Homebrew Homebrew app Switch-OC-Suite

[binkinator]

The allow unsafe frequencies is not to unlock full OC support under battery, it's just to go beyond the very known safe power delivery OC for each Switch on Erista is to go above 1785mhz in the CPU for example...

Sadly they had that in the tesla overlay, but on latest updates they removed it, now it has to be manually added if you want it by doing the following:

going into: /config/sys-clk/config.ini and adding this to like the bottom:

[values]
allow_unsafe_freq=1

(in their FAQ they say to only add the allow_unsafe... but it doesn't work without the "[values]" specification).

ps: Hope they bring back the option to select "allow unsafe frequencies" in the Tesla overlay like before...

I get it though...all we need is a bunch of n00bs blowing themselves up on prime-time MSNBC. It will be the next vape pen epidemic!
*politician clutches pearls*
If banning the Nintendo Switch can save just one child's life...it's worth it!

Maybe this is an opportunity for Konami code?

 

[ChanseyIsTheBest]

I get it though...all we need is a bunch of n00bs blowing themselves up on prime-time MSNBC. It will be the next vape pen epidemic!
*politician clutches pearls*
If banning the Nintendo Switch can save just one child's life...it's worth it!

Maybe this is an opportunity for Konami code?

View attachment 352354

Ngl I seriously want to make a suggestion on Kazu's github now to have an option to unlock the unsafe frequencies by entering the Konami code in sys-clk. If you're reading this thread Kazushime please do it lol.

 

[Anxiety_timmy]

The allow unsafe frequencies is not to unlock full OC support under battery, it's just to go beyond the very known safe power delivery OC for each Switch on Erista is to go above 1785mhz in the CPU for example...

Yes and no. 2091 by itself isn't enough to actually damage anything since its under 18W, and Erista regulators can handle around 2.295. The problem is that basically anything that is GPU intensive will push it over, Which reminds me I need to test what gpu frequency stops drawing above 18W with 2091.

Post automatically merged: Feb 9, 2023

Ngl I seriously want to make a suggestion on Kazu's github now to have an option to unlock the unsafe frequencies by entering the Konami code in sys-clk. If you're reading this thread Kazushime please do it lol.

Konami code unlock that's amazing

 

[guily6669]

I get it though...all we need is a bunch of n00bs blowing themselves up on prime-time MSNBC. It will be the next vape pen epidemic!
*politician clutches pearls*
If banning the Nintendo Switch can save just one child's life...it's worth it!

Maybe this is an opportunity for Konami code?

View attachment 352354

Well they always had the option there and it had a clear warning on it if you wanted to activate it and by default was OFF...

normal sys-clk without OC suite also had that.

ps: I prefer the option there my self cause even though I don't use them, when testing I have to edit the ".txt" file just to unlock it and before could be unlocked under miscellaneous option of the sys-clk overlay on-the-fly.

Yes and no. 2091 by itself isn't enough to actually damage anything since its under 18W, and Erista regulators can handle around 2.295. The problem is that basically anything that is GPU intensive will push it over, Which reminds me I need to test what gpu frequency stops drawing above 18W with 2091.

Post automatically merged: Feb 9, 2023

Konami code unlock that's amazing

Well still the safe clocks by OC Suite is 1785mhz for Erista CPU and I'm guessing that's into consideration of having both GPU + CPU at their <<max safe frequencies>> and probably considering already 100% use in both all CPU gaming cores + GPU...

I'm not a electronics engineer so I'm not even contest what can be considerable safe or unsafe or even if it's 18W, less or higher...

 

[swutch]

a few benchmarks:
Cooler3D 2400 MHz kip & modified hekate | EMC Vddq = 650mV | EMC Vdd2 = 1250mV
(I've lowered the EMC Vddq Voltage from 700mV to 650 mV)

KazushiMe kip with the following settings:

2131 MHz
NO_ADJ_ALL
EMC Vddq = 600mV | Vdd2 = 1125mV

 

[Cooler3D]

(I've lowered the EMC Vddq Voltage from 700mV to 650 mV)

a few benchmarks:

I did not expect to see here tests of an experimental, non-public fix to 4IFIR

Such fixes, technically not intended for normal use, their task is to confirm or refute conceptual hypotheses.

In this fix, for example, all optimizations are deactivated - not directly related to the frequency of the memory, for the purity of the experiment. proof of concept of 2400MHz frequency mode stability - on all memory models, including Hynix and low rated models.
For setting records 4IFIXes will not a best choice. Hovewer, the next public update of 4IFIR itself is ideal for this. I advise you to compare OS Suite with comming 4IFIR update. It will be interesting to compare them head on.

Fun fact about this fix:
The 2400 MHz used in the fix not only does not require memory overvolting, it is designed for memory undervolting, 4IFIR ECO Mode. The value of 700mV was in the fix by accident, lol. Neither 700 nor 650mV are required for stable operation of the mode. It can even function like this:

 

[swutch]

I did not expect to see here tests of an experimental, non-public fix to 4IFIR

Such fixes, technically not intended for normal use, their task is to confirm or refute conceptual hypotheses.

In this fix, for example, all optimizations are deactivated - not directly related to the frequency of the memory, for the purity of the experiment. proof of concept of 2400MHz frequency mode stability - on all memory models, including Hynix and low rated models.
For setting records 4IFIXes will not a best choice. Hovewer, the next public update of 4IFIR itself is ideal for this. I advise you to compare OS Suite with comming 4IFIR update. It will be interesting to compare them head on.

Fun fact about this fix:
The 2400 MHz used in the fix not only does not require memory overvolting, it is designed for memory undervolting, 4IFIR ECO Mode. The value of 700mV was in the fix by accident, lol. Neither 700 nor 650mV are required for stable operation of the mode. It can even function like this:
View attachment 352376

Yes I thought it would be interesting to compare the Switch-OC-Suite to 4IFIR.
I followed your telegram link and those files were the last ones you have uploaded there.
It's impressive to see 2400 MHz is bootable.
But performance is more or less the same as 2131 MHz,
and it is not stable with or without OV.

So with a Samsung 4266 Mbps Ram i don't see any advantages.
Maybe it is a different story with a 3733 Mbps Ram or your final Version.

 

[Cooler3D]

and it is not stable with or without OV

This issue is known and has already been fixed. Fortunately, it is not connected with overclocking of memory.
Now that the concept of the performance of such memory frequencies during undervolting has been proven, this feature will be implemented in the main branch. Let's see what results we can get in combination with other 4IFIR optimizations.

Spoiler: Comparative Tests

OС Suite in the latest versions, better than ever. There is no better opponent for 4IFIR.
Since 4IFIR is positioned as the most advanced overclocking solution, I propose to consider any result other than a clear superiority as a defeat for 4IFIR.

I will run and post a variety of tests, and try to be open-minded.
Open to any criticism, wishes and questions about the testing methodology - from users with a rating, tackable activity, which has never been a member of the 4IFIR and RetroNX community.

Attempts of provocations, trolling and the like will be ignored. I reserve the right to challenge any of my conclusions through a constructive discussion, with participants comparable or superior in my competence.

Tests will be posted gradually, as they are created and free time.

I reserve the right to veto any of my findings to KazushiMe.
In case of disputes, I leave the last word to KazushiMe.

If my conditions are fair, and someone is really interested to see what this will lead to, mark this post as a reaction, so I will understand that it is worth it to invest my time to this.

 

[guily6669]

At that low voltages in such clocks my Switch would have a good performance penalty at least with OC suite, don't know with your 4IFIR mod...

Also I don't think it would handle such memory frequency with such low voltage either, but I never tried.

ps: Couldn't InfoNX be lying on the voltages there???


---------------------------------------------------------------------------------------------------------------------------

@ChanseyIsTheBest I just saw that KazushiMe replied to the request you made for me on the github:

Setting CPU/GPU voltage coefficients for every frequency is possible in their dvfs tables.
Dynamic voltage-frequency scaling (DVFS) for DRAM does not exist in HOS, so the 3rd one cannot be implemented.
As for the 4th, there are an enormous amount of timing parameters to consider. Only the table for maximum DRAM frequency will be exposed and customizable.
All these require rebooting to be applied. Adjusting on the fly is not feasible.

How is that done? I wanted is to be able to do it on OC SUITE configuration webpage...

But in real-time and being able to change and save voltages would really be the cherry on top of the cake if possible but I would really be more than happy already if he could add on the online configurator...

 

[ucupetuks]

a few benchmarks:
Cooler3D 2400 MHz kip & modified hekate | EMC Vddq = 650mV | EMC Vdd2 = 1250mV
(I've lowered the EMC Vddq Voltage from 700mV to 650 mV)

View attachment 352362View attachment 352363View attachment 352364


KazushiMe kip with the following settings:

2131 MHz
NO_ADJ_ALL
EMC Vddq = 600mV | Vdd2 = 1125mV

View attachment 352366
View attachment 352367
View attachment 352368

try test 921 or 998, you would see different performance. but it wont stable for switches, to risky for small gain

 

[ChanseyIsTheBest]

a few benchmarks:
Cooler3D 2400 MHz kip & modified hekate | EMC Vddq = 650mV | EMC Vdd2 = 1250mV
(I've lowered the EMC Vddq Voltage from 700mV to 650 mV)

Thank you for showing how just because the RAM frequency is higher it doesn't necessarily translate into better performance especially with the extra voltage you have to put towards it. I'll add these to the main post as well.

 

[Cooler3D]

I knew you guys would get to the bottom of 700 millivolts, so I put it there on purpose. I imagined your faces when you suddenly realized under what voltage my proof of concept of 240 was actually designed. Check it out:

There are no tricks here, they are stable, and work at a voltage that is not enough for the stock 1600 MHz.
As soon as I combine this with other optimizations, the effective performance will increase noticeably.

Obviously, if the solution were as simple as setting the voltages in three places, you would not be standing in front of an insurmountable barrier at 2131 MHz, any attempt to overcome which fails, regardless of the voltage used (if I did not miss anything). If you get tired of guessing on coffee grounds, you can ask me directly about how I did it. I'm in a good mood, and if you ask politely, it will probably succeed.

Returning to the topic of a comprehensive comparison of OS Suite with 4IFIR. Worth doing? If no one is interested, I have no other tasks left in this thread, and I will go about my business.

Personally, I would be most interested in whether 4IFIR is really as good as it is commonly believed, or in fact - we will see parity between forks. This would mean that 4IFIR no longer makes any sense.

 

[ChanseyIsTheBest]

I knew you guys would get to the bottom of 700 millivolts, so I put it there on purpose. I imagined your faces when you suddenly realized under what voltage my proof of concept of 240 was actually designed. Check it out:

View attachment 352472

You're only on the home screen menu because if you try doing anything like booting a game it will crash and is super unstable. It's not surprising that you got worse results on the benchmark because you took and used Kazu's code he got rid of. I wonder why he got rid of that code? Probably because it wasn't very good.

 

[Cooler3D]

You're only on the home screen menu because if you try doing anything like booting a game it will crash and is super unstable. It's not surprising that you got worse results on the benchmark because you took and used Kazu's code he got rid of. I wonder why he got rid of that code? Probably because it wasn't very good.

Do you really think that I came here with such an obvious deceit, to risk my reputation? If this is a deception, then it is enough to ask me for a proof, and the illusion will dissipate. Ask.

In the meantime, I propose to agree that the indicators in the screenshot are so impressive that in your opinion they cannot be true, and there is probably a catch hidden somewhere. Did you mean it?
Ask for proof...

 

[ChanseyIsTheBest]

Do you really think that I came here with such an obvious deceit, to risk my reputation? If this is a deception, then it is enough to ask me for a proof, and the illusion will dissipate. Ask.

In the meantime, I propose to agree that the indicators in the screenshot are so impressive that in your opinion they cannot be true, and there is probably a catch hidden somewhere. Did you mean it?
Ask for proof...

Post it. And even then if it works it only works on a few units that have miracle RAM at that voltage. And I seriously doubt how stable that might be. It could boot into a game and then later crash but you wouldn't show that. I could undervolt shit slightly on my Switch as well but it would only work for my unit because every unit is different. It doesn't mean much. Still doesn't change the fact that your 2400Mhz was getting lower performance then 2133Mhz.

 

[Cooler3D]

Post it. And even then if it works it only works on a few units that have miracle RAM at that voltage. And I seriously doubt how stable that might be. I could undervolt shit on my Switch as well but it would only work for my unit because every unit is different. It doesn't mean much.

I have the most ordinary, average device, the same master sample on which I test 4IFIR. 3733 Mbps (Rated) SAMSUNG K4U6E3S4AM-MGCJ nothing special.
My method is stable on any memory chips, including Hoonix. If what I show raises doubts, it is enough to ask me to provide a patch, and make sure it works for yourself.

And since you already asked for proofs, give me a minute, I'll shoot it now.

 

[Keylogger]

How can I adjust settings directly on switch and depending of games like sys-clk manager ?

 

[ChanseyIsTheBest]

How can I adjust settings directly on switch and depending of games like sys-clk manager ?

1. Boot into game and then open up Tesla menu using L+D-Pad Down+Right Stick Click.
2. Choose sys-clk OC from the Tesla menu
3. Ensure it is enabled (Enabled=on), and then click edit app profile
4. Change the clocks you want in edit app profile

There's also a sys-clk manager that you can edit games included which you can launch using title override (Hold R while booting game)

https://github.com/KazushiMe/Switch-OC-Suite/releases/tag/1.4.1

 

[Cooler3D]

Post it. And even then if it works it only works on a few units that have miracle RAM at that voltage. And I seriously doubt how stable that might be. It could boot into a game and then later crash but you wouldn't show that. I could undervolt shit slightly on my Switch as well but it would only work for my unit because every unit is different. It doesn't mean much. Still doesn't change the fact that your 2400Mhz was getting lower performance then 2133Mhz.

Done. Sorry for keeping you waiting. I hope this is what you expected:

VHQ Video | VHQ Mirror (watch or doianload) | YouTube Video

I made this video, taking into account all your wishes. It was recorded in one 27-minute take without gluing, on it you can see at least every pixel of each frame. I was not too lazy to use a tripod and a camera.
Obviously, it is not mounted, there are no tricks in it, it is extremely unsophisticated. This video is a testament to the fact that I somehow managed to do what no one else could before (at least on Switch). You, CTC, PSam, even KazushiMe, whoever. None of you not only did not understand how to achieve a similar result, it seems impossible to you, and I know why:
Stable 2400MHz on Nintendo Switch on undervolted volages? - Soind like a joke.

Now you are faced with a choice of two:
A. You can go on picking on minor details, looking for the most improbable arguments - designed to devalue my work, looking for fraud and extremely unlikely explanations for how it happened.
In that case, I'll just end the pointless dialogue and move on.

But there is another option:
B. The explanation for the phenomenon lies on the surface, it is extremely unpleasant to agree with it. According to this explanation, I was able to do (not for the first time) what others could not, because I understand this better than other Switch scene developers.
It is unpleasant to agree with this, I agree, but this option is accompanied by a bonus. You can ask me to teach you what I know. Take for granted my expert, and ask to explain to show how this is even possible. That way, you'll find out almost instantly, was the decision correct?.

I've wasted some of my time trying to get you to the right conclusions. If I didn’t succeed, it’s okay, you don’t owe me anything.

I'll go to import the innovation into the mainline 4IFIR. I wonder what results will be achieved in tandem with other optimizations, like timing adaptation, etc.

Some of the conclusions that I made during the successful "proof of the viability of the hypothesis":

• The values obtained are far from the limit of what can be obtained for free, remaining even within the limits of the standard voltage, not to mention a slight overvolting.
• It is able to function perfectly stable.
• MEMORY MATTERS! It still remains the bottleneck of the system. It is not obvious to everyone, but the increase in memory bandwidth allows you to get the same performance as before, at lower GPU / CPU frequencies, and as a result - less power consumption.
• It seems that I will be able to reproduce the method on ERISTA samples as well.

P.s. If the first option is chosen, for God's sake, don't feel bad about reverse-engineering my waste. Just ask me how it's done and I'll explain, saving you time.

 

[ChanseyIsTheBest]

Done. Sorry for keeping you waiting. I hope this is what you expected:

VHQ Video | VHQ Mirror (watch or doianload) | YouTube Video

For one you blatantly lied - that is not 500mv that is 550mv while the image a couple posts above that you previewed is 500mv.

I knew you guys would get to the bottom of 700 millivolts, so I put it there on purpose. I imagined your faces when you suddenly realized under what voltage my proof of concept of 240 was actually designed. Check it out:

View attachment 352472

Secondly your results on tinymem bench are worse then swutch's at 2133 meaning that your 2400 performs worse then Kazu's. The numbers don't lie, why would someone use yours when Kazu's runs at a lower frequency which means it's more stable and gets better performance. Who cares that 2400 is higher then 2133 when it runs worse.
From your YouTube video

Swutch's 2133

Thirdly, I don't care about you playing some random games and checking the memory for less than half an hour. It can take gameplay for hours on end for corruption to start happening.

Fourthly, there's no reason to undervolt RAM past the default. RAM barely draws any power and undervolting it increases instability.

Why the hell would some use something that is closed source, unstable, and performs worse then something publicly available. There's a reason why Kazushime never had to advertise his project - its quality speaks for itself. I made this thread because I wanted to spread good information around.

It is absolutely disgusting how you shit on the people who originally made sys-clk - the people who know what they are doing ZachyCatGames, CTCaer, m4x. It's appalling how you tried to pawn sys-dvr as your own with no credit to the original author. It's gross how you barely ever acknowledge your sys-clk is a poorly edited Switch-OC Suite by KazushiMe. It's beyond indefensible how you do not advice users to make an immediate backup of their EmuNAND and SD card when the chance of corruption is very real and likely.

And ultimately all for what? A script-kiddie mod of Switch-OC that runs worse. The numbers don't lie.

Next time you post on my thread about your precious cobbled together 4IFIR, don't expect that post to stay up for very long at all. Why don't you go back to your 4IFIR thread - I wonder why the mods haven't opened it yet?