My experience with all sorts of Problems regularly mentioned here.

[ChristophSmaul1337]

It is not matter of being respectful. It is matter of being solid and reliably proving your case. 90% or more input laggers here were or are unable to prove that they face the issue. You don't know if they are good players or not.

It very much is. I don't understand why you would want me to prove my case. To actually prove the case, I would need scientific test equipment used in labs to certify components, some highly specialized tools to locate and quantify interference both from the mains and over the air. I would need to show that interference is higher in my house than in the average house, which would involve conducting a survey at dozens of different homes to estabilsh a baseline. And so on... You get what I mean. Nearly impossible to do.

Yes, there are 90% or more users here that are unable to prove that they face the issue, because guess what? It's very hard to prove the presence of input lag (without spending lots of money for an LDAT uint or a monitor with a built-in measuring tool), and it's virtually impossible to scientifically prove the presence of desync, bad hitreg and overall lag that isn't caused by network lag. With desync, one can always just argue "it's the server, it's the routing, it's the games netcode". You would need full access to the game server, the entire datacenter where the gameserver is hosted, the whole entirety of any ISP's network that is involved in getting the data to you and all the connections that are able to influence your own, like neighbours' lines on a VDSL vectoring connection. You would need to basically stop every bit of network traffic that can influence your connection to the game server to isolate only your own connection and remove all variables from the equation. Good luck with that. I bet more than 90% of users, actually 100% of users, won't be able to do that.

Once I saw a guy who cried about input lag and tried to share some tweaks and then I saw he plays on windowed full-screen.

These are the vocal minority. It's been like this for anything on the internet. There are just a few people who are like that. You just notice them a lot because they're the "loudest" ones in the forum. Just because these exist, there isn't any need to automatically be dismissive about every other user just because there isn't "proof" attached to their posts. Innocence until proven guilt, not the other way around. Give people, like myself, the benefit of the doubt and assume that their problems are real.

I am picking on you, because your case is something that finally brought my attention.

And I am picking on you, because if you were a little bit more open-minded about your approach to different opinions and suggestions, you might already have found a solution yourself, but because you've read something and the person hasn't "proven" it, you might be dismissing it and didnt try their supposed solution which could've been the one that helped you.

While you decided to post a wall of text I was pretty upset that you didn't post any video of you gameplay.

Because I don't have to, once again. To add to everything said so far, these videos aren't even actual proof of anything. Think about it, the 4 examples in CS:GO and VALORANT could've all been lucky shots, coincidences, server lag, or anything you want outside of "desync". My gameplay could've looked exactly like today the entire time. I could've cherry-picked the examples to fool you. But guess what? I am a real person, with a difficult real life, who doesn't have time to troll on a forum. If I were a troll I wouldn't have taken the time to type up wall of texts to give people the whole picture, to make finding a solution easier.

You're lucky I'm a calm and reasonable person, but let me give you some good advice. If this is your attitude towards people looking for help, you'll eventually get in trouble with someone that might not have this much patience. It's not my job to tell you this, just take it as genuine advice.

Enough of this subject now, I won't respond to this particular matter any more.

If you actually decide to order an overkill PSU, be prepared that it won't do anything for you. Good luck anyway.

[gadpaw]

My game time change (more smooth) when i change sseasonic focus gold 650 -> super flower gold 850

[Chief Blur Buster]

It is not matter of being respectful. It is matter of being solid and reliably proving your case. 90% or more input laggers here were or are unable to prove that they face the issue. You don't know if they are good players or not.

Cursed Gamer, please read this reply:

It very much is. I don't understand why you would want me to prove my case. To actually prove the case, I would need scientific test equipment used in labs to certify components, some highly specialized tools to locate and quantify interference both from the mains and over the air. I would need to show that interference is higher in my house than in the average house, which would involve conducting a survey at dozens of different homes to estabilsh a baseline. And so on... You get what I mean. Nearly impossible to do.

Yes, there are 90% or more users here that are unable to prove that they face the issue, because guess what? It's very hard to prove the presence of input lag (without spending lots of money for an LDAT uint or a monitor with a built-in measuring tool), and it's virtually impossible to scientifically prove the presence of desync, bad hitreg and overall lag that isn't caused by network lag. With desync, one can always just argue "it's the server, it's the routing, it's the games netcode". You would need full access to the game server, the entire datacenter where the gameserver is hosted, the whole entirety of any ISP's network that is involved in getting the data to you and all the connections that are able to influence your own, like neighbours' lines on a VDSL vectoring connection. You would need to basically stop every bit of network traffic that can influence your connection to the game server to isolate only your own connection and remove all variables from the equation. Good luck with that. I bet more than 90% of users, actually 100% of users, won't be able to do that.

This is all correct.

While there is EMI induced issues, I need to remind you that EMI problems are so snowflake-unique (could be power supply, could be grid, could be a bad component, etc). It's hard to trace.

While the onus is often on OP to prove EMI issue, the problem is that EMI is such a hard problem to specifically prove (even though EMI problems are defeinitely real). Usually many separate line items in an infinite pie chart, in one of those ultra tiny slices:

If you actually decide to order an overkill PSU, be prepared that it won't do anything for you. Good luck anyway.

Correct. It definitely has helped, but it's not an EMI-problem-solve guarantee. There's over one million EMI problems, with over one million UNIQUE ways to solve the EMI problems -- that's the problem -- I think ChristophSmaul1337 is at least being reasonable in his replies.

It's fine and dandy to goad/coax an OP to try to provide proof of EMI, but sometimes the OP is smart enough to add such disclaimers like these. OP has genuinely solved an EMI problem, in my opinion, but I also am smart enough to know that there's no one-size-fits-all EMI solutions.

NVIDIA engineers are paid big bucks to solve lots of kinds of on-chip EMI stuff, those jobs actually exist, but you can only shield yourself so much. No electronics is immune to nuclear bomb EMI (aka Electromagnetic Pulse), but there's a lot of weaker EMI that still puts a computer to its knees (e.g. lags and slowdowns caused by error correction behaviours etc). Just like Magnitude 9 earthquakes exists, not all cities in the world are built to survive a Magnitude 9 earthquake. Likewise, not all computers are able to resist the world's worst electrical grids (or a malfunctioning power supply, or a bad capacitor on motherboard, or whatever million causes of EMI that has no unique solution...)

While posters are encouraged to give proof of much simpler problem-solves, EMI is so tough to provide proof (Without the equipment companies like Intel, NVIDIA, AMD has), so I appreciate posters that caveat that their solution is not universal (and OP has correctly caveated as such it seems).

There is no use in bullying the OP into providing proof because consumers don't have the equipment that the big companies do. Even chinese manufacturers don't always as well EMI-proof their equipment, especially all the cheapened motherboards and power supplies and other components. Sometimes rebuilds don't solve the problem, and sometimes they do. And so on.

[Chief Blur Buster]

It is not matter of being respectful. It is matter of being solid and reliably proving your case. 90% or more input laggers here were or are unable to prove that they face the issue. You don't know if they are good players or not. Once I saw a guy who cried about input lag and tried to share some tweaks and then I saw he plays on windowed full-screen.

Enough, buddy. OK?

EMI is an asterik exception here. Please read the above why.

I think the OP did an fantastic job following proper EMI "caveating" etiquette, that appears to be above reproach (so far), with proper caveats added.

Please study the Infinite Pie Chart closer, and the million-dollar budgets required for some ultra-thin slices. So the OP gets a free pass here, given the proper caveating done. My bigger concern are the fake (and well-intentioned) "universal EMI solution" posters here, because there are NO universal solution to EMI issues. EMI is a universe of infinite number of frequencies (that you cant see) and an infinite number of strengths (from weaker than a single slow-moving electron, all the way through supernova-league EMP bursts), and no computer can resist ALL of them.

There are NO one-size-fits-all solution. The chase for the EMI Holy Grail, means you generally need to block all those alleged "unversal EMI solution", and remember to be respectful to those people who managed to solve their difficult EMI problem. I know it's easy to be jealous and resentful of those people who did (successfully) solve their unique EMI problem, but sometimes it DOES require million-dollars worth of equipment for a much more broadband (>50%) EMI gamut coverage far beyond those cheap EMI meters. Most people who solve their EMI problems don't even know which frequency and which decibel-strength their specific EMI probelm was, just that they solved it by a change (e.g. component or location or whatever).

Due to the difficulty involved (false negative by cheap EMI tester, and only getting a true EMI positive via equipment costing more than the original rig) -- even providing even partial proof (e.g. swapping power supplies back and fourth) is very time consuming -- and will NOT solve most peoples' EMI problems.

Still -- power supply induced EMI problems may even be one of bigger 1% or 5% odds slices rather than one of those 0.001%-or-less EMI-cause slices, but it's still a major lottery what you do to try to solve EMI issues, and sometimes it's easier to do a shotgum approach (not even guaranteed either, since it's only merely one of the bigger slices of the infinite pie chart) like entire-computer-rig-replacements and offgridding (large power station), etc.

[cursed-gamer]

It is not matter of being respectful. It is matter of being solid and reliably proving your case. 90% or more input laggers here were or are unable to prove that they face the issue. You don't know if they are good players or not. Once I saw a guy who cried about input lag and tried to share some tweaks and then I saw he plays on windowed full-screen.

Enough, buddy. OK?

EMI is an asterik exception here. Please read the above why.

I think the OP did an fantastic job following proper EMI "caveating" etiquette, that appears to be above reproach (so far), with proper caveats added.

Please study the Infinite Pie Chart closer, and the million-dollar budgets required for some ultra-thin slices. So the OP gets a free pass here, given the proper caveating done. My bigger concern are the fake (and well-intentioned) "universal EMI solution" posters here, because there are NO universal solution to EMI issues. EMI is a universe of infinite number of frequencies (that you cant see) and an infinite number of strengths (from weaker than a single slow-moving electron, all the way through supernova-league EMP bursts), and no computer can resist ALL of them.

There are NO one-size-fits-all solution. The chase for the EMI Holy Grail, means you generally need to block all those alleged "unversal EMI solution", and remember to be respectful to those people who managed to solve their difficult EMI problem. I know it's easy to be jealous and resentful of those people who did (successfully) solve their unique EMI problem, but sometimes it DOES require million-dollars worth of equipment for a much more broadband (>50%) EMI gamut coverage far beyond those cheap EMI meters. Most people who solve their EMI problems don't even know which frequency and which decibel-strength their specific EMI probelm was, just that they solved it by a change (e.g. component or location or whatever).

Due to the difficulty involved (false negative by cheap EMI tester, and only getting a true EMI positive via equipment costing more than the original rig) -- even providing even partial proof (e.g. swapping power supplies back and fourth) is very time consuming -- and will NOT solve most peoples' EMI problems.

Still -- power supply induced EMI problems may even be one of bigger 1% or 5% odds slices rather than one of those 0.001%-or-less EMI-cause slices, but it's still a major lottery what you do to try to solve EMI issues, and sometimes it's easier to do a shotgum approach (not even guaranteed either, since it's only merely one of the bigger slices of the infinite pie chart) like entire-computer-rig-replacements and offgridding (large power station), etc.

Why do we talk about EMI? He replaced his PSU and it has solved his issues, how do we know if it's even worth start talking about some EMI? I quit with EMI stuff already long time ago because there is nearly zero chance, that my PC would work exactly in the same shitty way in places located hundreds of kilometers from each other. What if it's just about high current load?

Every PSU has the sticker with current limits on each rail.

3.png (275.84 KiB) Viewed 4512 times

How do you know if particular PSU manufactured years ago is capable of handling modern GPUs, CPUs and MBs in terms of high currents under heavy load (WITH NO PERFORMANCE DROPS)?

[dervu]

PSU's can also behave differently in different environments.
I have read about case in my country directly from PSU manufacturer, where manufacturer could not reproduce high pitch sound on customer's GPU at their place with his PSU, but issue was present at customer place. At the end of the day they had to take 8 PSU models to this customer and only one of them did not have that issue.

[ChristophSmaul1337]

Alright, you've successfully garnered the Chief's attention so here I am responding to it a last time.

While the onus is often on OP to prove EMI issue

It's fine and dandy to goad/coax an OP to try to provide proof of EMI

While posters are encouraged to give proof of much simpler problem-solves

I'm not going to question these rules, and if you insist, then please by any means ban me for not providing this proof, but how would you go about actually validating that some user, who's claimed to have fixed his alleged EMI issue, did indeed fix it? How do you deem something as "proof" of an EMI issue being present? How would you deem something as "fixed"? I don't feel like anybody can do that, really. But again, your forum, your rules.

So the OP gets a free pass here, given the proper caveating done.

Please don't. I don't want to be treated different from others. If your rules call of it, then please take the necessary action. I don't think I should've been given a "free pass" just because I've added a disclaimer. If your rules demand proof, do what's the right thing to do. I can't provide you with actual proof. I didn't know these were the rules, but ignorance isn't protecting from punishment.

Why do we talk about EMI? He replaced his PSU and it has solved his issues, how do we know if it's even worth start talking about some EMI?

Fair enough question. And honestly, I don't even know myself if EMI is actually the reason for my problems and the new PSU is actually mitigating something regarding EMI, or it's an entirely different reason. Despite all of this, I can safely assume that it's not the PSU itself causing the problem. As you can see from the testing I've done (appended table), every PSU i have laying around here eventually becomes "usable", given that I reduce the systems power draw by swapping in low-power components. Another hint for EMI is that I have 2 different 850W units laying around here: Let's call them A and B. Troubleshooting theory 101 tells us that if unit "A" is defective, swapping in known-good unit "B" should then solve the problem, as it's known to not be defective. If both showed the exact same problems, then you can disregard this component as the root of the problem. And that's exactly what I've done. Both 850W units show the exact same problem, so I'm pretty sure that the actual PSU itself isn't to blame. Furthermore, I can use both of those in a low-power draw build and they all work just fine, without showing weird problems.

What if it's just about high current load?

Simple. The example PSU you provided is great.
Let's take my main computer as an example. The CPU, by default, is limited to 125W power draw via the BIOS, but it can go over that for a specific amount of time. That's called TAU, and in my case the CPU is allowed to draw 250W for a duration of 54 seconds by default. So, let's use 250W of power draw for the worst-case scenario.
The GPU is even simpler. It's TDP is set to 300W, because it's an aftermarket card with an improved cooler. The Founders' 2080 Tis have a TDP of 250W. GPUs usually don't come with a TAU or dynamic power limit. The power limit is set via the VBIOS and can be viewed and/or modified with any overclocking software of your desire. I'll assume a power draw of 300W for this GPU.
Now, the PSU example you provided immediately tells me that it has one 12v rail, and it can provide 70.8A on that rail. 70.8A at 12V equates to 849.6W, as shown on the sticker. Adding the power limits from the two components of my computer together, we get a worst-case power draw of 550W on the 12v rail. The PC's fans do also run on 12v, so let's factor in another 50W if you really go all-out and plaster your PC with fans (EDIT: RAM of course also pulls its current from the 12v rail, but typically uses <10W at most). All the other components, like SSDs or USB which are on 5v, are on different rails, so they're not factored into the 12v rail. This means that - when looking at the worst case scenario - my computer would draw 600W from the 12v rail - something that this unit is very well capable of and is not even close to the limit. This unit, on paper, should have zero issues with powering my system, but I can assure you that this unit would also cause my PC to show the usual problems.

How do you know if particular PSU manufactured years ago is capable of handling modern GPUs, CPUs and MBs in terms of high currents under heavy load (WITH NO PERFORMANCE DROPS)?

Because a load is a load. Modern GPUs and older ones aren't different from the view of a PSU, they draw the "same" current (not in an "amount of current" sense, but the same "kind" of current, because there is only one "kind" of current). Current is current, the PSU can't differentiate if it's hooked up to a 2080 Ti or a 980 Ti or a GTX TITAN or a 750 Ti or a GT 710. You know that these "old" PSUs are capable of handling this hardware, because they were capable of handling the highest-end hardware in the past (assuming we're talking about quality 700W+ units here).
If i overclock the hell out of a 4790k, maybe even apply liquid metal and custom watercool it, I can easily push 250W through that thing. Pair it with a watercooled, heavily overclocked 780 Ti (base TDP: 250W) and there you have your "modern" system in 2014, in terms of power draw. And I just happen to know somebody who has used that exact combo with a 750W unit and never had a problem with it. So yes, these PSUs are capable of handling these loads. Sure, these older ones might not be as efficient, meaning that they'll draw more power from the wall than they're actually providing for the PC, but apart from that, there is no difference.

[Chief Blur Buster]

I'm not going to question these rules, and if you insist, then please by any means ban me for not providing this proof, but how would you go about actually validating that some user, who's claimed to have fixed his alleged EMI issue, did indeed fix it? How do you deem something as "proof" of an EMI issue being present? How would you deem something as "fixed"? I don't feel like anybody can do that, really. But again, your forum, your rules.

Your follow-up post about the PSU is one of the "good posts" to me -- I enjoyed reading your above-average-quality followup post. My reply is because just peeved at people pressuring a little too hard you to try to provide futher proof when you already provided a fair bit of detail that is reasonable for an end-user.

In other areas, there's a need for scientific proof of a lot of the Blur Busters initiatives but EMI is a real tough one given it overlaps BOTH
...the Ph.D professionals (there are positions at NVIDIA semiconductor design that has EMI disciplines too!)
*AND* the conspiracy tinfoilhattery
*AND* lots of false-positive
*AND* lots of false-negative
Both EMI placebo and EMI non-placebos galore.
It's very easy for forum to torpedo/derail/threadrap on EMI because of this, so often scares away people.

So it's a very hard topic to moderate since I often lose-lose if I try to moderate one way or another. Metaphorically there's a lot of babies and lots of bathwater (to reuse that old English saying about babies and bathwater). The best solution I've found so far is to silo these into a Niche Issues forum, out of the main view area, rather than banning these topics altogether. I understand why some people wish they haven't started a forum in the first place (it was started over a decade ago as part of a G-SYNC contest giveaway).

OP's can send me a PM to personally request their threads to be closed, if followup-replies are starting to annoy things after a completed (even if partially uncertain-causes) solution. I totally prefer those approaches (OP messaging me to ask for their threads to be marked closed as completed) over letting a completed thread degrade over time with replies that slowly start to derail over time.
Let me know if you want this topic closed on a more positive note pre-emptively.

[cursed-gamer]

Why do we talk about EMI? He replaced his PSU and it has solved his issues, how do we know if it's even worth start talking about some EMI?

Fair enough question. And honestly, I don't even know myself if EMI is actually the reason for my problems and the new PSU is actually mitigating something regarding EMI, or it's an entirely different reason. Despite all of this, I can safely assume that it's not the PSU itself causing the problem. As you can see from the testing I've done (appended table), every PSU i have laying around here eventually becomes "usable", given that I reduce the systems power draw by swapping in low-power components. Another hint for EMI is that I have 2 different 850W units laying around here: Let's call them A and B. Troubleshooting theory 101 tells us that if unit "A" is defective, swapping in known-good unit "B" should then solve the problem, as it's known to not be defective. If both showed the exact same problems, then you can disregard this component as the root of the problem. And that's exactly what I've done. Both 850W units show the exact same problem, so I'm pretty sure that the actual PSU itself isn't to blame. Furthermore, I can use both of those in a low-power draw build and they all work just fine, without showing weird problems.

What if it's just about high current load?

Simple. The example PSU you provided is great.
Let's take my main computer as an example. The CPU, by default, is limited to 125W power draw via the BIOS, but it can go over that for a specific amount of time. That's called TAU, and in my case the CPU is allowed to draw 250W for a duration of 54 seconds by default. So, let's use 250W of power draw for the worst-case scenario.
The GPU is even simpler. It's TDP is set to 300W, because it's an aftermarket card with an improved cooler. The Founders' 2080 Tis have a TDP of 250W. GPUs usually don't come with a TAU or dynamic power limit. The power limit is set via the VBIOS and can be viewed and/or modified with any overclocking software of your desire. I'll assume a power draw of 300W for this GPU.
Now, the PSU example you provided immediately tells me that it has one 12v rail, and it can provide 70.8A on that rail. 70.8A at 12V equates to 849.6W, as shown on the sticker. Adding the power limits from the two components of my computer together, we get a worst-case power draw of 550W on the 12v rail. The PC's fans do also run on 12v, so let's factor in another 50W if you really go all-out and plaster your PC with fans (EDIT: RAM of course also pulls its current from the 12v rail, but typically uses <10W at most). All the other components, like SSDs or USB which are on 5v, are on different rails, so they're not factored into the 12v rail. This means that - when looking at the worst case scenario - my computer would draw 600W from the 12v rail - something that this unit is very well capable of and is not even close to the limit. This unit, on paper, should have zero issues with powering my system, but I can assure you that this unit would also cause my PC to show the usual problems.

How do you know if particular PSU manufactured years ago is capable of handling modern GPUs, CPUs and MBs in terms of high currents under heavy load (WITH NO PERFORMANCE DROPS)?

Because a load is a load. Modern GPUs and older ones aren't different from the view of a PSU, they draw the "same" current (not in an "amount of current" sense, but the same "kind" of current, because there is only one "kind" of current). Current is current, the PSU can't differentiate if it's hooked up to a 2080 Ti or a 980 Ti or a GTX TITAN or a 750 Ti or a GT 710. You know that these "old" PSUs are capable of handling this hardware, because they were capable of handling the highest-end hardware in the past (assuming we're talking about quality 700W+ units here).
If i overclock the hell out of a 4790k, maybe even apply liquid metal and custom watercool it, I can easily push 250W through that thing. Pair it with a watercooled, heavily overclocked 780 Ti (base TDP: 250W) and there you have your "modern" system in 2014, in terms of power draw. And I just happen to know somebody who has used that exact combo with a 750W unit and never had a problem with it. So yes, these PSUs are capable of handling these loads. Sure, these older ones might not be as efficient, meaning that they'll draw more power from the wall than they're actually providing for the PC, but apart from that, there is no difference.

Yeah your calcultions are reasonably fine but you don't know if there was any serious leakage current.

[ChristophSmaul1337]

My reply is because just peeved at people pressuring a little too hard you to try to provide futher proof when you already provided a fair bit of detail that is reasonable for an end-user.

Got it, chief. I'm sorry to have bothered you, and dealing with this should've not been on your agenda in the first place. I could've - and honestly, should've - just not provided videos just to please one person who is sceptical. But on the other hand, I'm not a contentious person, so I just went with it to avoid more discussion.

In other areas, there's a need for scientific proof of a lot of the Blur Busters initiatives

Absolutely, that's totally reasonable and I would love to contribute to it, but sadly I'm a 27-year old basement dweller who simply enjoys playing video games a lot. I was working on PCs since I was 5 years old, so I do have some experience with it, and I've always had a deep interest for how these things work, despite never getting a degree. Most of the time, I find more fun in binge-watching some YouTube channel explaining how every single component in a PC works than to sit down in a university lecture. Heck, I've even built my very own working computer on breadboards. It's just fascinating and I would say that I do have a solid understanding of these things, even without visiting the university. But again, of course, for the real in-depth stuff like EMI, I absolutely agree that this would need at least a university degree to reliably diagnose and deal with. I actually contemplated buying some test equipment for EMI, but despite being in a lucky financial position to put it lightly, I'm just not willing to spend 5 figures on equipment that might turn out to be useless, as for all I know, the problems still couldn't be related to EMI at all.

So it's a very hard topic to moderate since I often lose-lose if I try to moderate one way or another. [...] The best solution I've found so far is to silo these into a Niche Issues forum, out of the main view area, rather than banning these topics altogether.

You're doing gods work here. I am eternally grateful for this forum to exist, because as you said, if you were to post topics like this in another forum, you'd immediately be dismissed as some sort of conspiracy nut, whos of the opinion that games, players, the electric company, the government, the ISP and the electricians of the world are all against you just so you suffer in a video game. I genuinely appreciate the effort that many people go through to make this work, like you and your moderation team, and I also appreciate the general friendly vibes in this subforum. Again, some loud minority doesn't speak for the well-behaved, vast majority of users. There have been a ton of good suggestions in here, and while none of them helped, they were still highly appreciated and could just as well have been the solution.

Let me know if you want this topic closed on a more positive note pre-emptively.

Will do, chief. When figuring out that my problems were gone, I thought it would be great to close the thread right out the gate, but now I'm more inclined to leave it open just a little bit longer. There are some good questions in here after all, and I'm happy to answer them and to maybe even find out why this was happening to my setup in the first place. Who knows, maybe someone can piece it together, the more info I can provide? Maybe this can also help others? And, if it does indeed derail, there's always the possibility to close it later down the line.

Yeah your calcultions are reasonably fine but you don't know if there was any serious leakage current.

Alright, I don't know what kind of "leakage current" we're talking here, so I'll cover the two things that came to my mind first.

The first, what I do understand of the term "leakage current", is when a device is drawing current from the mains but returns the current not over the intended neutral conductor, but rather over the earth/ground wire, for example when there is a ground fault. This isn't going to happen in my case, because german law requires something called a "FI Schutzschalter", a residual current device (or a ground-fault interruptor), to be installed and functional. These things measure the current going out on the live wire, and weigh that against the returning current from the neutral. If there's a discrepancy, the device will trip, cutting the power. My setup, the entire house in fact, is protected by this device, so there's no potential for this kind of leakage current.

The second meaning, although that's not what you'd call leakage current, might be the inefficiency of the PSU. Let's say my example computer needs 600W to operate under maximum load, and the PSU is happily supplying that power, but there is a higher power draw from the wall. This is because PSUs aren't and can't be 100% efficient, and every PSU has an efficiency curve. You can easily measure the power draw from the wall though, and in my practical case, it never exceeds 500W.

Let's take another look at the latter thought. Let's assume that my computer is under the worst-case scenario the whole time. Let's be generous and assume that to be 650W of continuous power draw. This power draw would be about 76% of the PSU's maximum certified wattage of 850W. Here's the efficiency chart for the example PSU you provided:


It's a bit hard to read, but the efficiency at 76% load is roughly about 92%. I'm basing this off the white 240v curve. This means that the PSU will draw 52W more power from the wall than what the components actually need. This would place the overall power draw from the wall at 702W, still comfortably in range of the certified 850W.

All of this is of course ignoring that the rating on the label usually is showing how much DC power the computer can draw, not how much AC it can safely pull from the wall. Which means, if the PSU would be running at 850W, and according to the chart it's 90% efficient at that wattage, it could safely pull 935W from the wall and still be within its rating.