Special K can drastically reduce latency

[jorimt]

In your case it seems a bit different with the case that they pretty much know who you are, it probably stings a bit. But I assume that the people who linked to your past works are not in the correct department for marketing and collaboration, thus they most likely didn't have the power (or thought) to do these things.

Thanks, I know some who have connections, but I myself have zero. Everything I've done regarding this has been virtually all volunteer-based, and I haven't even considered monetizing any of it.

The squeaky wheel gets the oil, and that I am not; I haven't a self-promoting or self-marketing bone in my body, and I think that shows in my (nonexistent) relationship with Nvidia. I also don't announce myself in other outside forums, so have gotten treated like a rando in threads such as Guru3D on one than more occasion (of which I now avoid posting in for their sake; too many cooks in the kitchen, apparently).

But enough about "An Introvert's Guide to Being Overlooked" ( ), again, appreciate the tip. I've since contacted Andy via ResetEra and will update here if he replies (no matter the answer).

If I do manage to get my hands on one, users here can expect to eventually start seeing test results for various synced scenarios on the forums (in my off-time; have a day job). That said, I don't yet know the qualifications or stipulations for receiving one, so could very well end up being a nonstarter.

[diakou]

Extra post showcasing framerate limiting aspects of special k in horizon zero dawn vs other external framerate limiters before a lot of major updates the last 1-5 days

keep in mind that these are not the latency oriented claims done earlier in this post, those will eventually come soon. These are simply showing the immaculate framerate limiting methods/performance by special k in a very controversial performing game on PC (horizon zero dawn)

https://babeltechreviews.com/the-horizo ... -review/3/

[jorimt]

Reach out to AndyBNV on ResetEra, use the start a conversation tool. Nvidia didn't contact Kaldaien first either.

An update, as promised; I'm on the list for the next shipment. No apparent stipulations or requirements for me (as of yet) beyond my contact info and address. They must really want to get it tested by interested parties.

Again, greatly appreciate your heads up, as I wouldn't have imagined I qualified, let alone known where to make contact.

As for Special K, assuming he and I receive the device around the same, I will wait for his results before I start testing or posting any of my own for supplement or corroboration. That said, once I break it in, barring any possible NDA (timed or otherwise) on result sharing, visitors can ultimately expect to start seeing some test scenarios and numbers generated from the device around here periodically, so long as it ends up producing useful/valid results (and according to Gamer Nexus tests, it does when compared to a 1000 FPS [EDIT: 240 FPS; hmm, I might have to run my own verifications after all] high speed setup: https://youtu.be/0SZ7ZCac38A).

Future Blur Buster material isn't out of the question as well, but I'll have to consult with the Chief on that one.

[diakou]

Reach out to AndyBNV on ResetEra, use the start a conversation tool. Nvidia didn't contact Kaldaien first either.

An update, as promised; I'm on the list for the next shipment. No apparent stipulations or requirements for me (as of yet) beyond my contact info and address. They must really want to get it tested by interested parties.

Again, greatly appreciate your heads up, as I wouldn't have imagined I qualified, let alone known where to make contact.

As for Special K, assuming he and I receive the device around the same, I will wait for his results before I start testing or posting any of my own for supplement or corroboration. That said, once I break it in, barring any possible NDA (timed or otherwise) on result sharing, visitors can ultimately expect to start seeing some test scenarios and numbers generated from the device around here periodically, so long as it ends up producing useful/valid results (and according to Gamer Nexus tests, it does when compared to a 1000 FPS high speed setup: https://youtu.be/0SZ7ZCac38A).

Future Blur Buster material isn't out of the question as well, but I'll have to consult with the Chief on that one.

Nice! That's great to hear, I need to become friends with Darktalon at this point, he's made multiple people at this point reach out and get the tools they deserve, the latency warpath needs more people like him who are both well-spoken, resourceful, helps people reach out and is knowledgeable on many areas (He's made a nice post about why 10gb of VRAM is enough for now in the outcry of "VRAM being too little" over at resetera https://www.resetera.com/threads/vram-i ... gh.280976/ He went against the grain on this one, but has since had notable people recognize this thread including Andrew Burnes from NVIDIA)

Although you mention it in your post, I hope you'll be able to test a few "competitive" games with Special K and see if there's benefits in competitive games (given that anti-cheat has it whitelisted / the game has no anti-cheat) when the framerate limiter version of it reaches final stages (I think he wants to make it stand-alone, for easier integration for other devs/people) If I could make a personal request, I'd love to see how it would do in fighting game ports on PC like Street Fighter 5 / Tekken 7 / DBFZ etc I can gift you the games on steam myself. BUT! These things are all done in free-time, just saying for the unknown future when you get the time and Kaldaien has done his tests as well.

The Gamers Nexus video was very detailed on LDAT, indeed very helpful for future LDAT users without having to tinker themselves for a long time.

[jorimt]

He's made a nice post about why 10gb of VRAM is enough for now in the outcry of "VRAM being too little"

Not my primary subject of focus, but from what I know, yes, an efficient game will take advantage of (aka cache) as much VRAM as you have, but it doesn't mean it is all being used at once, even if filled, so that worry is unfounded. 10GB, even at 4k, should be plenty for the time being (especially with the new management methods available and implemented).

Traditionally, the only thing that should be worried about, is if there isn't enough VRAM to prevent swapping to the page file or system DRAM, and I don't see that being the case here for a while. Memory speed is more important in this instance, and the 3080 has it. That, and more memory at these speeds would have brought up the price of the 3080 significantly, so I think it's an acceptable balance.

Although you mention it in your post, I hope you'll be able to test a few "competitive" games with Special K and see if there's benefits in competitive games

Now that I'll have a method of testing input latency that won't take weeks or months at a time (my original setup was extremely low-res and time consuming), I'll be able to explore those scenarios much more frequently. I obviously can't provide an ETA, but your requests have been noted.

[disq]

Congrats @jorimt, well deserved getting LDAT!

Thanks to @Darktalon for sharing the info and of course to AndyBNV from ResetEra as well!

All for better findings

[deama]

I did some amateur tests using a 240fps camera on a 120hz screen (with black frame insertion), the results were basically the same as using rtss.

I'm on windows 8.1 and I used grim dawn for my test game and the frame rate has been capped to 120 in all cases, except for the rtss case, I used scanline sync at the specified value (gets rid of the tearline for me at that value) in the rtss scenarios.

no dwm:
special k -- 8-9 frames
special k with reduced input lag mode -- 8-9 frames
rtss scanline sync @1280 -- 8-9 frames

with dwm:
special k -- 12-13 frames
special k with reduced input lag mode -- 10-12 frames
rtss scanline sync @1280 -- 10-12 frames

I was hoping special K could deliver input lag comparable to without dwm while dwm was on, but it doesn't seem like it.
I've zipped the video files for anyone who wants to check out, but it's 2GB and my upload speed is pretty bad, so I'm gonna later edit this post with the link, probably tomorrow as I'll leave it overnight to upload.
EDIT: https://drive.google.com/file/d/1DyJiOq ... sp=sharing

I'll stick with RTSS because of the scanline sync feature wherein I can adjust the tear line, special K doesn't have that, I asked in the forum if there was a way or if he was planning to add it but he said it was unnecessary, which he's right, it's not, if you play with the dwm, but if you disable the dwm, or go into fullscreen exclusive (which should disable it), then it becomes a problem.

[jorimt]

Congrats @jorimt, well deserved getting LDAT!

Thanks to @Darktalon for sharing the info and of course to AndyBNV from ResetEra as well!

All for better findings

Thanks @disq

They're waiting on a new batch (ETA is sometimes this month), but I'm on the list to receive one.

[andrelip]

It could not beat CPU time + GPU time, which is the minimum amount of input delay in a scenario where the CPU is faster than GPU for all frames (the queue should be under 1 for almost all the frames / you can test that using Frame view). What should be going on is that the hook method has some side effects, which is exploiting the measurement. Sidewinder said many times that RTSS decreases input lag, which is a lie as you can check by all button-to-pixel tests. You cannot reduce input lag by adding artificial delay; you can only avoid GPU-queue using it.

Also, frame-pacing almost always is an illusion. Your frame could be presented to the display at a fixed interval, but if the CPU timer differs too much from each frame than your animation will be all messed up. Most timers like RTSS hook present(), but that happens after the positions were set by the CPU, so it will do nothing to make things smoother. It does help to stabilize the total time so you can control the time that the CPU starts processing, but you cannot control what happens inside this processing pipeline and how long it will take. The GPU part is just to draw the scene, every position and animations have already been chosen.

[Chief Blur Buster]

An update, as promised; I'm on the list for the next shipment.

Me too.

Also, I also have an internal BlurBusters-developed LDAT-like tool (that works on all GPUs) that I am considering commercializing, but right now it's mainly exclusively used for my contracting services and Blur Busters Approved. Trying Kickstarters and hardware sales is too risky during COVID at the moment. However, I'm open to commercialzing at some point in the future, as I have features LDAT cannot do.

Also, NVIDIA agreed that LDAT and other lag test tools should have consistent industry standardized results.

Improves trust in lag tests.