I never claimed it would reach that temperature. He claimed thats the only way things would fail and i was making a point that the chip even reaching that temp would be dead already. We already covered the power delivery aspect previously.
He previously stated essentially that
" Increased voltage does not lead to direct degradation of the element base. Increased temperature leads to degradation."
This is false, and has been proven so on systems as far back as Pentium 4 that i know of personally and as recent as Ryzen 5000 where if you push IF or core voltage too hard you will degrade the chip.
"In order to face the notorious degradation, you will have to make Tegra work for a long time at temperatures above ~ 136 degrees Celsius. Spoiler alert: You will not succeed, because on the way to this figure you will encounter at least three barriers (foolproof)."
This is also not true, the CPU like i said would die before this, not to mention that the switch temperature monitoring is far from "foolproof" and the biggest concern isnt even the SoC but power delivery.
My point was that saying that was incorrect then adding information as to why the CPU running at 136C to begin with doesnt even make sense and not acknowledging anything but temperature as limiting factor is ignorant, misguided, or negligent depending on the mindset.
They claims "I have been an overclocking enthusiast for nineteen years" but show none of the knowledge or concerns associated with anyone who has done it when going as far as to post a modification that is titled as "The simplest way to overclock" . We've already had people in here who dont even know how to check their DRAM type in hekate, they need to be properly informed in the OP of the risk theyre taking. Dozens of places around the internet claim there is absolutely zero risk to overclocking Erista and its just not the case and when you make it simple enough its bound to draw in people far too new.
That’s precisely why I found the notion of “150C fears” kind of silly - that’s not what’s going to kill a given console, overcurrent will, because the components are specifically designed to be driven at certain current limits and while the ratings are generous, they *are* ratings. Again, realistically, “silicon degradation due to heat” is not a concern here, components popping up and down the power distribution due to overcurrent is. It is inherently unsafe to drive any of those components out of spec, or even close to peak for extended periods of time - peak ratings describe momentary, not constant use. That’s neither here nor there though if you make it clear that this is *unsafe* and *probably shouldn’t* be the kinds of settings you use as a daily driver.
There wasnt 150C fears anywhere, i was making a point of the previous 136C claim by OP being silly
Of course you will degrade a chip gradually when you run it out of spec. You will degrade a chip if you run it *in spec*. My point was that it’s not exactly a huge worry here because it’s a gradual, not a spontaneous process, and it will not kill the system. I can think of several things that *will* kill the system *before* you even get there, as I stated before. I’m well-aware of the Ryzen experiment you’re talking about and degradation after the long and arduous stress test was minimal. That’s expected behaviour of all silicon. As a side note, it *is* the temperature that causes degradation of silicon - it messes with silicon doping, but it has nothing to do with the maximum rated temperature. In reality silicon degrades under stress in general, and degrades faster under higher stress, we just happen to know general temperatures at which it can fail spontaneously. There is no magical temperature at which silicon won’t degrade at all (besides maybe absolute zero, when all electron movement stops lol), that’s not what these ratings denote.
You didn’t exactly make that clear, but fair enough, I suppose? I still don’t see how it’s encumbent on anyone to protect any imaginary “new user” from doing things they shouldn’t be doing to their console. None of this is advertised as “safe” anywhere from what I can see - it is listed very clearly as inherently unsafe, which is true.
Im not sure you read any of my previous posts earlier in the thread which would mean that you arent aware of what has been previously stated in regards to this.
As for claims of "imaginary new users" , are we REALLY going to make the claim that everyone encountering this knows EXACTLY whats going on and what the limitations and risks are here? This is incredibly strange that this forum has no interest apparently in informing would be ignorant users and possibly saving them some unnecessary hurt when overclocking dedicated forums are more than happy to inform and explain. a simple "yeah, is not safe its your fault if bad happens" doesnt acknowledge or inform of the risks.
Silicon should not reasonably degrade within spec during its usable lifetime, that would be silly. CPUs will last decades or more operating within spec, RAM runs into other faults besides degradation as do other components
Yes overclocking is inevitably going to do this, I wasnt saying this was the worry, the worry is power delivery as i and others have said previously and the misinformation that only tempurature is a concern. Youre arguing something that isnt being argued, again was making a point regarding OP previous statements of voltage not being a concern and the silly 136C claim.
We are very interested in informing new users of possible ramifications and the danger inherently associated with overclocking - that’s why the OP includes a notice stating as much and why this discussion is allowed to continue. A “new user” can very easily read what everybody had to say.
I can tell you with absolute certainty that every single piece of silicon, even one that has never been used at all (nothing is eternal et all, but that might be a bit too metaphysical for this discussion
Will it degrade over the course of a century? Probably - doping isn’t exactly permanent), will degrade over time - that is exactly why they have designated “lifetimes”. In fact, the operating parameters of electronic components are (partially) derrived from their degradation level over time at given settings. To put it bluntly, the reason why a CPU you purchase in a store has a certain number on it is because the manufacturer can guarantee that, on average, the gross majority of the chips will reach those specific values or exceed them if ran within spec. That does *not* mean that the chip doesn’t degrade over time at stock settings and it is foolish to say otherwise - they absolutely do, and that degradation is in large part priced in. The chip can degrade excessively if overclocked, by which I mean degrade faster than the manufacturer had predicted, but that’s *obvious* considering those are not the settings the manufacturer recommended. Nowadays even those ratings make *no* sense on a box given the fact that CPU’s and GPU’s are, by and large, only power limited and current limited - they will boost as hard as they’re allowed to, often exceeding maximum stock ratings if the host system allows it, and that’s *expected* behaviour, not even overclocking. These matters are, and always have been, very touch and go. Even something like TJmax is calculated differently depending on whether we’re taking about an Intel or an AMD chip because of gratuitous amounts of wanketeering.except all it says in the long short is "Im not responsible if something bad happens" and the rest of the warning is
the OP saying how they assembled these different parts of the mod from different projects. There isnt any relevent information there.
Unused silicon wouldn't degrade any relevant amount within our lifetimes unless put in an environment conducive to it doing so. Everything dies eventually lol, but in this case id bet a CPU in a box stored decently would outlive me.
We arent even disagreeing that it does Im just saying within operating spec the amount is so minimal that its not even worth talking about. Overclocking is a different story, and this is a pretty hefty tangent i should probably put the wrap on.
at least with AMD you have FIT for kinda determining what your specific chips "stock" voltage is. "Stock" and "expected" clocks are way weirder these days than in the past with dynamic boosting of CPU/GPU and means something different for any given system. The days of simple overclocking are seeming to be far behind now.
Last edited by Indoo,
Oh, certainly not within our lifetime, no - I made that correction after the fact to be more specific. My point was that it will, because it has to - it’s part and parcel of how computers are made. Overclocking will absolutely reduce a lifespan of a device, so no, we’re not disagreeing. What I said, from the start, is that even minimal overclocking will reduce the lifespan of a component if it goes beyond the manufacturer-recommended spec. Anything past 1GHz goes beyond the core clock recommended by NVidia themselves on a stock X1 - forget Nintendo for a moment here, they didn’t make the chip. Since this is the case, what’s this argument about? You’re *all* going beyond it, some further than others. The *far* more pressing risk here is pushing 20W+ through a circuit designed to handle what, 18W? On a good day? Close to 40W, even? *That* will kill it *far* sooner than any appreciable level of silicon degradation, let’s be real. That was my point, and it was my impression that you were questioning that. Silicon degradation, even in the extreme, means that you won’t be able to reach peak performance values you used to use as stable over long periods of time. Massive overcurrent means death of the circuit, and possible fire risk. Let’s have our priorities straight here for a moment.except all it says in the long short is "Im not responsible if something bad happens" and the rest of the warning is
the OP saying how they assembled these different parts of the mod from different projects. There isnt any relevent information there.
Unused silicon wouldn't degrade any relevant amount within our lifetimes unless put in an environment conducive to it doing so. Everything dies eventually lol, but in this case id bet a CPU in a box stored decently would outlive me.
We arent even disagreeing that it does Im just saying within operating spec the amount is so minimal that its not even worth talking about. Overclocking is a different story, and this is a pretty hefty tangent i should probably put the wrap on.
Thats exactly what we have been saying this whole time
This has been the concern, you came in when i was trying to make a point to inform the OP that even suggesting the system running at 136C would be whats needed to cause harm is silly. Everyone against the Erista mods has been so due to the fact that it can pull some big boy power numbers when overclocked
Then we are in agreement, somewhat. My question is as follows - who’s responsible for knowing that operating at such parameters may damage the device? In my opinion it is, and always has been, the user pressing the buttons. As long as the user knows he or she is operating well beyond spec, I don’t see a problem. If you believe the thread needs a specific warning that the OP should attach, and have a given phrasing in mind, you’re welcome to suggest that - it’s a discussion. In fact, I started the moderation process in this thread by asking that question. In my opinion the warning is sufficient, but I’m also a man who used to run his Pentium 3 in a tank of mineral oil with a Peltier attached to it, so maybe I have a slightly different perspective on personal responsibility.Thats exactly what we have been saying this whole time
EDIT: Hey, ‘member peltiers? What an unexpected comeback, who would’ve thunk.
I simply would like the warning not in a spoiler box and to make it known the issues of power draw possibly damaging the Switches power delivery when pushing these kinds of overclocks and that it was only designed for roughly 18w. If OP is willing to acknowledge this and doesnt want to write it ill do it if needed. Theyre obviously going to do w.e they want but at the bare minimum i want license compliance and a proper warning.
IDK if id say "comeback" lol
Theyre still absolutely awful for efficiency
Last edited by Indoo,
I would like to see all that, those are credible concerns. Worrying on behalf of theoretical users is not. I’m glad that we arrived somewhere sensible, and would like to see @Cooler3D mention all that in his OP, in a clear fashion. The settings he uses apply to his machine - they may or may not apply to others. Silicon lottery is a cruel mistress and a lot of stars have to align for a circuit to be driven orders of magnitude beyond its capacity without magic smoke. Thing is, I consider magic smoke as a given.
Let's not forget that at the very heart of all of this is the fact that Cooler3D did not develop or design any of this, not a single part, and that all of the developers responsible for the patches and projects being used here have explicitly expressed that this should not be done on Erista systems or at the minimum avoided it.
Cooler3D seems to simply be insistent in putting an all in one noob friendly Erista package together, one that has potential to do great damage. There are any number of people who could have done this but they don't because they understand where the line is between experimentation and sheer stupidity, and they respect the work at hand.
I struggle to see what the motivation is here and can only surmise that the Russian Switch community is less well informed on the dangers of such levels of overclocking, and that Cooler3D enjoyed the praise and credit received from the users there and wanted more of the same from the gbatemp community.
Cooler3D seems to simply be insistent in putting an all in one noob friendly Erista package together, one that has potential to do great damage. There are any number of people who could have done this but they don't because they understand where the line is between experimentation and sheer stupidity, and they respect the work at hand.
I struggle to see what the motivation is here and can only surmise that the Russian Switch community is less well informed on the dangers of such levels of overclocking, and that Cooler3D enjoyed the praise and credit received from the users there and wanted more of the same from the gbatemp community.
Last edited by UnT7oh,
I second this. It's been made too easy. Just drag and drop. Many do not realize it's that dangerous. Aside from that, I'd rather have it on Github where the source is uploaded as source and an all in one package in the releases page, like every other thing for the Switch. To make it download as easy as now, just post a link to either the latest release (github.com/user/repo/releases/latest) or make a direct link like github.com/user/repo/releases/version-x.x/thepackage.zip. Links are just an example. If he makes clear warnings (including about the power draw AND the current) on Github too, then it's fine. Keep in mind that Mariko isn't a holy thing, it already exceeds power draw and current just like Erista. (regulators rated for 5A per phase; 7.5A being drawn with a little over 18W)
imagine being a mod that wants to allow a switch mod that can easily brick switches even with many of the smarter individuals of the community saying its a bad idea. so now basically when it does happen its going to take you down with it for supporting it
Last edited by deathblade200,
KazushiMe stubbed ValidateAcidSignature so yes. But I recommend using the patches from iTotalJustice. Unless Cooler3d compiled the package himself and didn't do it. Patches will behave very weird if he didn't stub it.
I’m not responsible for what people do with their Switches, and neither is the OP. They’re the ones responsible for the modifications they run on their systems, there are no “training wheels” here. If you don’t know how something works or why, you shouldn’t be touching it and you don’t get to cry foul if you fry something. If the OP makes some mods or configs that work for him, however Ill-advised they may be, he’s entitled to share them on the site with others. I’m not anyone’s dad here - if you massively overclock, you’re taking a massive risk. This is no different.
The “peltier inefficiency” myth stems from a misunderstanding of physics. It’s a complicated subject, but this is a handy playlist.IDK if id say "comeback" lol
Theyre still absolutely awful for efficiency
https://youtube.com/playlist?list=PLeC9Tqil0aqxfTjB99wqds3GIDsS7PMFz
Tl;dw define efficiency. High power consumption doesn’t equal inefficiency. They “came back” in a major way, considering Intel’s cryo coolers, manufactured by CoolerMaster and EK respectively, are both peltier-based and tap directly into Intel’s dedicated API to avoid condensation. They’ve never seen a mainstream adoption like this before. I haven’t seen a peltier used in a commercial PC cooling product for years, and I’ve never seen one released with the blessing of one of the big two.
I do not seek praise and do not expect recognition. I just help beginners, as far as possible and free time. I convey to them the work is the result of the work of the best members of the console community, slightly optimized and compiled in an accessible form. The modification contains a number of redundant components, and even this is done in order to maximize the chances of getting a workable result by simply unpacking the contents of the archive onto a memory card.
If I were more irresponsible, I would not have sat on New Year's holidays with the Oxford Dictionary, translating the discussion in this thread. If I were less responsible, I would have ported the modification to Erista in a couple of evenings, rather than spending weeks on optimization, calculations and debugging.
I do not agree with the fact that supposedly all the developers of the original projects were against. In the thread, I see only one of the authors of the code, whose code was significantly rewritten already in the project that formed the basis for my modification.
How one can come to conclusions about my ignorance is also incomprehensible to me. I've been an overclocking enthusiast for about 18 years. During this time, I overclocked a wide variety of chips. I shunted the RTX Quadro 5000 with my own hands on my new Razer Blade Studio (Understanding perfectly well that there are no Razer service centers in Russia). Over the years, not a single device has suffered that I overclocked. I am (most likely) the only person on earth who fit components and an internal cooling system into a compact ATX case, capable of removing two kilowatts of real thermal energy around the clock. A little less than a kilowatt of heat is removed from the processor alone (Threadripper 3990x) in my actual system. Every time I come across the fact that users who are not versed in the laws of physics and who do not know mathematics claim that I am making a mistake and my rig will burn out. However, my assemblies continue to function properly after many years, and that is not fortune. The above system, for example, is almost two years old. Daily use, intermittent loads of 100% across all GPUs and all CPU cores. I clocked my grandfather's Pentium with jumpers. I overclocked the microcontroller and set up a webserver on my wife’s toothbrush. I have root rights in my TV, kettle and refrigerator. My vacuum cleaner runs on Linux compiled from source. I ran GTA San Andreas on my wristwatch many years ago.
Do you still think I do not know what I am doing when overclocking my consoles, and did not informed enough?
Last edited by Cooler3D,
I'm rubbing my eyes in disbelief here.
Did you really just use a vacuum cleaner running linux, an overclocked toothbrush and webserver access to your wife (?!) as your final defence?
I accept defeat.
Last edited by UnT7oh,
- Status
