Latency Caused By Power Instabilities Caused by Computer Fans

[joseph_from_pilsen]

The difficulties with REDUCING input lag causing worse aim i can confirm, i tuned down my input lag by disabling ULMB (200 nits on VG259QM is really dimmy even at night) + switching from vsync capped to vsync fast (uncapped) and my input lag decreased for about 3 miliseconds to approx. 6-7ms. The aim is over sensitive now and during prime time its hard to hit anything, its oversensitive to EVERY STUTTER which happens over day. But now at 4am when the servers FINALLY run smooth i had a nice try to deal as much kills in dm without missing head, ended with 68 of 70, 1 opponent crouched and stood up and another one was just standing behind first one who got headshoted So the lower latency of aim is a double edged sword, its like a missing anchor when the aim goes "too light".
Btw the only fan in my PC except the PSU and CPU/GPU fan is me - im a pc fan

[r0ach]

I wonder how AiO’s are impacting this issue if this is the case?

I imagine running PWM fans off motherboard four pin connectors would bring the same issue + additional overhead from your system having to monitor and pulse power on and off in duty cycles. If you're paranoid about your computer having weird issues, you should probably not be using any four pin fans connected to the motherboard in the first place.

I believe most plain old, cheap, RPM fan controllers connected to three pins use a constant voltage and just bleed off the excess amount you don't use as heat onto a heatsink (or eventually blow up the mosfets on an ultra cheap, non-heatsinked model). The four pin models running in PWM mode also supposedly generate more EMI.

I'm not a fan scientist or anything, but I'm only seeing potential cons and no real pros to running four pin connectors off a motherboard for anyone on an input lag forum. For anyone that wants to try removing all fans from the power supply, the easiest option is probably buying a CPU heatsink with vertical fins instead of horizontal, use a dremel tool to cut a circular hole in the cover of the case, then attach a 200-240mm fan (has USB plug then use USB to wall or power strip adapter).

[Chief Blur Buster]

Fans are a known very legit interference source (interaction with power supply, voltage dips, sounds in speakers, or EMI/sparkgap radiation causing ECC-derived lags). And of course other domino effects (e.g. performance throttling when a chip detects electrical instabilities, since downclocking stabilizes a chip during unstable power).

However, the more appropriate forum for electrical issues is Rare/Electrical Issues Affecting Latency — Interference, EMI, EMF.

So I have moved this thread there.

[r0ach]

Fans are a known very legit interference source (interaction with power supply, voltage dips, sounds in speakers, or EMI/sparkgap radiation causing ECC-derived lags).

However, the more appropriate forum for electrical issues is Rare/Electrical Issues Affecting Latency — Interference, EMI, EMF.

So I have moved this thread there.

I don’t really see this as an EMI subject, more like an electrical engineering 101 subject where the fan motors spinning creates noise which then bleeds back up the cable into the PSU. If this problem cannot be solved by things like ferrite beads or in-line capacitors (I actually don’t like these on PCIE cables) to filter out this noise being fed into the PSU due to their operation, then I guess it becomes an electrical engineering 102 problem. I imagine the problem is solvable but the solution raises the fan price 50% so it’s just not done.

[Chief Blur Buster]

Ok, renamed to "Latency Caused By Power Instabilities Caused by Computer Fans"

Yes, there can be other mudane things like performance throttling (when certain chips detect voltage instabilities they sometimes throttle to prevent crashes, caused by power load of other things like fans). Some chips DO automatically downclock when they detect power supply instabilities, because lower clocks are less likely to crash during unstable power. And the variable power transients by poor-power-factor-correction devices (e.g. high power PWM-driven RGB LEDs) can sometimes brownout/ripple a power bus enough to trigger downclocks, increasing lag slightly.

Or other Rube Goldberg / Chaos Butterfly / Domino Effect stuff occuring. That does happen, of course. Often misdiagnosed, but a lot of computers have lots of mysterious lags. Some never discovered properly.

Maybe this is not it, but there's a downright MANY "mystery lags" in the ecosystem.

As humankind pushed the last vestiges of Moore's Law, we pushed safety margins dangerously close, to the point where chips even have to downclock at the drop of the hat when something breathes incorrectly (heat, voltage stability, whatever). So chip throttler logic can trigger on other stimuli than heat -- including unstable power. Like how turning on the laundry dryer machine sometimes dims the lightbulbs in the house; but at the micro scale. It's on the power buses (power supply wires, or VRM on circuit board, and other power rails subsystems etc) and the chip is trying to "protect itself from crashing" (and becoming more laggy as a result). It's like how overclocked chips really need very good/stable power. Those circuit traces don't violate Ohm's Law, and you have brownouts on circuit traces. Now even for unoverclocked chips, we're well inside that dangerous S/N territory, we improved power margins so much (2020s power supplies are so much cleaner than 1990s power supplies), to baby-along modern chips. With all that, humankind necessarily had to inherit a lot more mystery lags.

However, this thread shall remain in this other forum as it is more of a mystery-lag factor.

I'm trying to keep the other forum only more about game-lag optimization, lag-chain talk (USB CPU GPU mouse move/click measurements etc), and talks about lag-tester devices, lag-testing methodologies (etc). Basically that "level" without the S/N being polluted by these niche lag subtopics.

I let these topics exist (there's merit in some of it, just too noisy to let it interfere with lag-tester-device talk), just a separate lag forum. Subject to change at my will over time, as lag discourse changes over time. I've even renamed the forum now, to "Niche Issues Affecting Latency — Electrical, Interference, EMI, Unexpected Causes, etc.".

As long as you respect the way I prefer to silo/adapt this forum, you're good, just honor the rules too. Thank you.

[Hyote]

The fan issue is kind of a complex subject. On my setup it feels like each additional fan brings an additive effect to creating heavier cursor movement (and maybe some types of fans are worse than others - no idea yet). If someone prefers heavier cursor movement for aim they might not see this as a problem. I could see people removing four fans from their PSU and noticing mouse movement feels 1/10th as heavy and actually playing worse.

Some people hit the lottery on their builds and it just winds up ideal for their aim with the random parts they bought. For other people, you have to add or remove components if something is causing a problem. For my current setup, I run 7800xt with modded/stripped AMD drivers installed via driver only mode. Mouse movement is more sensitive than any Nvidia drivers released in the last several years. With the combination of removing all the fans + the stripped AMD drivers, I play better at 1440p because it's actually overly quick/sensitive at 1080p.

So as you can see, depending on your preference for heavier or light cursor movement and each person's hardware variables, issues like this are going to range from no problem to enormous problem for some people. Having four fans plugged in feels overly heavy for me with any combination of variables, though.

Could you share the details of the stripped AMD driver pretty plz