Does cloning your display cause increases in Input Lag?or FPS drops?

[nursejoy]

Hello I am seeing mixed opinions about whether or not cloning your display for dual PC streaming is the way to go. I want to do it as I want to retain the Gsync capabilities of my monitor as well.

I hear from many that cloning a display does not cause any issues, FPS drops, or input lag increases. But From others I hear the opposite.

I am using 1080p currently at 360hz.

[daviddave1]

Hello I am seeing mixed opinions about whether or not cloning your display for dual PC streaming is the way to go. I want to do it as I want to retain the Gsync capabilities of my monitor as well.

I hear from many that cloning a display does not cause any issues, FPS drops, or input lag increases. But From others I hear the opposite.

I am using 1080p currently at 360hz.

Fr33thy tested the inputlag difference between cloning and running OBS on gaming and streaming PC and then projecting at 4k. He also mentions 1080p in the video if I am not mistaken. Check out https://youtu.be/TCp1AYI7BlM?si=sbazZCpR5t6j8gZQ

If I remember correctly hdmi 2.1 or dp capture cards with passtrough is best for 360 hz 1080p so u can use passtrough. But dp cards are not here yet. There is a 2.1 hdmi aver media card and a Asus tuf one coming soon.
But there are no 360 or 540 hz 1080P hdmi 2.1. monitors and even if these monitors would be here 240hz is the max passtrough at the moment on the 2.1 hdmi aver media and asus cards.

[vlad54rus]

The last time i tested it (Windows 10 21H2) - using display cloning prevented borderless fullscreen games from getting tearing (even though they still engage Independent Flip), increasing input lag a little.

[nursejoy]

Hello I am seeing mixed opinions about whether or not cloning your display for dual PC streaming is the way to go. I want to do it as I want to retain the Gsync capabilities of my monitor as well.

I hear from many that cloning a display does not cause any issues, FPS drops, or input lag increases. But From others I hear the opposite.

I am using 1080p currently at 360hz.

Fr33thy tested the inputlag difference between cloning and running OBS on gaming and streaming PC and then projecting at 4k. He also mentions 1080p in the video if I am not mistaken. Check out https://youtu.be/TCp1AYI7BlM?si=sbazZCpR5t6j8gZQ

If I remember correctly hdmi 2.1 or dp capture cards with passtrough is best for 360 hz 1080p so u can use passtrough. But dp cards are not here yet. There is a 2.1 hdmi aver media card and a Asus tuf one coming soon.
But there are no 360 or 540 hz 1080P hdmi 2.1. monitors and even if these monitors would be here 240hz is the max passtrough at the moment on the 2.1 hdmi aver media and asus cards.

Thanks so much. This video was so good. Turns out cloning doesn't add lag at 1080p, but its 4K where the issues happen. Possibly because of Bandwidth!

[Chief Blur Buster]

"Depends"

As of 2023, the latest fast-memory-bandwidth cards can clone a display in under 1ms.

But please, use identical EDIDs (identical display) or EDID overrides (ToastyX CRU).

Some displays emit EDIDs that are 240.001Hz and others emit EDIDs that are 239.999Hz.

Even 4K 60Hz EDIDs can vary slightly -- even a different porch/sync can cause the signals to slew out of phase = LAG

This problem means you will get weird varying sawtooth lag (as the phases of displays slew against each other) if ANY EDID signal numbers are different, such as "Sync" or "Porch" or "Pixel Clock" or "Horizontal Scan Rate" or "Horizontal Refresh Rate" or "Vertical Refresh Rate" is different. They absolutely must be perfectly identical or you will get massively more lag

MOST IMPORTANT TIP: Use fast-memory-bandwidth + use ToastyX EDID overrides to make the signal timings perfectly identical for synchronous mirroring output at darn nearly zero lag (<1ms)

Most people forget to do this.

And also use a high-memory-bandwidth GPU for bitblt cloning or for double-readouts from port transceivers (memory-copying a whole 4K framebuffer in just tens/hundreds microseconds, or doing two memory reads per pixel without slowing down the rest of the GPU).

4K cloning can now go perceptually lagless if you're using RTX 4090 -- MOAR MEMORY BANDWIDTHZ

Or just use identical displays, if you want plug-n-play synchronous outputs. This works best for VSYNC OFF (tearlines are even in identical positions on both displays) or for VRR (displays refresh concurrently). Set display side by side, and flick turn willy-willy high-contrast vertical edges while looking for tearing. You can spot-check by verifying that tearline positions are identical on both displays. If you achieve that, you've got lagless cloning..

The Blur Busters LITMUS TEST: "Identical VSYNC OFF Tearline Positions On Both Displays" = "Lagless Cloning"
*Why? Tearline offsets is a latency, as proven by Tearline Jedi, as tearlines are just raster splices at scanout. If you're familiar with beam racing, raster interrupts or such retro concepts, it's the same thing. I helped Guru3D with RTSS Scanline Sync, and I helped Kaldaien with Speical-K Latent Sync, both raster-interrupt-style beam raced steering of VSYNC OFF tearlines.

YMMV...

[Nolan]

"Depends"

[...]

YMMV...

Are there any tutorials for this sort of thing I have a AverMedia LiveGamer 4K and a Zowie 240Hz with dyac+

[Chief Blur Buster]

Are there any tutorials for this sort of thing I have a AverMedia LiveGamer 4K and a Zowie 240Hz with dyac+

No, this stuff is generally so flippin' niche, that my reply is probably the definitive guide (litmus test for latency).

TL;DR: VSYNC OFF tearline positions perfectly in same positions on both monitors during mirroring = you fixed mirroring lag differentials. VSYNC OFF tearline positions are raster time-based along the display scanout, so if you have the tearing in the same positions, you've eliminated the cable/splitter/mirroring/etc latency difference. Not necessarily the display lags itself nor any GPU lags itself, but at least you'd have solved a latency difference in VSYNC OFF mirroring on your two monitors

[nursejoy]

"Depends"

As of 2023, the latest fast-memory-bandwidth cards can clone a display in under 1ms.

But please, use identical EDIDs (identical display) or EDID overrides (ToastyX CRU).

Some displays emit EDIDs that are 240.001Hz and others emit EDIDs that are 239.999Hz.

Even 4K 60Hz EDIDs can vary slightly -- even a different porch/sync can cause the signals to slew out of phase = LAG

This problem means you will get weird varying sawtooth lag (as the phases of displays slew against each other) if ANY EDID signal numbers are different, such as "Sync" or "Porch" or "Pixel Clock" or "Horizontal Scan Rate" or "Horizontal Refresh Rate" or "Vertical Refresh Rate" is different. They absolutely must be perfectly identical or you will get massively more lag

MOST IMPORTANT TIP: Use fast-memory-bandwidth + use ToastyX EDID overrides to make the signal timings perfectly identical for synchronous mirroring output at darn nearly zero lag (<1ms)

Most people forget to do this.

And also use a high-memory-bandwidth GPU for bitblt cloning or for double-readouts from port transceivers (memory-copying a whole 4K framebuffer in just tens/hundreds microseconds, or doing two memory reads per pixel without slowing down the rest of the GPU).

4K cloning can now go perceptually lagless if you're using RTX 4090 -- MOAR MEMORY BANDWIDTHZ

Or just use identical displays, if you want plug-n-play synchronous outputs. This works best for VSYNC OFF (tearlines are even in identical positions on both displays) or for VRR (displays refresh concurrently). Set display side by side, and flick turn willy-willy high-contrast vertical edges while looking for tearing. You can spot-check by verifying that tearline positions are identical on both displays. If you achieve that, you've got lagless cloning..

The Blur Busters LITMUS TEST: "Identical VSYNC OFF Tearline Positions On Both Displays" = "Lagless Cloning"
*Why? Tearline offsets is a latency, as proven by Tearline Jedi, as tearlines are just raster splices at scanout. If you're familiar with beam racing, raster interrupts or such retro concepts, it's the same thing. I helped Guru3D with RTSS Scanline Sync, and I helped Kaldaien with Speical-K Latent Sync, both raster-interrupt-style beam raced steering of VSYNC OFF tearlines.

YMMV...

I was wondering how is this affected if I was going to clone my display to my capture card (Avemedia Live Gamer Ultra 2.1). I was going to run it at 1440p/360hz to the main monitor and then clone it via HDMI 1440p@60hz to my capture card. Would that cause any issues in regards to lag? All I was going to do is use CRU to strip VRR from the HDMI signal so I would be able to use VRR on my main monitor while capturing with my capture card. I was wondering if I was going to get added input lag this way. I saw this video https://youtu.be/UChcqp62pNA where this person had no added input lag cloning to his capture card and obviously the resolutions are completely different with different EDIDs as well. He just used CRU to strip VRR for the capture card signal (HDMI)

[Chief Blur Buster]

To verify and test, you may want to temporarily turn off VRR on both displays and use VSYNC OFF. Use a tearing test such as RTSS moving line (vertical line that moves sideways). Use RTSS to cap near Hz, to slow down the tearline movement so it's easier to compare tearline positions on both displays. Find out how tearing behaves. If the tearline positions are identical on both, you're generally getting lagless cloning.

If you see a small position offset, then you can approximate the lag differential via how vertically spatially offset the tearline is. If the tearline positions are 1/10th of a display height vertically different positions on the two displays (e.g. tearline is 1/10th of a display height downwards on one display) -- then you're getting very approximately 1/10th of a refresh cycle worth of mirroring latency on one of the displays. This is a clever way to visually estimate display-mirroring latency (e.g. drivers copying a framebuffer to the other display), since tearline position is temporally based on display scanout.

Remember... this visual litmus test only works if you're using VSYNC OFF.

Now you can resume by turning on VRR on both displays. Please note that VRR latency overheads vary from display to display, but as long as you're using the same identical VRR EDID, you will probably avoid mirroring overheads if both displays passed the VSYNC OFF lagless-mirroring litmus test.

[Selexore]

idk how it works . idk how to get rid of VRR with CRU and when i try stuff it infinitely turns off and on my monitor when i clone .