did get the "black screen" problem, fixed it by unplugging my HDMI monitor. I forgot to take the film off the lenses and it was still better than my old Rift S. Pretty awesome, but I'll state for the record, the aggregate user interface is a complete cluster****. At default settings everywhere except for the undervolt and memory timings, there was a lot of stuttering in DCS 2.5 but that was with my old shader files, and high quality shadows (which I fear may still be a bridge too far). Presently playing around with environments in Steam Home while a years' worth of DCS updates download at a glacial pace.

Just because you have 22ms between frames doesn't mean your frame time is 22ms. You can theoretically paint the picture in 11ms, and only display it 45 times per second. Several things contribute to discomfort. It depends on what kind of game you're playing, for example. Frame rate isn't one of them. Minecraft will make ya puke every time, even at 90fps. Way I look at it, you're gonna watch Top Gun at 30fps, and you'll like it. You could also watch it at 90Hz, but if it kept dropping frames, stuttering and flipping between 90Hz and 30Hz, it could get pretty annoying. That's the problem we are trying to solve here. Think about it. If only 1% of your frames are rendered in say 21ms, at 45fps you'll never drop a frame. At 90fps you'll drop a frame every second. Oculus figured that out a long time ago, it's always locked at 45fps. They specify a 90Hz monitor because of the response rate. How quickly you can turn a pixel on or off affects latency, which is the main factor that makes you puke. You don't want to be turning your head while the world is standing still, or (heaven forbid) the opposite. That'll send you to the Gents-or-Ladies quickly. Question is, what to do with this hot mess. I feel OP has a key insight; we need to figure out how to cap this beast at 45fps. Either by forcing motion reprojection, or by setting limits in Adrenaline (and there are a couple of places to do that). Throwing money at the problem ($3K for a 3090? Ouch!) is an answer, but it's not the answer. There is no way anybody is gonna run DCS in VR with a consistent 11ms frame time. It's not just that we are driving the equivalent of two 4K monitors; the CPU has to render each frame twice. We gotta figure something out, and we will. (Note, I've figured out how the human brain works, but not how a GPU works, so take this with a grain of salt)

So far so good, I have the beast running. It does not appear that "one click" undervolting does anything. Followed this guide to manually undervolt and temps went down 10%, performance up 3% under Unigine Superposition. Wasn't able to push power limit as I can't keep the durn thing cool but it's better than nothing and puts me in the pack. Pretty sure fan profile had the most effect. Starting with anti-lag, fast VRAM timing, and SAM. That's about it. Next to wrangle the G2 setup.

Ted, you are the absolute boss. You had me doing the Newegg shuffle every day looking for a 3080ti. No joy so far (but I did get the chance to buy a 3060 for my son-in-law, so, it does happen). Unless I win Monday morning, I'm slapping in the 6900XT Monday night. Looking forward to comparing notes. I like your strategy of capping FPS at 45. I've read articles suggesting it's not precisely FPS that keeps you from puking, it's latency. Everything between the time you pull the trigger, and see the results in your headset. Which is primarily a function of render time, assuming your USB ports are working Speaking of which, I'm assuming AMD has fixed things at the motherboard level, no? Have you been able to run PCIE 4? Agree with your thoughts re: the GPU-mounted USB-C port. Which I've only seen it on reference models. Agree completely with your thoughts on thermals. Glad to see the "one click" undervolting method actually does something. It's been a couple of years since I messed around with VR settings in DCS. Before the new clouds and tress, and right at the beginning of VR optimization. From looking at published results since then, I get the feeling things are quite a bit less sequential, and quite a bit more parallel, making it difficult to sort out what's affecting what. Back then, it was clear that MSAA, along with total number of pixels, was all on the GPU. With that in mind, the fact you're able to run native resolution with any MSAA at all is a testament to the 6900XT and all the blood sweat and tears that went in to your tuning efforts. Standing ovation for that. That's yuge. As for the other settings, where we are primarily trying to regulate the CPU work load, I imagine there are tradeoffs. My plan is to set vis range to medium, which I personally feel is realistic. With the old trees, that allowed you to turn the settings down without the weird "pop-up" effect; unknown if that trick still works. But I agree, for jet pilots, trees and grass is kind of a waste of CPU cycles. For water, medium worked well enough for me back in the day. I'll try all that and see if that gives us any wiggle room with clouds and shadows. (While I agree flat object shadows are fine, I shudder to think how much money I've spent on those doggone cockpit shadows, so, I want em.) I'm also planning to do a study of spotting distance, to see how undersampling and anti-aliasing affects that in the G2. My theory is, native resolution is gonna be as good as it gets, but we will see. Thanks again boss. I may come whining sometime during the next week.

It's gonna be tough. I'm in the middle of building a rig for G2, although before I succumbed to mission creep I was looking at quest 2, having flown the rift S successfully back in the day. Look up the thermaltake reactor 370. That should be about the right spec to get ya started, and be advised, it's tough to beat that price even if you win the Newegg Shuffle and source the remaining parts on your own, and build it yourself. So, figure something in the neighborhood of $2300 us... If you're not against going used, I would not personally be against sourcing a 3600x on a b450 mobo. Can't think of a used gpu I really like though. The 2000 series didn't improve much, if any on the 1080ti, which was great on the rift s, in older versions of DCS but that was then. Considering the price the used ones are going for I don't feel it's a good investment. GPU-wise I just don't see a good alternative to a 3070 or better

Well, you could try turning down your "graphics" settings to take some of the load off your CPU. In quotes because pretty much every setting, other than anti-aliasing, affects your CPU at least as much as your GPU. I'd look at vis range first. Don't expect much though. To give you an idea how important RAM is... Back in the day, using the same CPU, I was driving a Rift S at 40Hz. Assuming you're playing at 60 hz on your monitor, you have ~17ms to render the scene once, I had 25ms to render the scene twice and I was able to make it work. You have plenty of CPU. And I doubt upgrading your GPU will help much in MP. For that generation of hardware, if it worked in SP it worked in MP. You need RAM my brother. Don't be sad. DCS is still cheaper than golf, and not nearly as much a waste of time.

Yeah. VR is all about the latency, everything that happens between when you click the mouse, and see the result on your screen. If you're having usb problems that's an issue for pretty much everything except the mouse innards I guess, which fortunately seem to be working at present. Between that and power spikes, VRAM management, PCIE issues (did I forget anything?) I guess AMD's spanking new chipset is a work in progress. Intel's consumer-level article on VR optimization here Here's how I look at it. Of the three things we might want -- 1. Resolution (MFD clarity, spotting)* 2. Eye candy (clouds, shadows) 3. Anti-aliasing (supersampling, msaa) -- with any of the top tier cards (3080, 3090, 6900xt) and a top tier headset (eg Reverb, Pimax) it seems safe to say you can get your choice of two, but perhaps not all three at the same time. My only question is, is it really a meaningful test, to try to max out all three on a busy PVP server. There, you're inclined to turn your anti-aliasing down, if not off altogether to make spotting easier. And of course shadows and trees are irrelevant, in fact shadows could be a distraction if you're really competitive. You get absolutely nothing by setting vis range to anything more than medium, and run the risk of being CPU bound if you do. I'd run the MP test without MSAA, with vis range medium, and with supersampling at 100% (or to be more precise, leave the barrel distortion correction in at 140%, but no more than that -- set the *sliders* at 100% and ignore the pixel count). Granted, my money is still on the 3080 -- suspect its memory bandwidth gives it an advantage no matter how you slice it -- but I think it would be a more relevant test, and speaks a bit more to what some of us perceive to be the weakness of the 3000 series cards; that we are paying (a lot) for technology we aren't realistically gonna fully leverage in DCS. Ebay prices as of first week of Jan 2022: 6900xt $1350, 3080 ~$1800. You make a strong argument, that the 3080 is the superior card. My question is, is the delta worth it? ---- *ie, not having to under-sample to handle the load

Thank you so much for posting your results. I always had this idea in my head that the CPU did its thing, then the GPU did its thing, then we see a picture. I think that work flow has been diced up quite a bit in the process of optimizing VR, which I think means CPU and GPU frame times should covary. I think. Back in the day, it was pretty clear that almost every "graphics" setting in DCS affects CPU as much as GPU, with the notable exceptions being supersampling and msaa. I think what your data shows is, the 6900xt can hold it's own in the supersampling and msaa department, for example when you're just flying around by yourself. But when things get busy it has a hard time keeping up with all the stuff the CPU is demanding of it. I think it follows, then, that tuning the 6900xt may be a matter of limiting the CPU workload, for example by reducing vis range or trees. Good news for people who like their anti-aliasing I guess. Unfortunately my computer is in a million pieces right now so I can't test this hypothesis, but I suspect I'll do what I've always done -- set vis range to medium, turn off shadows and trees in MP, and just be thankful I can read my instruments now 5600x, 6900xt (at $1350, the least bad buy on the market imo), and G2, soon.

Hot tip: ergo360.com has the ironically-named "sliders" that'll work for any chair. Little feet that replace the wheels. That...um... keep it from sliding. They can fit any chair, including Ikea. Which I guess is a thing now, gaming-chair wise. $69 for a mesh gaming chair is a bargain imo. Cool thing is, Monstertech can fit em with Hotas mounts for only $350. See how you can save money in this hobby by being a smart shopper? DCS is like Kmart, there's a bargain on every aisle

Wags hinted they might let a third party take it over. It's not a popular map, would need support from the community to set up some meaningful MP scenarios. It's been done, I had the privilege of working in a flight school based on that map a while back. Heavily Navy oriented, which is why I was asking. A training map, by nature; but surely there's some interest in that. And sight-seeing; it's a striking landscape irl and with cosmetic upgrades at the level we've come to expect from Ugra I imagine it would be something to see. Being 80% done already shouldn't hurt anything.

Yeah. Personally I feel it would be helpful if people could post their CPU and GPU frame times, which is arguably relevant even for pancake mode. eg http://www.mvps.org/directx/articles/fps_versus_frame_time.htm Granted, this article is about software optimization. But still relevant I think. When you use the same GPU to compare one game against another, yes that's about software optimization. When you compare one GPU against another in a rather uniquely unoptimized game -- which applies to DCS and maybe all flight sims, as they are up to something subtly different than your average mass market game -- it's about how a particular card's features or engineering happens to work in that particular environment. Which is difficult to predict. I can think of a couple of outliers. The Vega56, which punched way above its class in DCS back in the day. Likewise the 1080ti these days. I think maybe frame times might get us a little closer to figuring out how efficiently a given card works for DCS, rather than gross FPS. For example, see this: In MSFS, interestingly, the 6900xt spanked the 3090 in this test, FPS wise. In VR, which is a pretty rigorous test, as each frame has to be rendered twice at high resolution. As well as in flat screen mode, where the difference was even more dramatic. Note the reviewer still preferred the 3090. "Just seems smoother," he says. I imagine he's exactly right; if he had reported frame times instead of FPS -- and more specifically, some measure of frame time variability -- I'll bet he would find the 3090 had quite a bit less variability. "Frame rate only tells half the story." Anandtech has written extensively on this, can't quite put my finger on the editorial I read yesterday but arguably this is relevant to pancake mode, albeit at a much more subtle level. As a new 6900xt owner, it gives me no pleasure to say this, except to the extent I saved about $800. I imagine I could smoke the 3080, FPS-wise, but wouldn't be surprised if the 3080 gave a better and smoother experience at 4K, flat screen or no. Beyond that, who cares about frame rate? I'm gonna watch Top Gun Redux at 30fps and I imagine I'll enjoy it just fine. Likewise, in VR anyway, whether I'm in DCS or MSFS or whatever, I'll take anything over 30. With good enough frame times -- or more specifically, low enough latency -- I'm confident I won't puke. Whatever frame rate you want (60? 90?) the question is, what's the experience like *at that rate*. I think frame time might get you closer to the nugget of the thing. It's not so much the dude needs to buy a headset, and let himself in for that particular brand of life-complexity. But that maybe we should all rethink what variables we want to monitor. Yes? No? Also, MSFS makes you appreciate DCS, don't it?

https://teamwadafak.com/how-to-run-dcs-dedicated-server-on-linux/ I understand there's an argument for playing on your own server. Offload some of the work to somebody else's cores, right? What about say a NUC? What about a VM?

^Awesome. I thought they were more than that... that's about what a Big Bertha driver costs, and unlike DCS, golf is a waste of time. I've only solved half my problem, with VKB pedals. Which oddly aren't much more than CH or TM plastics. Great value; I figure I can find a brake axis somewhere. Perhaps with a new VKB stick! See how ya save money in this hobby, using logic?

I will admit, VTOL *might* hold my attention for a moment or two but, but I really need to be working on my understanding of geometry after the merge and maybe not quite so much how, precisely, I enter brc into the UFC, ya know? Check out xplane, they are all over the VR controller thing.

VR controllers are clunky and break immersion. Typically we rely on a pointing device to interact with the cockpit and everything else is by feel. It's possible to put button boxes in about the right spots and work on your muscle memory. I've always thought a landing gear lever would have utility, otherwise I don't prefer to search around for stuff. That's just me tho. Voice attack is immersive. When you fly in real life it's common to say your checklists out loud. "gear down, flaps full" etc. And of course anything involving a radio call is very immersive. "Wheels up." "Fence in." "Ready pre-contact" etc. The point ctrl device is really interesting too. If you go to the guys website, you'll see he has plain plywood panels laid out like an instrument panel. When he's working switches and stuff he touches the panels. Looks bizarre from the outside but it's brilliant when you think about it. Haptics on the cheap with no muscle memory required, no link to the default world. Which is the grail we seek, I think.

I'm starting over with a new build. About the only thing available, to me at least is 6900xt so I'm doing a build around that. The market is kinda nuts right now, ain't it? I'm flipping over to reverb but kinda miss oculus, quest 2 strikes me as a great deal and the software is a pleasure to use. So many gpu's (2070s for example) would be great with that rig but even the used market is nuts. Sold that Vega for $500. Crazy.

How are your thermals?