Is it true Ryzen has higher input lag than Intel? Is there conclusive data to prove this?

[deama]

People may have already responded in the course of the topic, but I'm from Brazil and I don't understand much English, making it difficult to understand and read message by message, or watch any video.
I have a Ryzen 7 3700x, 2x8gb 3200mhz and RTX 2070. It means that someone with an Intel processor equivalent to my Ryzen and the same setup, has the most responsive game that I have? How much? Is there any way to improve this on Ryzen to look like Intel? I ask this because I like to play competitively and use a 240Hz monitor for example. Thank you

It depends on your CPU GHz and memory latencies and clocks, you've given your memory clock, but what's your CPU clock speed and what's the memorie's latency speed?

My CPU is around 4.25ghz on all cores. The memory teams (I did a light OC) are:
tCL: 16
tRCDWR: 18
tRCDRD: 19
tRP: 19
tRAS: 38

Memory seems fine, and assuming the 4.25GHz is fixed, so it doesn't boost or anything, always stays at 4.25ghz, then I believe you'll get about 8-10ms less input lag if you upgraded to a good intel CPU and overclocked it to about 5Ghz.
You could try to disable SMT for your AMD CPU, it's AMD's version of hyperthreading; disabling it for me gave me about 4ms less input lag on my 3600 at 4.2ghz, and allowed me to overclock to 4.2ghz from 4.1.
Other than that, you can try doing some windows tweaks, but those would generally improve stability/performance rather than raw input lag.

You could also try to see if you have the dwm still enabled during gameplay, in your case the dwm would add about 5ms of input lag.
If you play games in exclusive fullscreen, and you're not on the 2xxx windows builds, you should be fine as I heard the 2xxx builds have issues with exclusive fullscreen.

[Caronizeeee]

People may have already responded in the course of the topic, but I'm from Brazil and I don't understand much English, making it difficult to understand and read message by message, or watch any video.
I have a Ryzen 7 3700x, 2x8gb 3200mhz and RTX 2070. It means that someone with an Intel processor equivalent to my Ryzen and the same setup, has the most responsive game that I have? How much? Is there any way to improve this on Ryzen to look like Intel? I ask this because I like to play competitively and use a 240Hz monitor for example. Thank you

It depends on your CPU GHz and memory latencies and clocks, you've given your memory clock, but what's your CPU clock speed and what's the memorie's latency speed?

My CPU is around 4.25ghz on all cores. The memory teams (I did a light OC) are:
tCL: 16
tRCDWR: 18
tRCDRD: 19
tRP: 19
tRAS: 38

Memory seems fine, and assuming the 4.25GHz is fixed, so it doesn't boost or anything, always stays at 4.25ghz, then I believe you'll get about 8-10ms less input lag if you upgraded to a good intel CPU and overclocked it to about 5Ghz.
You could try to disable SMT for your AMD CPU, it's AMD's version of hyperthreading; disabling it for me gave me about 4ms less input lag on my 3600 at 4.2ghz, and allowed me to overclock to 4.2ghz from 4.1.
Other than that, you can try doing some windows tweaks, but those would generally improve stability/performance rather than raw input lag.

You could also try to see if you have the dwm still enabled during gameplay, in your case the dwm would add about 5ms of input lag.
If you play games in exclusive fullscreen, and you're not on the 2xxx windows builds, you should be fine as I heard the 2xxx builds have issues with exclusive fullscreen.

What is DWM? How did you get to those game delay numbers?

[deama]

What is DWM? How did you get to those game delay numbers?

If you go through the pages on this thread, some people posted videos of a guy comparing a 4.2ghz ryzen cpu vs a 5ghz intel cpu, he got an input lag difference of about 8-10ms I believe. I think he also did hyptherthreading off/on comparisons? Also, he did a ryzen comparison with an intel cpu with similiar ghz ratings, and that gave only about a 2ms difference or so.
Also, I've done some measurements myself using a 240fps camera on my setup, I got 4ms less input lag by disabling SMT and overclocking cpu to 4.2ghz from 4.1ghz. Keep in mind that some games can use the hyperthreading to give you more fps, so you might lose some fps by disabling hyperthreading, on some games.
I did tests with memory as well, but it didn't seem to affect much of anything, maybe mouse felt a bit more stable if I modified a specific latency timing.

DWM is the desktop window manager (or desktop compositor), it used to be called aero back in windows 7.
It adds vertical sync to everything, but that ends up adding input lag.
You can easily disable it on windows 7 by switching to a non-aero theme. On windows 8 it's a bit finnicky to disable, but can still be done fine. On windows 10 it's pretty hard to disable, generally it's not recommended as it breaks a lot of things, can't even run dx12 with it disabled. But on windows 10 it's supposed to disable by itself when you start playing a game in exclusive fullscreen mode, but that's started to change back in 19xx windows version, they did something and it kinda broke it, so you'd have to do a lot of tweaking to get it to work again. They recently broke it again on 2xxx windows version, I think you can still get exclusive fullscreen to work properly, but it's more finnicky to get working.

Generally the dwm adds 1 cycle of your monitor's hz as input lag, so if you have a 120hz, it'll add about 8.1ms +2ms more in games because of some extra processing. If you're on 240hz, it would add about 4.1ms +1 or 2 more ms in game. I measured mine as adding 10.3ms or so with the dwm on, on a properly configured RTSS scanline sync setting, without the scanline sync setting the ms would fluctuate between 15-18ms, some games it would be 11-16ms, sometimes just a 1-2ms fluctuation, depends on the game, highest was 18ms for me. I've got a 120hz monitor, so I'm not sure how much this would equate to a 240hz, but it would give a minimum of 5ms of extra input lag I would imagine, if you have scanline sync setup properly, if not then probably more, and it could also add in fluctuations that may make it even worse.

Try this video where he talks more about dwm and how to disable it in the newer versions of windows:
https://www.youtube.com/watch?v=7FSzofnsmW4

[axaro1]

Also, I've done some measurements myself using a 240fps camera on my setup, I got 4ms less input lag by disabling SMT and overclocking cpu to 4.2ghz from 4.1ghz. Keep in mind that some games can use the hyperthreading to give you more fps, so you might lose some fps by disabling hyperthreading, on some games.

4ms is exactly in the margin of error if you use a 240fps camera with a 240hz monitor, I wouldn't consider it a reliable benchmark unless you are using a 960+fps camera.

Do you concretely feel a difference? I may give SMT off a try...

[deama]

Also, I've done some measurements myself using a 240fps camera on my setup, I got 4ms less input lag by disabling SMT and overclocking cpu to 4.2ghz from 4.1ghz. Keep in mind that some games can use the hyperthreading to give you more fps, so you might lose some fps by disabling hyperthreading, on some games.

4ms is exactly in the margin of error if you use a 240fps camera with a 240hz monitor, I wouldn't consider it a reliable benchmark unless you are using a 960+fps camera.

Do you concretely feel a difference? I may give SMT off a try...

It was always 1 frame difference when I did a bunch of tests with and without SMT and the different clocks. At 4.1ghz without SMT, it didn't produce a difference, but disabling SMT and overclocking it to 4.2ghz produced a 1 frame difference, so about 4.16ms. The idea here is disabling SMT will allow you to more easily overclock your CPU, I'm not sure if disabling SMT itself would yield a difference as I'd need a more precise camera for that from the looks of things.

I didn't really notice much of a feel difference, but I'm generally bad when it comes to these things, I asked a friend of mine to come and test it for me, and he said he could feel a small difference.

[axaro1]

It was always 1 frame difference when I did a bunch of tests with and without SMT and the different clocks. At 4.1ghz without SMT, it didn't produce a difference, but disabling SMT and overclocking it to 4.2ghz produced a 1 frame difference, so about 4.16ms. The idea here is disabling SMT will allow you to more easily overclock your CPU, I'm not sure if disabling SMT itself would yield a difference as I'd need a more precise camera for that from the looks of things.

I didn't really notice much of a feel difference, but I'm generally bad when it comes to these things, I asked a friend of mine to come and test it for me, and he said he could feel a small difference.

I tested SMT, I'd say that I felt a very small improvement, like maybe 1ms better (I definitely notice the +2ms when I enable strobing on my monitor).

Too bad that my fps tanked in both Overwatch and Rainbow Six so I actually got worse frame times due to the way the game engine handle 12 threads vs 6.

[Boomchakadah]

Is it true Ryzen has higher input lag than Intel?

No.

phpBB [video]

[MaxTendency]

Is it true Ryzen has higher input lag than Intel?

No.

Okay I'm confused, did we even watch the same video? Because to me it seems like the answer is a resounding YES. And this is coming from a 3700x owner so there's no bias here. I think everyone is familiar with the concept of 0.1% lows and why "Avg fps" graph without a "0.1% low" is quite useless, since you can have high avg but still stutter crazy.



This is the csgo test most of intel's entries are around 13-22ms mark (underlined blue) where for ryzen most of the entries are at 19-28ms and it even spikes up to 37ms! (underlined orange) Not to mention intel has more than double the entires at 13ms mark compared to ryzen.



Fortnite similar results , ryzen spiking WAY above intel. Also majority of intel's entries are at 14-16ms where ryzen majority of entries are at 16-18ms mark. Once again intel beats ryzen on the lowest entry of 12-14ms mark, though not by a huge margin compared to csgo test above.



Same story on overwatch, if I'm eye-ing this correctly on the lowest entry intel scored almost 3x higher than ryzen, while on the highest lag entry (28-30ms) ryzen is spiking almost 3x higher compared to intel. With majority of the ryzen entries sitting at 24-26ms mark while intel sits at 20-22ms mark.



Once again ryzen is spiking way higher than intel. Although ryzen does beat intel on the lowest entry point this time.


If we're just gonna look at avg and completely disregard the spikes then we might as well go back to benchmarking fps without including 0.1% or 1% lows (which would be a HORRIBLE idea btw.)

[1000WATT]

Common sense prevailed in the end.

[axaro1]

It might be placebo but I find that using ISLC to set the system timer to 0.5ms feels so much better on my ryzen machine with HPET ON in BIOS(OFF in Windows strictly with disabledynamicticks YES, useplatformtick YES).

Lately I switched back to Game Mode ON after playing with it OFF for 6 months and I truly don't know what kind of optimization it does create on the Windows ecosystem but my mouse completely stopped feeling floaty or laggy ( Microsoft didn't really disclose what it does specifically, I feel like it improves the way my input is processed, like a pass-through directly hooked up to the game engine but it most likely just disable background services).
From my testings Game Mode ON makes a huge difference in games like Overwatch.

I highly recommend everyone here with a ryzen system to try the two tweaks I described, it truly makes a massive difference.

I use some other tweaks like MarkC mouse fix however my system is not too tweaked except for stripping the OS from telemetry/bloatware with BCUninstaller (with a built in registry cleaner).