[Futurist Technical Talk] VSYNC is Obsolete: Future Framerateless Displays

[Chief Blur Buster]

Futurist thread:

Some talk in a different thread opened a pandora box about VSYNC, so I've split this into a separate subtopic.

<PandoraBox state="OPEN" interlock="OVERRIDE">

entire industry cannot snap out of using 1940 display vsync tech for monitors. response time is measured for marketing(on 25 deg celsius, with overshoot so huge they have ghostbusters on the sides with guns to bust Casper after they perform the measure-ritual). PC performance is measured in fps which is a useless metric only to be used as basic check of "playable". there is none who is aware of temporal vectors describing reality and how it translates to man made reality(video games). for example a factor vector of frame distribution, frame latency, consistency, skew... are all great measures. but such reasoning exceeds human perception.

On this note, there's an Thread in Area 51 Display Science, Research & Engineering discussing why the raster methodology has been around for a long time.

It's older than that. Raster sequential generation of pixels is actually more than two centuries old (Jacquard loom in year 1801, weaving pixellated patterns automatically in fabric). And early 1870s telegraph fax machines used a pendulum to raster-scan across a surface (with metallized ink to complete a telegraph circuit).

Raster is a conveniential serialization of a 2D image into 1D mechanism (data transmission, whether the 1801 loom punchcards running in an infinite loop, or a 2020 DisplayPort connection spewing binary at a display), and will continue to probably be viable methodology for a long time to come.

You could do other things like using video compression or JPEG images or whatever, but uncompressed and (near)lossless compression tends to be fundamentally serial anyway. One might start to send 3D geometry instead, where the display uses a futuristic GPU to decode theoretically H.268 or H.269 framerateless resolutionless streams. Then we can say goodbye to raster. But raster workflows will still be needed for a long time to come -- well beyond the next century, too.

Eventually vsync will just be a 1-byte comma separator (actually, DSC almost sort of does that already, due to the redundant blanks/zeros in there), but vsync will still exist between frames of a finite-framerate delivery methodology.

</PandoraBox>
ERROR 1001: Failed to close the PandoraBox element. <PandoraBox> is still open.

[schizobeyondpills]

its good research only if used to create a solid and actually valid research by conducting a non flawed benchmark with a properly optimized system using an environtment in which subjects are not aware of being tested(analytics on a tournament). using it to prove impact of latency in esports not so much.

one day maybe someone will figure out that Pythagorean theorem is not about triangle but about lower dimensional intersection to higher one. from 1d lines into 2d plane and that it can be used to compute necessary latency bottleneck of sending a 2D frame down a 1D protocol. (so u dont walk x+y steps to get to a 2d point by bottlenecking yourself in one dimension. ) such as current display tech. and then imagine if some schizo opens the pandora box of doing that for 3rd dimension of per second of frames as 4D protocol! such wizardry is black magic even for tearline Jedi. unfortunately as i said display industry is too busy figuring out how to transfer all this huge pile of data over a broken bridge that collapses under a leaf falling onto it, so they make you lose sight until they transport frames across the river underneath(G sync)

as i said previously. one must wait for a system to burn to ashes and collapse to build a new one on firm and evolutionary solid foundations of a flawed fallen system to be able to fix reality. doing it sooner while the flawed system still exists only shows impatience and rushing to mess up cycles of evolution.

esports is seen as some apex of engineering integrity inside this flawed evolution cycle where people rush to compensate into what they need being the right thing as if they are the center of the universe. this whole thinking of "i dont play esports at the top to need good response time display" is the reason why we have so many flawed engineering in all of our products. in actual truth esports is at the bottom of engineering integrity pyramid while evolution and pushing bleeding edge tech and optimizing it sits at the top.

theres no need for vsync. its flawed thinking. only latency is the limit between human engineering (digital) and analog reality.

maybe it takes an insane schizophrenic to reinvent the wheel and turn it into a 4d sphere

[Chief Blur Buster]

theres no need for vsync. its flawed thinking. only latency is the limit between human engineering (digital) and analog reality.

Terminology Disambiguation Alert
- VSYNC as in driver setting
- VSYNC in the perspective of framerate-vs-refreshrate motion smoothing (fps=Hz)
- VSYNC as in an analog signal embeded between refresh cycles, as part of VBI
- VSYNC as a wasteful digital equivalent of analog signal
- VSYNC as a miniaturized marker or comma-separtor between compressed frames
- etc, etc, etc.

For any finite refresh rate system system (VRR or non-VRR, high-Hz or low-Hz), a derivative of a frame separator (synonym of one sub-definition of VSYNC, under our Blur Busters terminological standard (which is sometimes optimized to fit advanced-mainstream; whether you like it or not).

Another word is VBI (Vertical Blanking Interval) of which the VSYNC signal is a subset thereof. Sometimes VBI and VSYNC is used interchangeably (by accident or by intention) because they separate refresh cycles, though VSYNC as a signal, is just a subset of VBI (also including Vertical Front Porch and Vertical Back Porch); also covered in Custom Resolution Utility Glossary as well as Quick Frame Transport Threads or the Next-generation scanrate: Global Scanrate/F-SRR for readers who want to read some Popular Science style explanations of the raster video signals and their synchronization-related markers such as VSYNC or VSYNC-equivalents.

NOTE: Obsoleteness of VSYNC as a motion-smoothing technology definitely increases as the higher the frame rate and the higher the refresh rate, a display begin to behave like a per-pixel variable refresh rate display, thanks to the ultrafine granularity of refresh cycles. That's why 50fps stutters are dramatically less at fixed-Hz 240Hz than with fixed-Hz 240Hz. Stutter amplitudes 1/4th as much, and stutter/jitter often at higher frequencies that begin to blend into flicker fusion thresholds and/or the display's own persistence motion blur (tinier stuttiers hidden in the display persistence when running low framertes on a fixed-Hz ultra-high-Hz display). Latencies of VSYNC ON, VSYNC OFF, and VRR also converge closer to identicalness (zero), the higher the refresh rate you go, so this is one angle of "VSYNC becoming more obsolete". At 1000Hz (and continuing well beyond where engineering-feasible and beneficial; Vicious Cycle Effect permitting); almost any framerate looks like native Hz / VRR framerate (on lower Hz) even if the 1000Hz is a fixed-Hz monitor.

However, from a signal layer topology, a form of synchronization separator between refresh cycles will likely continue to be used for a long time for fixed-Hz and variable-Hz -- as long as framerates/refreshrates are still finite. But the old-fashioned "VSYNC" display signal terminology will continue to carry-over to such refreshing workflows, even if VSYNC becomes slowly a terminological misnomer when it's just a H.268 frame marker flag or a GPU-framerate-amplification delivery mechanism instead of a raster based delivery mechanism, or some other as-yet-invented display signal that replaces a raster workflow. Until things go framerateless encoding -- if ever -- though even that probably will still have things like a 'recommended display rates' mechanism encoded within;

VSYNC terminology is fuzzy but... Until framerates become analog (framerateless encoding mechanisms), synchronization markers or timestamps or frame-separators or refresh-separators (roll your dice and choose your favourite term) continue to be necessary in all framerate signals, files, and encodings. Whether be H.264 compression or be a DisplayPort or an analog TV broadcast or when boundaries later starts to blur (e.g. framerateless DisplayPort 7.0 in year 2043 or 2046 or whatever becoming a subpart of framerateless H.269 or H.271 standard or whatever) -- and sometimes VSYNC terminology are appolied to these anyway when used in an interactive display context, to chagrin of some by-the-book technicians. But for the purposes of "all-definitions-of-VSYNC-obsolete", it likely won't happen until finite-framerate systems become future conceptual framerateless/vectorized/4dimensional/timecoded-photon/whatever encoding workflows.

Yes, we even think that far ahead to the eventual sunsetting of raster this century.

Mind you, VSYNC-required workflows isn't necessarily superior or worse latency-wise; considering the overlapping latency venn diagrams of ultralowlag high finite-framerate workflows and slightly more laggy complex-encoded framerateless workflows. There's electronics overheads, whether from processing/transmission/decoding/etc considerations.

Also, thanks to limits (like the brick wall of CPU processing speeds stopping increasing) -- higher refresh rates and display rates may require some parallelism in the display, see Custom OLED Rolling Scans -- Custom Built OLED Monitor but that can also piggyback onto Modularized jumbotron LED panels, which impacts latency considerations (delivery and/or buffering for shingled parallel refreshing of multiple trailing refresh cycles).