I’m beginning to suspect that anyone with a recent M.2 purchase is sworn to secrecy. Did you get a knock at the door from the SSD men in black @BeachAV8R? 
1 Like
Yeah…I have to say, I would not be able to run cables and get everything into a case the way the guys at MSI did. It really does look stunning on the inside. My other computer, which I largely built, has SSDs hanging from SATA cables like…well…hanging SSDs.
She gets an i7-4770k, a 1080, and a Rift…! What’s not to love? She’s going to be a Beat Sabre champion!
3 Likes
So the really good news is that X-Plane runs perfectly fine from the new 4-bay network drive via USB-C. Which is a good thing because my current ortho-heavy X-Plane directory is sitting at 2.76TB. That would not even fit on a singe m.2 drive.
I haven’t gotten into any formal process for benchmarking or anything. Running X-Plane off the USB-C connection takes about 18 seconds to the X-Plane menu. And that is loading a bunch of plug-ins. From the main menu, if I load a complex aircraft into ortho scenery and custom payware scenery (for instance - Salt Lake City), takes 1 minute and 14 seconds to be in the sim and flying. I don’t know that would necessarily be any faster on the m.2 drives…but maybe.
I did some flying in both 2D and VR today in X-Plane and (I’m running 3440 x 1440) I found I can pretty much put all the sliders all the way to the right except I have World Objects to the middle and texture to Maximum (but not Maximum w/ no Texture Compression). That gives me solid FPS in both 2D and VR.
DCS World is on the m.2 drive and launches to the main menu in 46 seconds on the initial launch.
Once in DCS World - launching an F/A-18 mission (I chose Ode To Mongo) takes 46 seconds to get into the cockpit. Launching the mission a second time (like hitting Refly) I’m in the cockpit in 4 seconds. All of this is with maximum detail settings and 3440 x 1440.
Closing DCS and relaunching it a second time brings me to the main menu in 21 seconds.
Running the F/A-18C Mongo mission with the following settings I get 100FPS (2D)…
I’ll have to check and see what my VR FPS is, but I found I had to back off MSAA and bump shadows down to preserve very high FPS.
3 Likes
I also noticed that under load, even when the fan kicks on the MSI is very quiet. I do wish that it had an HDMI and/or DisplayPort ports on the front so I could plug my O+ in via that - it would make it a little easier if I wanted to plug it in, and would buy me another two feet of cable, but it would also be susceptible to being kicked, so maybe it is better that it is in the rear. I do like it has USB-C on both the front and back.
2 Likes
It’s hard to tell, for me, with respect to the M.2. I have not given DCS a go yet, bt X-Plane went from a HDD to an M.2 SSD … so it’s obviously better off. Load times where significantly improved! 
I will be in DCS on Saturday, at the earliest (travel to visit my parents on Friday). I suspect it will be faster than it was on my old school SATA SSD, but I am also overclocking, so there is that.
If I was to suggest anything … by an M.2 drive for OS and primary games, but maybe a SATA SSD for ‘less important games’. My OS still resides on my SATA SSD and I can’t be arsed to move it. I will be rebuilding a new PC in 6-10 months, so I will have to go through all of that then
3 Likes
I just added a 1TB Samsung 970 EVO Plus NVMe to my rig. Was on a SATA SSD before. I cloned windows and whatnot over without issue thanks to the Samsung app and am running everything from the EVO now.
Maybe a couple seconds faster for DCS startup. It’s close. But hoo boy, copying large files is lightyears faster. More much needed storage and less clutter in the case. Sorta future proofed for a while. Overall I’m thrilled with the upgrade.
4 Likes
So with a nudge from @fearlessfrog to “publish the numbers man!!” I sneaked away this early morning Mother’s Day to try to grab some stuff. No worries, I’m making crepes for the wife whenever she awakens…that has to buy me some good will right?
In order to keep things consistent and repeatable, I just used the DCS World pre-set graphics settings for MED and HIGH for these tests. I was also running fpsVR, a Steam application that is really cool that binds a performance table to your hand controller so you can reference it. What I don’t know is if settings in fpsVR trump the ones you’ve set in SteamVR. The reason I say this is that in SteamVR, I have super-sampling set to just 100% to keep things sort of baseline. But in these screens, I see that fpsVR has Global Resolution Setting at 200% giving a 2018 x 2516 render per eye. So maybe fpsVR forces those values above and beyond what you’ve selected in SteamVR (I’ll follow up with a test setting the fpsVR to 100% and see what we get).
At the preset MED settings in DCS World, and 1.8 pixel density, and perhaps 200% SS as indicated by the pop-up menu, I was getting around 45 FPS. The average FPS is showing 37.6, but it was weird because the left hand number wasn’t really dipping at all below 45, so I don’t know where that average comes up. Anyway - the sim looked fabulous, was perfectly playable and smooth at these settings. You’ll see that compared to the next image, the MED settings change some of the shadow effects, which is probably the greatest visual difference between MED and HIGH. That dark stripe you see down the center is not what you see through the O+, it is just a screen refresh thing with the iPhone I was using to take the picture.
So next I bumped the DCS preset to HIGH. You can see the shadows cause some of the cockpit to be darker. FPS on the left was showing around 30 and the right (average) was 37.3. I don’t know which to believe. It was pretty solid at 30 on the left though, and again, the sim was very playable at that FPS.
Next I went full bore and put all of the graphics menu settings to their maximum (or ultimate), full MSAA and anisotropic filtering, and all of the preload, trees, grass, clutter, etc. sliders fully to the right. Just maximum everything. Oddly enough, the sim looks much worse with these settings. The colors are washed out, the brightness is too high, and obviously performance drops - 19 in the left and a not believable 37.5 on the right. I halfway wonder if the right number (average) isn’t just using a single session…maybe there is a button to reset that average that I need to hit. Hovering around 20FPS, the game is not comfortable at that refresh. I think 30 and above is playable, 45 is fantastic.
EDIT - I see now there is a RESET STATS tab in FPSvr that I obviously should have been hitting. So I’d disregard the average readout and the left current FPS is actually pretty much accurate as an average.
I will repeat some of these tests with the fpsVR settings at 100% and maybe also see what dropping the pixel density settings does. And I’m aware that there are certain menu settings (I think MSAA) that have very large effects that, when turned off, result in substantial improvements to FPS without really dropping visual quality…so I’ll mess around with those and try to get some MED and HIGH numbers with maybe one or two of the killer menu items turned down (I’ll take any suggestions on what those are…maybe shadows to FLAT, blur OFF, and MSAA to none, aniso to 4x or 8x…something along those lines).
4 Likes
I see now there is a RESET STATS tab in FPSvr that I obviously should have been hitting. So I’d disregard the average readout and the left current FPS is actually pretty much accurate as an average.
Project Cars 2 - with all sliders maxed and 200% Steam Global Resolution Setting (2018 x 2516) was getting 90FPS steady (is that the max display I will see with the O+…meaning…can that only display the max refresh rate of the O+ screens?)
Smooth as silk obviously at that frame rate…and this sim is impossible to get a through the lens shot with all the juggling of phone, controller, headset…LOL…so apologies for the crap shot…
4 Likes
fpsVR used whatever SteamVR has set. Because you put the ‘Video’ manual override to ‘200%’ and all the applications left at 100% (they are cumulative) then every app will be at 200%, and that’s what fpsVR is showing.
The DCS pixel density and SteamVR combine, so you are running an enormous and unviewable resolution! Either set the PD to 1.8 and then your global SteamVR Video to ‘100%’ or put PD to 1.0 and use the SteamVR ‘200%’ but not both together. I personally use the latter, as I like all games oversampled, but it is essentially the same thing.
If you press the ‘Windows’ button on your VR controller you get a nice set of icons, with the ‘screenshot’ one putting an image in your ‘My Pictures / My Captures’ folder. Really handy for grabbing fpsVR values rather than in-lens screenshots.
MSAA in an 16k resolution(!!) is incredible that it worked that well. Put the PD to 1.0 and a 200% then you’re still running 4K per eye (HP Reverb like). Because of so many pixels and oversampling the O+ you shouldn’t really run MSAA above 2x, as MSAA as an algorithm is quadratic in scaling, which basically means it performs worse and worse the higher the resolution, so it’s diminishing returns. Do put AF at x16 though whatever, to help shimmers.
Interested to see DCS with the PD 1.0 and SteamVR 200% - it should look pretty much identical to what you’ve done, but being a little quicker. Shadows are still CPU bound, so that (other than MSAA left at x2) is really the only tweakable on high end stuff like this in DCS.
Thanks for doing this @BeachAV8R!
2 Likes
LOL…ah…nice to know I’m way oversampling and can get even better. I’ll make those adjustments and report back.
Great…I can’t wait to try it out. Hmmm…too bad it isn’t Father’s Day… 
1 Like
So interesting results. Lowered the in game PD to 1.0 and put fpsVR to 200% global. As you predicted @fearlessfrog, performance was much improved over all settings.
Here is the HIGH setting preset, a solid and fluid 45FPS (I did reset the stats prior to each test)…
With all sliders set to their maximum (or ultra) manually, and all sliders such as terrain preload and all that to the right, I was still getting 45FPS and fluid.
This was interesting. With the VR setting preset I was getting a solid 45 although average was showing 63.
And this was REALLY interesting. With all of the sliders set to the MAX or ULTRA, and shadows flat and SSAA off, the sim looked really good (not sure if FLAT turns cockpit shadows off though), lettering was crisp, and the performance was awesome at an average FPS of 65-ish.
What is harder to describe is the smoothness of the motion. Like rolling the airplane there isn’t the slightest judder or anything…it feels buttery smooth. I’m wondering if a heavy tree area will be different, so I’ll test some on the Caucasus map in a bit.
5 Likes
That looks amazing!
I think DCS is unusual about reprojection, in that it seems to ‘lock’ to 45 fps regardless if you have WMR reprojection on or off (or SteamVR reprojection or the ASW on/off the Oculus side of things). It’s like they do their own thing for frame rendering and unless you are looking at the sky don’t tend to allow framerates like 60 fps - you either get 45 fps or 90 fps, but it is clamping between the two. It does make things smoother, as no computer on Earth so far can run DCS at a solid 90. If you look around it will flip between the two, so impacting the average FPS counter.
It is unusual as normally when the reprojection toggles are made outside the game then you get the choice and the game respects the settings, but on DCS I don’t think so much. If I had to guess, it looks like they are repeating the previous frame and clamp to 45 if it can’t make 90, as their own form of reprojection. I’ve never seen it confirmed either way on the ED forums.
1 Like
I think there’s a target FPS setting somewhere in the config file to configure the vsync. That used to work for conventional rendering, dunno if it still does and how things are for VR but possibly worth a try.
1 Like
On 2D I did use to have something like a Saved Games / DCS.openbeta / Config / autoexec.cfg file with this in it:
disable_write_track = true
options.graphics.maxfps = 120;
…but I don’t think people have anything for maxfps by default. I renamed mine, as 120 was a distant dream in VR.
@BeachAV8R, you could check you don’t have an old autoexec.cfg file in your DCS.openbeta / Config folder I guess.
2 Likes
Curious, i don’t even have an autoexec.cfg in my config folder any more. Must have deleted my DCS saved games folder at some point. Can you confirm it is still used?
It was always something the user could create as a new file after 2.0, and I don’t think it made it as a new file in the open beta sort of timeframe. The flags in it get set after the main install directory Config ones do, so it was good for overrides outside the main install location. The flags like the disable_write_track still work in it, although I don’t think it impacts VR for maxfps, but not 100% sure.
1 Like
I really look forward to DCS (and X-Plane) using more cores. That i9 9900K has 8 cores, and I bet just core0 is pegged at 100% giving you this overall ‘14% usage’:
The move to Vulkan isn’t so much about making the existing code more multi-threaded, but about getting away from DX11 only allowing a single core to issue instructions. Even Vulcan batching calls off the main thread will have a big impact, hopefully without changing too much of legacy DCS.
3 Likes
No autoexec.cfg file in that folder…just
authdata.bin
imgui
network.vault
options.lua
And I agree…I seem clamped at 45. I’m excited to see about this VR fix they have come across that is promising even more performance. I can play all day at 45 though…I do need to try a tree heavy area though.
1 Like
Should I give it a whirl?