Looking Glass - Triage

I haven’t had the reset bug with the RX590.

This thread is going off topic, please take this discussion elsewhere.

1 Like

Hey, @gnif, I remember you’d mentioned something about an OBS plugin that works with Looking Glass. Did you ever do anything to that end?

It’s still planned, just not done yet :slight_smile:

Okay.

I’d be more than willing to pay for a license, if that ahem motivates. :stuck_out_tongue:

1 Like

The plugin will be open source, so no licenses, etc. Things have slowed a little here as I have to go international to a friends wedding (sigh… I hate travel), so it will likely be a few weeks before I can get onto new code.

2 Likes

Ah, I’m with you.

You’re not coming to Cali by any chance, are you? If you are, I’ll buy you a beer.

No worries, I know your time is limited and valuable.

Haha, thanks for the offer but I doubt you’re in Vanuatu :stuck_out_tongue: .

Edit: btw, glad to see you back here and active again, it’s been a while :slight_smile:

Nope, I’m not lol

Yeah, I’ve been out of the scene for a bit. I was head-down on work stuff for a while.

I saw you released Beta 1, congrats! (I know it’s late)

1 Like

If anyone is interested there is a test branch of LG (fbtest) that incorporates a chunked memory copy between guest and host. This was an idea I had today during my shower (where most good ideas come from) that allows the host and client memory copies to run asyncronously, which should decrease latency.

Note, you will need to build your own host application to test, the host has changed to support this feature

2 Likes

I’m looking for a dummy-plug recommendation for my setup. I’ve got a bit of a unique requirement since my primary display is 3840x1600 native.

I’ve got a couple dummy plugs that work fine at 1080p, but they drop down to 17hz when I set a custom resolution. :frowning:

Any advice on that front?

Okay got a few questions, but before that. I pretty much have the setup up and running per the POP_OS guide (and lots of other research on code 43, compiling the client, apparmor for the shared memory file etc…) I have a few issues and wondering if there is an easy fix, its working that way, or I need a hardware upgrade.

If I leave my monitor plugged in, however, I swap inputs the client screen freezes and receives no more updates. I can get normal usage when I toss in a dummy plug, but I swear it worked a time or two before if I just swapped inputs. Is this intentional?

I will assume here that the GT710 is not enough to run POP_OS on/looking-glass client, while the 1070 is passed to Win10. If I see the games/benches report 60-100FPS, but it feels like a slide show on the client side. I am really guess the 710 cannot perform the job and properly copy/display the guest. I can get my hands on some cheapish 1060s 6GB, or is there another alternative?
Edit: I realized that POP_OS at 4k, already stressed the GT710 to 80%. I am now more convinced that the GPU just cannot handle looking glass.

Outside of synergy is there any other way to jump in and out of the host, kinda like ctrl+alt on libvirt?

This is expected behavior. You need a dummy plug because DXGI needs a “display” to render to.

I’d say that’s the most likely culprit. Can you give full system specs?

Are you going to be doing anything more intensive than Looking Glass and video acceleration on the guest? If not, I’d look at the gtx 1050 or maybe an RX 550.

1 Like

@gnif, is there an option on the host to choose which IVSHMEM device looking glass tries to use? I’ve got two on my system and it’s grabbing the wrong one by default.

EDIT: to be clear, I’ve got a looking-glass-host.ini file in the same directory as the exe, with the following contents:

[os]
shmDevice=1

Regardless of what I do, it doesn’t grab the other device. I’m assuming I’ve configured this improperly.

1 Like

Looks right to me, have you tried doing it on the command line?

looking-glass-host.exe os:shmDevice=1

1 Like

I haven’t. I’ll give that a try when I get back to the office.

Unsure how the quote/reply function works here, so I’ll answer in order.

Thanks, ill keep using the dummy plug.

Full system specs are overkill, 3900X on Asus X570 TUF, 32 GB Ram, GT710/GTX 1070. I had the 710 in place to facilitate GPU Passthrough. Then I saw this setup and wanted to give it a try, quickly I realized that my setup has one restrictive part.

As for intensive… Not really, the reason I am looking at 1060s and not much else since emy co-worker has a couple on hand that he isn’t using (and haven’t had luck selling). If I go to the local stores around me the 5700 is the lowest AMD card for 35,000 yen, and the 1650 go for almost 20,000 yen. I am getting a 1060 6Gb for $120 or so. The only other option is to drain the loop, slap an accelero IV on the 1080 ti and use the 1070 for POP_OS.

I would love to leave the Windows eco system, however, I have yet to get my Oculus setup to work/many other games that utilize DRM/anti cheat tech… I feel that we are getting closer, but we are also still far away.

This should help: New and Existing Forum Features 2017

That’s a solid system for this setup.

Ah, that’s definitely fair.

Yeah, there are very few games that have issues under passthrough right now, but there are still a few who don’t work well.

Hey, that worked!

Thanks!

1 Like

Swapping to the 1060 worked, I went from 80% usage to 1% on idle, and 13% usage on the client. Just need to look at some of these frame issues, might just be Heaven Benchmark. I will look into VM optimizations and client flags, see if any of those make the experience smoother (its already night and day difference). Also, looking into software KVM, seeing if there is a better way to interact than using Synergy.