Vive to Rift

[bobzdar]

Good question, and I should clarify that while there is a big window in the computer room, the shades are down and my tracking hiccups happen even at night. It's squarely related to very fast movement speeds.

Thus, I don't think it's a sunlight IR flooding issue, nor is it coming from the one lamp in the room that I usually keep on to have some ambient light besides my monitors (which is shielded such that none of the sensors can look directly at the bulb anyway).


Care to give me the $80-130 I'd need for one of those StarTech or Sonnet quad-channel cards, then? That'd clarify for sure if it's a problem with the add-in card.

Except I'm pretty sure that's also not the problem when it also happens with the motherboard's built-in Intel USB 3.0 controller. Something wrong's going on here, and I don't know how you're pulling off flawless setups, because mine isn't. And, no, downgrading to a triple-sensor setup is not a solution because of the obvious occlusion problems in the sensor-less corner that are far more game-breaking than occasional disruptions in very high-speed tracking.

Again, I must state that as it currently stands, tracking quality since the 1.12 update way back is 95% good, but that remaining 5% is the troublesome part. It's not something I can easily diagnose without money to blow on more controller cards or a whole new computer, either.


Drawing smooth, swooping curves with my setup is pretty easy. I just can't really do so cleanly by swinging my arms like I'm slicing and dicing some poor sap, gotta slow down a bit. If anything, it feels like the opposite problem people have with drawing tablets, where you often get better results by making fast strokes with the pen instead of slow ones, where jitter kills the line quality on N-trig digitizers especially (Surface Pro 3 and later).

I just can't figure out what's broken. I mean, my system isn't exactly budget hardware when it's built around a MAXIMUS VI HERO board, and it runs most things quite smoothly. Also, my card isn't a 5-port Inateck, but a 4-port QICENT with the same FL1100EX hub controller... wait a minute, Amazon, what the hell did you do to the product page linked in my order such that it no longer features my card? This isn't helping things one bit, but yeah, the card works normally, and the Fresco Logic tweaks come right up in Oculus Tray Tool.

Oh well, it's not like any of this ever stopped me from playing the hell out of Robo Recall, GORN or SUPERHOT VR in the 8'x8' space I have. (And that really isn't enough for my liking, seriously... can't even take one step away from center without putting a boundary in arm's reach.)

As testing, have you tried unplugging 2 sensors and seeing if the problem still exists? That can narrow it down to a usb subsystem issue or something else. I mean, it sounds like it's mostly fine and just has issues with fast movement, so not sure how much time you want to spend on it, but if the problem goes away with 2 sensors then it becomes a little easier to pinpoint the problem.

 

[BecauseScience]

Except I'm pretty sure that's also not the problem when it also happens with the motherboard's built-in Intel USB 3.0 controller.

FYI, none of the onboard USB3 ports on my months-old Rysen motherboard work with Rift at all. I can't even get past the sensor setup using them.

You said you have sensors and all are connected to USB3, correct? When I did my install, I read over and over that only two sensors should be plugged into USB3. Those should be the primary sensors with the best view of your playspace. Supplemental trackers should be in USB2 ports where they operate in some kind of bandwidth restricted mode.

I wonder if having all four of your sensors in USB3 is swamping something with too much high bandwidth data.

Did you ever say what CPU you're running?

 

[Mut1ny]

Just wanted to add my 2 cents here.

I have my Oculus on the following setup:

I have a Asus Z170i Pro Gaming ITX motherboard.

I have my setup just on one USB 2.0 port and one USB 3.0 port using two USB 2.0 repeater cables?

On the USB 3.0 port I have it going to one 16ft USB 2.0 repeater cable to a USB 3.0 4-port hub (non-powered) for the left and rear sensors.

Then on the USB 3.0 port I have a 16ft USB 2.0 repeater going to a 4-port USB 3.0 hub (powered) for the Oculus USB, right front sensor, adapter for wireless KB/M, and the cable for my racing wheel/pedals.

Have the Oculus HDMI hooked up through a "4K" passive repeater (like this one) on a 15ft HDMI cable straight out of my Zotac GTX 1080 FE. 15ft for sure, just the repeater, no external power or anything else.

So basically I'm using USB 2.0 for all USB connections because 3.0 port is on a 2.0 repeater cable. Yes I'm sure, they say so and aren't blue, etc...for sure USB 2.0.

I have no issues with latency or lag or anything like that. Picture works fine, controllers work fine, sensors, the whole shabang!

I'm pretty sure this breaks all rules of Oculus logic, no?

 

[NamelessPFG]

It's been a while since I bothered to dive back into this, but I figure I should report back.

Where did I say anything about reducing the number of sensors?

If you have a four sensor setup and using them all in USB 3 then maybe you should try using the USB 2 extensions cables that came with the extra sensors. Plug 2 front sensors into the USB 3 card, plug the headset into the USB 3 on motherboard and then plug the two rears (using the USB 2 extension cables) into any spare USB ports on your motherboard.

And because motherboard USB ports are so variable, you might need to try a few different combinations. As an example in my last computer, all my USB 3 ports were compatible with the Rift, but, when I plugged the headset into one port it would not work, and if I plugged a sensor into another port I got tracking errors.

You say you can't troubleshoot, but there are lots of things you can do to try and source the problem.

USB 3.0 extension cables. If you are using these, one of them could be having issues. Is there anyway to plug the sensors directly into the computer without using the extension cables? Just for testing purposes.

You could plug in two sensors see how your tracking works, then try three sensors, then four. See when the tracking problems occur. Then try different combinations of sensors. See do you get the tracking errors when you plug in a specific sensor or does it only happen when you plug in the fourth sensor regardless of what order you plug them in.

You can either troubleshoot your problem and fix it, or live with it.

Okay, let's go through all this once more. What I'm pointing out happens even before Oculus says anything about errors, and in fact, I haven't seen any "Wireless sync timed out" errors at all in recent memory. Software improvements in the meantime? Perhaps.

Meanwhile, only now does one of my sensors start throwing out "corrupted frame" errors. Argh. Unplugged it and went to a three-sensor setup; saw subtle tracking issues in the now sensor-less corner when using one hand in front of the other, as expected. Thankfully, it's a rear sensor. (I connected the sensor throwing fits to one of the motherboard's own USB 3.0 ports, and thus the Intel root hub, for the record.)

Moved the other rear sensor to a USB 2.0 port; no improvement. I can still get zig-zags by really chopping down in Quill, though moving my arm slightly slower straightens it out.

Added the other rear sensor back in; improvement in that one corner, no real change at high speeds.

Removed both rear sensors, still no real change at high speeds, zig-zagging is there regardless of whether I use the Intel or Fresco Logic USB 3.0 root hub controller. There's no errors mentioned.

Due to the placement of the front sensors, I can't take their respective USB 3.0 extension cables out of the equation entirely. There's not enough slack to get to my computer when they're up high in the corners. Even then, I have my doubts at this rate if I get zig-zags with just two sensors.

As testing, have you tried unplugging 2 sensors and seeing if the problem still exists? That can narrow it down to a usb subsystem issue or something else. I mean, it sounds like it's mostly fine and just has issues with fast movement, so not sure how much time you want to spend on it, but if the problem goes away with 2 sensors then it becomes a little easier to pinpoint the problem.

See above; it still happens with just 2 sensors. Errors in the Devices section might go away, but I still manage to run into the tracking system's limits somehow.

Again, it's not really a problem for most games and software, but it still feels like a limit that future tracking systems could negate. Sometimes I wonder if what those sensors really need is a refresh rate faster than 60 Hz; TrackIR runs at around 120 Hz IIRC. Just about everyone here knows what a big difference high refresh monitors make, so I shouldn't have to explain that in too much detail.

FYI, none of the onboard USB3 ports on my months-old Rysen motherboard work with Rift at all. I can't even get past the sensor setup using them.

You said you have sensors and all are connected to USB3, correct? When I did my install, I read over and over that only two sensors should be plugged into USB3. Those should be the primary sensors with the best view of your playspace. Supplemental trackers should be in USB2 ports where they operate in some kind of bandwidth restricted mode.

I wonder if having all four of your sensors in USB3 is swamping something with too much high bandwidth data.

Did you ever say what CPU you're running?

The CPU's an i7-4770K overclocked to 4.6 GHz, well above the requirements. It's also been a very stable overclock in my experience.

If you're curious about the GPU it's paired with, it's a GTX 980. Nothing special in this age of Pascal, but I told myself I wasn't spending another cent on a new GPU until NVIDIA brings Volta to market (and the 980 Ti/1080 Ti equivalent thereof, no less).

I would think that, but while adding the second USB 3.0 controller card and splitting up the sensors cut down on the "dropped frames" errors significantly (which would happen if connecting three or four sensors directly to the Intel USB 3.0 controller), it hasn't entirely resolved the issue.

Also note that in my latest findings, I'm generally NOT seeing "Wireless sync timed out" errors any more, but the tracking behavior stays largely the same even when it says everything is fine. I'm starting to think that whatever it is, it's not as much of a USB problem as everyone likes to think.

Why do I say that? Random stutters that only seem to happen in VR titles, too infrequently and inconsistently for me to pin down. Every time I see a framedrop while drawing a line in Quill, the line zigs and zags a bit - and note that Quill almost never has performance issues. Anything Unity-based like Onward, Bullets And More, COMPOUND and GORN is a complete crapshoot in terms of either running almost perfectly smoothly on my system, or seemingly framelocking itself to 30 FPS or less without Asynchronous Timewarp doing its job, which is about as pleasant as you'd expect. It's a weird bug I just can't pin down despite my best efforts, other than that launching outside of Steam (if possible) seems to end up with much better results for the Unity engine titles.

 

[BecauseScience]

Windows MR's Steam VR support is scheduled to drop on the 15th. Why not sell the Rift and pick up a Samsung Odyssey? No more tracker hassles.

 

[bobzdar]

It's been a while since I bothered to dive back into this, but I figure I should report back.


Okay, let's go through all this once more. What I'm pointing out happens even before Oculus says anything about errors, and in fact, I haven't seen any "Wireless sync timed out" errors at all in recent memory. Software improvements in the meantime? Perhaps.

Meanwhile, only now does one of my sensors start throwing out "corrupted frame" errors. Argh. Unplugged it and went to a three-sensor setup; saw subtle tracking issues in the now sensor-less corner when using one hand in front of the other, as expected. Thankfully, it's a rear sensor. (I connected the sensor throwing fits to one of the motherboard's own USB 3.0 ports, and thus the Intel root hub, for the record.)

Moved the other rear sensor to a USB 2.0 port; no improvement. I can still get zig-zags by really chopping down in Quill, though moving my arm slightly slower straightens it out.

Added the other rear sensor back in; improvement in that one corner, no real change at high speeds.

Removed both rear sensors, still no real change at high speeds, zig-zagging is there regardless of whether I use the Intel or Fresco Logic USB 3.0 root hub controller. There's no errors mentioned.

Due to the placement of the front sensors, I can't take their respective USB 3.0 extension cables out of the equation entirely. There's not enough slack to get to my computer when they're up high in the corners. Even then, I have my doubts at this rate if I get zig-zags with just two sensors.


See above; it still happens with just 2 sensors. Errors in the Devices section might go away, but I still manage to run into the tracking system's limits somehow.

Again, it's not really a problem for most games and software, but it still feels like a limit that future tracking systems could negate. Sometimes I wonder if what those sensors really need is a refresh rate faster than 60 Hz; TrackIR runs at around 120 Hz IIRC. Just about everyone here knows what a big difference high refresh monitors make, so I shouldn't have to explain that in too much detail.


The CPU's an i7-4770K overclocked to 4.6 GHz, well above the requirements. It's also been a very stable overclock in my experience.

If you're curious about the GPU it's paired with, it's a GTX 980. Nothing special in this age of Pascal, but I told myself I wasn't spending another cent on a new GPU until NVIDIA brings Volta to market (and the 980 Ti/1080 Ti equivalent thereof, no less).

I would think that, but while adding the second USB 3.0 controller card and splitting up the sensors cut down on the "dropped frames" errors significantly (which would happen if connecting three or four sensors directly to the Intel USB 3.0 controller), it hasn't entirely resolved the issue.

Also note that in my latest findings, I'm generally NOT seeing "Wireless sync timed out" errors any more, but the tracking behavior stays largely the same even when it says everything is fine. I'm starting to think that whatever it is, it's not as much of a USB problem as everyone likes to think.

Why do I say that? Random stutters that only seem to happen in VR titles, too infrequently and inconsistently for me to pin down. Every time I see a framedrop while drawing a line in Quill, the line zigs and zags a bit - and note that Quill almost never has performance issues. Anything Unity-based like Onward, Bullets And More, COMPOUND and GORN is a complete crapshoot in terms of either running almost perfectly smoothly on my system, or seemingly framelocking itself to 30 FPS or less without Asynchronous Timewarp doing its job, which is about as pleasant as you'd expect. It's a weird bug I just can't pin down despite my best efforts, other than that launching outside of Steam (if possible) seems to end up with much better results for the Unity engine titles.

That's really odd behavior. My thought is either some type of Oculus defective hardware issue (maybe another peripheral that's going rogue throwing the Oculus off?) or an issue with the motherboard itself that can't handle the usb load properly. The clue is that you also have issues running things at a solid framerate in VR, which given your hardware should not be an issue - my system is roughly the same performance wise and I have no software issues in Steam or otherwise using the Oculus. It's like something is causing a pci-e bus interrupt that's hurting the video card as well.

If you can, run a front facing 2 sensor configuration temporarily using the 2 cameras without extension cables and see if the problems go away - this would at least rule out the usb extension cables as causing usb latency issues that cascade into pci-e stalls or something crazy like that.

I think you can also send the log files to Oculus and they may be able to clue you in on why you're having stutters and issues as that's not normal behavior in a similar powered system at all.

 

[reaper12]

See above; it still happens with just 2 sensors. Errors in the Devices section might go away, but I still manage to run into the tracking system's limits somehow.

Again, it's not really a problem for most games and software, but it still feels like a limit that future tracking systems could negate. Sometimes I wonder if what those sensors really need is a refresh rate faster than 60 Hz; TrackIR runs at around 120 Hz IIRC. Just about everyone here knows what a big difference high refresh monitors make, so I shouldn't have to explain that in too much detail

You keep using your experiences of the tracking and applying to everyone. There isn't a high speed tracking problem. It's only you who are having these issues. I can swing as fast as I want. I can do sudden sharp fast movements.

You know that both the Vive wands and the Rift touch use IMUs for tracking, and the cameras/lighthouses for drift correction? The IMUs sample at 1000hz and report at 500hz. So I highly doubt that there is anybody alive that can move faster than either system can track. You are under the mistaken impression that the sensors do the majority of the tracking, they don't. The heavy lifting is done by the IMUs in both systems. Your monitor statement helped explain to me that you don't fully understand how tracking works, hope my post has helped. ​

The CPU's an i7-4770K overclocked to 4.6 GHz, well above the requirements. It's also been a very stable overclock in my experience.

If you're curious about the GPU it's paired with, it's a GTX 980. Nothing special in this age of Pascal, but I told myself I wasn't spending another cent on a new GPU until NVIDIA brings Volta to market (and the 980 Ti/1080 Ti equivalent thereof, no less).

I would think that, but while adding the second USB 3.0 controller card and splitting up the sensors cut down on the "dropped frames" errors significantly (which would happen if connecting three or four sensors directly to the Intel USB 3.0 controller), it hasn't entirely resolved the issue.

Also note that in my latest findings, I'm generally NOT seeing "Wireless sync timed out" errors any more, but the tracking behavior stays largely the same even when it says everything is fine. I'm starting to think that whatever it is, it's not as much of a USB problem as everyone likes to think.

Why do I say that? Random stutters that only seem to happen in VR titles, too infrequently and inconsistently for me to pin down. Every time I see a framedrop while drawing a line in Quill, the line zigs and zags a bit - and note that Quill almost never has performance issues. Anything Unity-based like Onward, Bullets And More, COMPOUND and GORN is a complete crapshoot in terms of either running almost perfectly smoothly on my system, or seemingly framelocking itself to 30 FPS or less without Asynchronous Timewarp doing its job, which is about as pleasant as you'd expect. It's a weird bug I just can't pin down despite my best efforts, other than that launching outside of Steam (if possible) seems to end up with much better results for the Unity engine titles.

Not sure what the issue is then. There is something wrong. Are you using supersampling at all? Maybe you should try turning that off completely. In fact you should lower all graphic settings to the minimum and see if it's actually a performance issue.

Do you have oculus tray tool installed? If not, try it and try using the registry tweaks for the fresno logic controller.

Random stutters and frame drops would led me to believe that you are either running at too high settings or you have a hardware problem.

Sorry that you continue to have issues, hope you get it sorted.

 

[reaper12]

An interesting post on Reddit about the tracking speeds. The first Eleven Table Tennis tournament is over and two Rift owners were in the final.


It looks like the Rift tracks fast movements just fine.

 

[NamelessPFG]

You keep using your experiences of the tracking and applying to everyone. There isn't a high speed tracking problem. It's only you who are having these issues. I can swing as fast as I want. I can do sudden sharp fast movements.

You know that both the Vive wands and the Rift touch use IMUs for tracking, and the cameras/lighthouses for drift correction? The IMUs sample at 1000hz and report at 500hz. So I highly doubt that there is anybody alive that can move faster than either system can track. You are under the mistaken impression that the sensors do the majority of the tracking, they don't. The heavy lifting is done by the IMUs in both systems. Your monitor statement helped explain to me that you don't fully understand how tracking works, hope my post has helped. ​

Not sure what the issue is then. There is something wrong. Are you using supersampling at all? Maybe you should try turning that off completely. In fact you should lower all graphic settings to the minimum and see if it's actually a performance issue.

Do you have oculus tray tool installed? If not, try it and try using the registry tweaks for the fresno logic controller.

Random stutters and frame drops would led me to believe that you are either running at too high settings or you have a hardware problem.

Sorry that you continue to have issues, hope you get it sorted.

I don't mean to say that everybody has tracking issues at very high speeds, especially given that I've only sampled one VR setup out of however many are out there. Nobody else I know locally is willing to plunk down the cash for VR; my own Rift setup is usually their first impression. They're usually, well, impressed. It doesn't spaz out like the dreaded pre-1.12 updates.

And yes, I know that the IMUs provide the bulk of the tracking, with the external IR reference points in either setup to correct for drift (which happens a lot with unassisted IMUs of any sort). I'm just wondering if the drift correction is happening frequently enough, or if errors in the drift correction for whatever reason (say, moving the Touch controllers at really fast speeds causes the IMU to buffer overflow somewhere and reverse for a split-second, causing the zig-zag effect when it errors and self-corrects mid-swing) are responsible for what I'm seeing. It's clearly not normal, even though anything short of a really hard swing gets tracked just fine.

Oculus Tray Tool is installed, of course. Fresco Logic registry tweaks are set; it's the first thing I had in mind when I ordered that card.

However, that means I've only got one setup to test and troubleshoot. It's not something I can easily troubleshoot, either, but I think I'll have a good opportunity to rule out performance-related stuff soon.

You see, a friend of mine wants me to build a new computer soon-ish, since the last one I built for him was an old i5-2600K/GTX 460 box from years ago. With a budget of around $1,100 tops (shooting for $800 to $1,000), that should be able to net me some pretty good hardware, at least Kaby Lake or Coffee Lake i5 and a GTX 1060 6 GB, shooting for better if holiday sales permit it. (1070 would be great, but likely out of budget.) I could give the system a VR test drive before I send it out, should he actually go through with this.

Other than that, though? I don't use supersampling at a global level (Oculus Tray Tool has it set to 0), preferring to adjust that in-game since most VR games have the option as part of their graphics settings. What's weird is that they generally perform just fine and dandy, except when they hiccup for just a sec. Maybe I better run some DPC latency monitoring utilities and see what that shows me.

Also, it's only VR games where it tends to happen; I've been playing the hell out of Destiny 2 lately, and that game's just smooth as butter. Can't even remember it dipping below 60 FPS, and it's usually hitting the 120 FPS cap I set (FG2421 can't refresh faster than that).

Even in VR, I just dove into a few minutes of Aircar when I found it in the Oculus Store again, and since it runs on Unreal Engine 4, it doesn't seem to have the "inexplicable low FPS cap" problem Unity games randomly get. Heck, even Raw Data seems to perform more consistently than said Unity games, and Raw Data is so demanding that I have to keep everything around low or medium for framerate's sake. Any Unity game in VR is a total crapshoot as to whether it runs flawlessly or caps its FPS into judder hell the moment I launch it, because I can tell you that GORN, COMPOUND, and so on will run just fine when they aren't capping themselves. Usually, this means launching them outside of SteamVR for whatever reason, as the alpha builds executed directly from File Explorer never did it, but launching them through SteamVR almost always did.

Another potential factor is that I don't have the drive space to put my games on proper SSDs, so I have a 4 TB total RAID 0 array (which only has about 276 GB left, so I better find a 6 TB or larger drive that isn't hampered with SMR soon). Anything that thrashes said drive array (which my VR games are obviously stored on) is going to have load times go through the roof, and it feels like Intel chipset software RAID hammers the CPU a lot more than necessary. Never mind that the OS is installed on an 850 EVO; HDDs in general seem to slow down computers greatly at this point.