Supmua

Just upgraded to 5800X3D from 9900K, on average my CPU frametimes drop by 3ms, and this is with somewhat crappy RAM with the Ryzen (3000MHz CL16 vs 3600MHz CL15 in the 9900K). I can probably get better results with better RAM. This pretty much cancels out whatever I lost from the 2.8 beta update (~2ms frametime) plus a bit more. EDIT: really like the DOCP function in ASUS Bios, making overclocking my old RAM a breeze.

I'm taking apart my PC, upgrading mobo and CPU from i9900K to Ryzen 5800x3D. If all goes well will grab some more frame numbers. But I think overall this update is somewhat busted (at least in the shadows setting) and shouldn't be relied on for true performance. EDIT: It's up and running and learned quite a bit from cross system upgrade--but unfortunately my RAM isn't compatible. My 4 yo old RAM sticks however work but I'm gonna have to put some money on new high speed RAM to avoid gimping the system, so won't be benching until the new RAM sticks come in. Also had a chance to reexamine the 16-pin GPU cable, no signs of melting, burn or crack...phew.

Just did Apache Hot Start Mariana. The fps was actually better around the airport which puzzled me. So I flew toward the Carrier that is slightly N of the runway (this is the usual culprit for fps drop), and it was no longer there but you can still see parked planes floating in the air in the spot where the CV normally was. At this spot normally my CPU FT would be around 30ms but it was like 22-23ms which is a lot nicer. So I opened up the mission in the editor and it appeared that they changed the CV object to a regular boat (still named CVN70, btw)--very sneaky, lol, might as well delete all those floating planes too huh. Anyway, I don't have anything else to complain on this mission. Averaging 56 fps, which is decent. Of course this was with reduced pixel rendering to 3022 horizontal per screen and visibility to medium. At this res it's not crisp like typical image quality that I'm used to but still plenty sharp and tiny texts are all legible.

Updated and did a quick test on a bunch of maps. On average both my CPU and GPU frametimes increase by 2ms, this is with VR preset so if you use higher settings the results may be worse. The lighting looks different so I'm going to have to redo my Reshade settings also. My testing also assumed that there's no different between the current NVIDIA driver and the previous one (my old data was obtained using the previous NVIDIA driver). EDIT: performance differences appear vary among missions also. I typically fly with VR preset plus High texture, Low shadows, High visibility, MSSAx2, ANS filtering 4x. With these settings the performance degradation isn't drastic. No longer am I getting 90 fps in free flights but it is still in 80s in those low requirement areas. Flat terrain shadows now look more like normal shadows with soft edges (no shadow fluttering which is nice). Ground objects appears more sharp grainy and shimmer like it underwent a sharpening filter, especially Persian Gulf, had to reduce my Reshade sharpening setting to accommodate. Still need to stress test with high requirement aircraft (Apache) and taxing maps such as Mariana but need to catch up with other games lol. After fudging around with the settings, I'm back up to 90 fps in some free flights especially the Harrier. Had to reduce visibility to medium and lower the horizontal pixels to 3022 per screen. If you play at high preset though, the performance taxing would be worse. This might explain why people who render at lower resolution and settings didn't see much performance impact. Typical Varjo setting at Very High on the headset is around 3600 pixels per screen without changing things in SteamVR, and that's a significant load.

This is the first performance gain post I've seen.

It looks like terrain shadows might be busted according to the general bugs thread.

I have a bunch of recent CPU and GPU FT numbers that I recorded from various aircrafts and starting missions (for 3090 vs 4090 comparison, also planned 9900K vs 5800X3D) but these aren’t meant to be compared with other people’s setup. So these would come in handy as they were obtained from the previous beta version.

Remember also that this is still beta branch.

That’s better than CPU FT getting hit in my book, there’re a lot you can do to compensate. I’ll mess around tonight.

Need to check whether CPU or GPU frametime is affected, or both. I guess it’s the new weather system with fancy clouds and rainbow, etc.

Ouch, there goes my 90 fps dreams. Does this happen with other people too? I thought we were getting VR optimization with this patch.

According to Igor's lab, the problem starts when you bend the cable sideways as the soldering on the edge of the connector appears flimsy and the wire connection can break off at that spot. It's a good thing that I'm already using ATX 3.0 PSU.

It seems more cable melting issues have been reported, so it might be a good idea to avoid using the NVIDIA’s flimsy adapter that comes with your 4090 if you can. So far there’s been no reported issue with proprietary cables. Either get those directly from manufacturers or buy an ATX 3.0 PSU that comes with the cable.

Yeah image quality perception can vary among individuals. The Varjo Aero for example, can exhibit significant color fringe unless IPD setting is perfect and the headset itself is in an optimal position.

The reported peripheral distortion is a bit of a letdown, but need to see more reports.

Nice! This is exactly what I wanted to see. It looks like having a large L3 cache does help quite a bit, up to 2-3ms CPU frametime reduction. This is a good test because not only CPU but also GPU are being equally pushed. How about 5800X3D vs an Intel CPU if you have access to one? The upcoming Ryzen 3D should be even better for sims. Still, this is nothing compared to GPU performance jump from 3090 to 4090, I'm seeing up to 9ms frametime improvement in GPU which is 3x of this CPU performance uplift.

nullMy ASUS TUF boosted pretty well, up to 3015 MHz, prob not going to push more just to be on the safe side. Temp is also nice, Tmax at 63 degree.

Clean install of what Windows or DCS? Try my settings if you want 90 fps on those free flights, moving your head inside the cockpit is a stutter free experience at this fps. I expect to see even better performance after upgrading mobo and CPU this weekend. Varjo: Very High, Vsync OFF SteamVR: SS 76% (should read 3144x2696 pixels per screen) DCS: VR preset but increase ground textures to High, shadows Low, MSSAx2

Create a post there and you should get in. There’ll be NDA document to sign.

This is usually due to disconnection or loose connection of the USB C cable, more often the side that goes into the headset itself. Unplug it (pull diagonally from the headset’s left upper compartment and reinsert, sometimes flipping the cable 180 degree can also make it tighter).

I can't say much but, super happy about the Aero now. I would encourage anybody with 4090 to contact Varjo's Arch via discord to be included in the beta as they said they would include everybody who has 4090. Stuttering is fixed, but also the IPD issue so now I can use a very thin facepad to maximize FOV especially the vertical FOV.

Of all the headsets I’ve had, the Aero by far. It’s the only headset that’s good enough to be used in place of a regular monitor because of high resolution and edge to edge clarity. However it’s got its own issues such as chromatic aberration (color fringes), narrow vertical FOV, narrow face gasket, missing features such as motion smoothing, mini LED local area dimming, and built in on/off switch. My ideal VR headset though would be something akin to the Vive Pro 2 with OLED panels and aspheric lens or anything but fresnel lens.

The bottleneck is always going to be there as long as GPU and CPU don't have equal power, as the less powerful one will always bottleneck the other. But it is such a relative term, for example, bottlenecking at 30 fps is no good but bottlenecking at 90fps is fine with me.

It doesn't hence the tiny size. Dynamic foveated rendering tech is sort of meh right now, I have it on my Vajro Aero and the performance increase is subtle because it also creates extra workload (from eye tracking calculations, etc.) resulting in minimal performance gain (Facebook/Meta's research has also come to this conclusion). Fixed foveated on the other hand can save more GPU frametime, but imaging quality outside the center area is not great when you glance around.

It’s got all the right specs for PCVR except maybe FOV, and it’s backed by an electronics giant Panasonic. It’s also made in Japan, so the build quality should be quite good.