I read somewhere, that it was not recommended to exceed some values to exceed 20W. And therefore damage the internal components of the Switch:
CPU: 1428 MHz
GPU: 768 MHz
The overclock configuration 2387/1536 seems dangerous, doesn't it?
CPU: 1428 MHz
GPU: 768 MHz
The overclock configuration 2387/1536 seems dangerous, doesn't it?
Tried it but had to revert for now whilst I debug my setup, I noticed that even though Exosphere ini etc... was enabled, my switch was not blocking the Ninty servers when I checked in the 90DNS app!! Just warning, I checked on the light side and my account seemed ok, but in case anybody gets themselves blocked.
The console has both software and hardware overheating protection. If the console feels uncomfortable, it will let you know (then go to sleep mode). If you disable software overheating protection, the console will simply shut down instead. I couldn't turn off the hardware overheating protection even if I wanted to. In view of only these two facts, further controversy about the possible harm is meaningless. I think so.
But the final decision is up to the users.
They shouldn't be there. Unless I accidentally left the modifications at the last update. I'll double-check.
Oh, yeah, of course. Their probably questionable testing procedure on a single console/sample also doesn't result in everything immediately exploding after a few minutes, so it's definitely safe and definitely works on all consoles.its fine, its not @ 136C
No but the original developers can ask the admins to take it down as it violates licenses.
All homebrew modifications to factory settings that go well beyond stock parameters are potentially dangerous and can damage a system - this has been a known quantity in the overclocking scene since the dawn of time. People go as far as soldering VRM bypass planes, flashing custom BIOSes or tricking sense resistors on GPU’s to remove power limits, risking having their cards (often worth thousands of dollars) combust like a firework. The “danger to users” concern is waived the moment a user modifies the console at all, I don’t consider that a valid “complaint” - everyone accepts this as a possibility when they choose to overclock. If a user doesn’t know how much power draw the VRM can take, or how much cooling capacity they realistically have, they shouldn’t be overclocking their system. That’s the user’s risk to take, not ours to police.
I do acknowledge a different complaint - namely, that this is originally your software, from what I understand. Is that correct? In order to re-release your original or modified code, the OP has to comply with the license it was originally released with. What they don’t have to do is keep their word that “if the original creator says this isn’t okay, they’ll take it down”. That’s just being an asshat, but it’s not illegal.
So, what is going on here, exactly? As long as the risk is a known quantity and is telegraphed to potential users, I don’t see an issue. This is a very simple disclaimer that needs to be included - “if you put X times more current through the VRM than it was designed to handle, or grossly overheat the SOC, or draw more current from the cell than its rated discharge current, your Switch may or may not turn into a fireball”. Is that what’s required? Because I can pin a message like that very easily.
If you have a different complaint, you should use the Report button. We’re not the copyright police, but we do enforce copyright-related legislation as is required of us, and to the best of our ability. Drama inside a thread is not considered a formal complaint.
Sys-clk itself has a beerware license so nothing is wrong there
The inclusion of Atmosphere and other tools does either require them to post source of some sort regarding any modifications to those or that the sys-clk modifications be released without those included afaik.
currently OP is not complying with MIT or GPL v2 which most of the included software in this minus sys-clk is licensed under.
My biggest issue besides license compiance is that the OP doesnt appear to entirely understand this either and is willing to tell people its 100% ok as long as temps arent high. This is factually false and needs to be stressed that on Erista specifically this is very much use at your own risk (and thats a high risk). Mariko has more room but also at a higher point its the same situation. Theyre going to do it anyway as has been seen and folks have asked them not to simply to protect the multitude of people that know just enough to be dangerous but oh well i guess.
The risks need to be clearly laid out and misinformation that this is perfectly fine to do needs to not be spread.
If people wish to experiment with stuff to see how far shit can be pushed, I'm all for it, that's what I originally wrote and distributed the patches for. The problem here is this user is distributing and advertising precompiled builds of these patches for more "normal" use, which they were never designed for and should never be used for, they can very quickly permanently destroy a console if used in that way (especially with the kinds of clocks/voltages they're including). Distributing them like this is dangerous, foolish, and irresponsible. Less of a problem, but they're also missing some critical information (eg. half of the time everything will be listed as if it's running at the higher speeds and you'll have the higher power consumption, temps, etc, but you won't get the performance benefits (it might even perform worse)).
I'm aware it's legal and I generally don't give a fuck if people use my stuff. But as you said they're an ass for saying they'd take it down if anyone was against it then proceeding not to after I responded to them within minutes saying I'm deeply against it. I'm trying to convince them to not distribute the Erista stuff because I don't want to see a bunch of valuable Eristas be destroyed for no reason.
In that case the OP needs to make their release compliant. Once he reads this message, he’ll be aware that he needs to read the license files and comply with them to redistribute - we can revisit the issue then. More thermal and power headroom is one thing, but users are still perfectly capable of making modifications to the hardware to accommodate for shortcomings of “less capable” silicon. I should know, one of my PS4’s has a heatsink I made myself out of a slab of pure copper skived fin stock. Will the average user do that? No, but that’s not anyone’s responsibility. As long as they know the risks, everybody’s happy.
Everything depends on what you consider to be “higher performance” - some people just chase the gigahertz for the sake of a higher number and don’t care about introducing microstutter or other assorted mess which reduces real life performance - they’re perfectly happy with “bigger number, and stable”, and I’m happy with allowing that - different strokes for different folks. If someone wants to hook up their Erista to an external VRM and pour LN2 all over it, first of all god bless them and second, can we all see? I’m generally against artificial limitations, but I do agree that any patches like this need to be distributed with a clear notice that they are to be used at the user’s own discretion and only the user is responsible for any incorrect behaviour or hardware failure, as it’s been the standard in the modding community since forever. The weird trend of gatekeeping and “protecting” the users from themselves is not something that I ever understood. It’s wrong when the original manufacturer locks users out, it’s no less wrong when homebrew applications meant to fix that problem do the same. There’s a difference between supplying suggested absolute maximums that are broadly considered “safe”, completely another to try and stop users from going beyond them. From my point of view, if I’m a dumb-dumb and I want to set that SOC to 10GHz just to smell the smell of napalm in the morning, that is my right, and I apply that principle across the board.
Thank God, I see an adequate, logical and detailed opinion! I was just thinking, maybe something is wrong with me, and I really do something bad. As if my opponents are not trying to dissuade me, they are trying to save humanity from extinction. And for salvation, they do not hesitate to resort to dirty methods. Distortion of facts, taking out of context, lies and misinformation, and so on. Simplest example:
Some are very disingenuous, claiming that I ignore the protests of the authors of the original code. The fact that the code of one of the participants in the discussion was taken, as a basis by the developer of the project (one from lots used) that I used in the modification does not mean that there is something left of his (original) code. I only mentioned him among the creators because he was mentioned in the "special thanks" section of one of the projects taken as a basis. It was thoughtless, I will correct this oversight.
It's hard for me to even guess what they are guided by when they claim that I violate the GPL. I answered them several times that I did not pursue the goal of appropriating the laurels of the original authors, and will gladly publish the source code of the project as soon as I bring it to a stable version, of which I will be sure that it is safe. I will put it in order, document it so that it would not be a shame to place it in the repository.
As someone who prefers the scientific method, I act on the basis of facts rather than speculation. Around the tenth page, I realized that the opponents of modification pursue other goals than communicating the truth. Convinced that I was tired of participating in meaningless polemics, they decided to invite you here and waste your time. They hoped that you would not delve into the essence of the issue, and even more so would not justify a newcomer with his muddy modification.
I apologize for the authorship of the topic, because of which your time was taken.
Sincerely,
User from "that RUS forum"
Last edited by Cooler3D,
what you really mean is one that agrees with you
people have given you plenty of logical and accurate feedback which you simply dismiss as anyone not agreeing with you doesnt know what theyre talking about because you think and i quote "In order to face the notorious degradation, you will have to make Tegra work for a long time at temperatures above ~ 136 degrees Celsius" (the silicon would die by that point but you know w.e)
EDIT: Just so you can know exactly why, by 150C the Silicon crystal structure within the chip begins to shift and change. Before 150C at your 136C mark (well actually before this but just for reference to previous temp claim) the doping materials used to change the characteristics of the silicon for semi conductor purposes begin to shift and become able to move around . This means that the transistors stop being transistors and once this happens the chip is dead forever. You have to build a chip very specifically for it to handle temperature extremes hence why NASA uses what appear to be very old MIPS and similar archs that have a developed process for building them this way. The Tegra X1 would stop being a CPU by the time it reached 136C for any prolonged period, and the traces within the chip will have begun to break down.
you have, repeatedly
you are, and continue to
you have been provided facts, and graphs even. You havent even applied this to what youre doing or your testing on the original repo.
ah yes, the syndicate here looking to bring you down, are devious ulterior motives brought to light by the glorious hero here to save the masses from darkness...
w.e at this point communicating with the OP is literally slamming your junk in a vice repeatedly expecting different results. Not worth the time.
Last edited by Indoo,
I do sort of want more public overclocking information, but I'm disappointed with how you are handling this. The power limitations are extremely important to keep in mind and are very much a real thing. You just seem to ignore them and claim it is fine because it isn't too hot.
And here is the author responding:
https://discord.com/channels/458005814121332757/543451772203565066/927017638750740481
You didn't say you would rewrite the code, you said you would likely stop distributing it: https://discord.com/channels/458005814121332757/543451772203565066/927010368981717113 (RetroNX discord)
https://discord.com/channels/458005814121332757/543451772203565066/927017638750740481
Your sample of 1 erista switch still working after a short time with extreme OC is not science. What people have done to determine the limits of the switch and boards in general is.
If the sigpatches are not included then is the zip you provided compatible with the sigpatches in this thread? https://gbatemp.net/threads/sigpatches-for-atmosphere-hekate-fss0-fusee-package3.571543/
The odds that he would ever reach that point, or anything near it, are slim to none, let’s not be overly dramatic here. 150C’s is “a lot of C’s” and he’s correct in saying that thermal shutdown is likely to occur long before he gets there, or he’ll simply run out of power to drive the chip. I don’t know how many of you are actually involved with extreme OC, but the far more likely scenario is that he will overdrive the power delivery - that’s what will conk out first because we’re working within a very narrow power envelope (as shown on the graphs). Modern silicon is perfectly happy working in elevated temperatures for extended periods of time and nobody needs to demonstrate that this holds true for the Switch because Nintendo already has - there’s no shortage of reports of early Switch units running so hot that the plastic shell partially melted and warped. That’s another scenario far more likely than spontaneous silicon degradation. That’s not to say that the SOC *can’t* fry, it absolutely can, but the notion that it would reach a 150C and stay there for any length of time seems odd to me - realistically it, or something else near it, will give out the magic blue smoke first. That’s not silicon degradation, which is a very slow and gradual process, that’s spontaneous cooking of a component driven out of spec.
Far more appropriate response. The theoretical OC limits of the Switch are mathematically derrived from the power envelope it operates in - it only has so much current to work with and the power consumption is what it is (and scales with temperature, among other things). There’s some silicon lottery wiggle room, but not much - settings that work for some units will not budge on others, or may even damage them - that’s life. The graphs shown in the thread aren’t pulled from anyone’s hat, they’re a realistic look at what the Switch can do - going beyond that would require more power, better cooling or both. That doesn’t stop anyone from trying, even if it’s foolish to do so. If someone wants to run a weird, asymmetric setup wherein the CPU section of the SOC is running full blast and the GPU is choked, coasting on whatever’s left, or vice versa, then that’s their deal and their Switch. What does have to be made clear is that overdriving the chips or the power delivery is not a good idea long-term because those components were not designed to handle this much power draw, not even momentarily, as we’re talking about clocks that go well beyond maximum factory settings. A “VRM” (let’s be generous here, even though the Switch power delivery is very modest) can and will supply what it is rated for at peak, or sometimes can even exceed its rating, but it is neither happy nor can it do that for long - the rating is called “peak” for a reason.
I’d also like to say that this isn’t a “he’s agreeing with him” or “he’s not agreeing with him” thing, or a “who’s in the right” thing, because it isn’t - it’s not a contest.
The problem is more that the Switch's components are made for up to 18W. ZachyCatGames did many tests and it even reached like 40W. That's way too high. The regulators are rated for 5A per phase. The overclock makes it draw over 10, maybe 20A. The problem is also that the OP thinks only heat is what kills the things. That's very wrong, it's current. Always has been. Heat is merely an accelerator, it does have to do with things and degradation, but it's certainly not the only thing.
- Status
