Gaming Homebrew Review 4IFIR / OC Suite - Showcases (performance / power consumption)

[Foxi4]

Spoiler

It is not true. You said yourself that you expected to see 360p with a slightly more stable frame rate than 20 fps. Or a slightly higher resolution at the same frame rate of 20 fps.

Rendering resolution is locked at 720p only on the first two screenshots, on both 12 fps (and in fact - lower, but let it be 12). If 12 is multiplied by 3, we get not 36 fps, but not 60 fps.

All subsequent videos and screenshots (captured on stock) were shot in 360p mode. The video shows that 360p in stock drops to 20 fps.

On 4IFIR-e 720p, 60 frames are rendered without drawdowns. And instead of 15 watts, which would be logical for such frequencies, it is fit into 7 watts (against 5 watts in stock).

In "+xX"-es - even according to the calculation scheme you proposed, it turns out significantly more than I stated.

Instead of denying the obvious, you can simply ask how I managed to implement this, and I will gladly explain. I wanted to tell you for a long time, but you did not believe that it was possible in principle. Today you made sure that I was honest from the very beginning.

Do not be stubborn, and we can assume that we have settled past misunderstandings, thereby you will return to the forum a unique, unprecedented exclusive, this time with repositories, etc. Everyone tends to err. Do not go on principle to the detriment of the forum.

Based on nothing. I told you that from the start - I need to see how the game runs at stock in order to make any worthwhile prediction about how it’ll run when you OC. How can I make an accurate one without any information? You asked me to guess, with zero numbers. Your screenshots align with my expectations - for starters, you tested in handheld, which gimps the GPU by half at stock from the get go - of course it ran like crap. You’d probably get decent performance at 30-ish when running the GPU at 768MHz, which (IIRC) is the max it can push when docked. You’ve near enough doubled it, so you got near enough double the frames. It’s not *quite* linear, but the before-after screenshots align with what I’d expect.

For the record, I’m not saying this is a bad result. This is a great result, and since you locked down the testing environment, people have a better idea of what to expect at what clock.

 

[P_H_I]

idk :d its is 3 am for me but no it is not a joke ; ) but this is still overclocking/modding so keep this in mind and how dare you compare yogurt to kefir dude... outrageous!

ay man as someone who had horrid GI issues for seven months, probiotic yogurt is yogurt all the same.

But yeah I'm definitely a fan of overclocking on the Switch so I'm interested to see exactly how this differs performance wise. Would be interested to see how titles like Xenoblade 2 stack up against it.

 

[Lamcza]

ay man as someone who had horrid GI issues for seven months, probiotic yogurt is yogurt all the same.

But yeah I'm definitely a fan of overclocking on the Switch so I'm interested to see exactly how this differs performance wise. Would be interested to see how titles like Xenoblade 2 stack up against it.

but @Foxi4 posted the differences, kefir is not a yogurt :/ pls keep more focus on overclocking at least just for safety of your console

 

[Cooler3D]

which (IIRC) is the max it can push when docked

I said that the dock mode adds absolutely nothing, except for the load, in my case...
I literally performed a miracle in front of your eyes, but you came up with an explanation that was more convenient for you, ignoring a number of facts.
Okay, I give up.

 

[P_H_I]

Ok so from what I'm gathering is this is a magic piece of overclocking software that is much more efficient in portable mode due to it drawing less power than other software. Why does it draw less power?

 

[Lamcza]

Ok so from what I'm gathering is this is a magic piece of overclocking software that is much more efficient in portable mode due to it drawing less power than other software. Why does it draw less power?

undervolting i guess and some magick sh***. Hope you're feeling better now regardless of the health issues you've been struggling with na zdrowie

 

[Foxi4]

I said that the dock mode adds absolutely nothing, except for the load, in my case...
Okay, I give up.

That’s what I’m trying to explain to you. Handheld mode imposes a hard limit on maximum clocks at stock - this is to protect power delivery and keep thermals in check. When you trick the Switch into thinking it’s docked, that limit is raised, but it’s up to the developer to use that headroom - they might not want to, for instance to maintain the same experience when docked and undocked. Overclocking raises that limit again, and on top of that, it allows you to set an override so that the machine runs at a higher clock than the game would otherwise demand. This is all theoretical until you see the actual figures the game is running at. Again, the performance increase is very clear, all I’m saying is that it’s in line with what I expect based on where you started and where you ended up.

 

[Cooler3D]

Ok so from what I'm gathering is this is a magic piece of overclocking software that is much more efficient in portable mode due to it drawing less power than other software. Why does it draw less power?

It draws much less power because bunch of low level optimizations, including deep undervolting of each possible component, and Atmosphere software tweaks.
It gives much more performance at the same CPU/GPU freqs because bunch of low level optimizations, including deep memory frequency and timings optimizations, which is the bottleneck of Switch.
Thus, at maximum frequencies, 4IFIR is twice as energy efficient, and twice as faster vs OC Suite. I'd rather not say anything about SysCLK. Comparing 4IFIR with SysCLK is like getting a baby out of a stroller to beat him up.
This is a short, and very incomplete answer. But he gives a general idea.
By the way, about OC Suite. For some time now, I am the only contributor to the project. So you still don't have a choice.

 

[Lamcza]

It draws much less power because bunch of low level optimizations, including deep undervolting of each possible component, and Atmosphere software tweaks.
It gives much more performance at the same CPU/GPU freqs because bunch of low level optimizations, including deep memory frequency and timings optimizations, which is the bottleneck of Switch.
Thus, at maximum frequencies, 4IFIR is twice as energy efficient, and twice as faster vs OC Suite. I'd rather not say anything about SysCLK.
This is a short, and very incomplete answer. But he gives a general idea.
By the way, about OC Suite. For some time now, I am the only contributor to the project. So you still don't have a choice.

oyy Cooler it is strange for me to ask as your first main alt account but can you make some only English channel/subchannel on telegram xD?

 

[Cooler3D]

That’s what I’m trying to explain to you. Handheld mode imposes a hard limit on maximum clocks at stock - this is to protect power delivery and keep thermals in check. When you trick the Switch into thinking it’s docked, that limit is raised, but it’s up to the developer to use that headroom - they might not want to, for instance to maintain the same experience when docked and undocked. Overclocking raises that limit again, and on top of that, it allows you to set an override so that the machine runs at a higher clock than the game would otherwise demand. This is all theoretical until you see the actual figures the game is running at. Again, the performance increase is very clear, all I’m saying is that it’s in line with what I expect based on where you started and where you ended up.

On 4IFIR, this does not happen. The console capabilities on 4IFIR are 100% identical, whether I emulate dock mode or stay in portable mode. 4IFIR comes with a custom Atmosphere and a custom bootloader for a reason. I can make hardware do literally anything, since the system is available to me at all levels.

In one of the upcoming updates, I will make the built-in notification system and the news application work, only instead of the Nintendo backend, the consoles on 4IFIR will access my server. I can literally get the built-in app store to work with Tinfoil's repositories. I don't do this for ethical reasons.

oyy Cooler it is strange for me to ask as your first main alt account but can you make some only English channel/subchannel on telegram xD?

It is multilingual. Write in English, I and most of the guys will catch you and answer.
Or just check ENG version of dev diary, here the lin... oh, you're a cunning ass, you go baited me for a ban. Links to 4IFIR are prohibited on this forum.
The third time I won't buy it.

 

[Foxi4]

On 4IFIR, this does not happen. The console capabilities on 4IFIR are 100% identical, whether I emulate dock mode or stay in portable mode. 4IFIR comes with a custom Atmosphere and a custom bootloader for a reason. I can make hardware do literally anything, since the system is available to me at all levels.

In one of the upcoming updates, I will make the built-in notification system and the news application work, only instead of the Nintendo backend, the consoles on 4IFIR will access my server. I can literally get the built-in app store to work with Tinfoil's repositories. I don't do this for ethical reasons.

You’re not good at taking compliments - I just told you it’s a good improvement in performance. You have to compare it to stock with the values disclosed though, otherwise measuring the effect is difficult. For instance, if you want to measure FPS increase, you must necessarily force the same resolution, that’s the only way to get a good measurement. You just must a common denominator, that’s why I insisted on it, regardless of how poorly it’ll run without an overclock.

 

[Lamcza]

On 4IFIR, this does not happen. The console capabilities on 4IFIR are 100% identical, whether I emulate dock mode or stay in portable mode. 4IFIR comes with a custom Atmosphere and a custom bootloader for a reason. I can make hardware do literally anything, since the system is available to me at all levels.

Post automatically merged: Apr 3, 2023

It is multilingual. Write in English, I and most of the guys will catch you and answer.
Or just check ENG version of dev diary, here the lin... oh, you're a cunning ass, you go baited me for a ban. Links to 4IFIR are prohibited on this forum.
The third time I won't buy it.

What? dude no xD and boys will still write that I am your alt LOL for the record Cooler doesn't like me and doesn't trust me
I just want a place where i can understand/read something know that i can ask in there but still, an English sub will be nice ; )

BTW link to your telegram is still there in your main closed thread soo if it is forbidden heh nvm whaat? whaat? Who said that@!?

 

[Cooler3D]

You’re not good at taking compliments - I just told you it’s a good improvement in performance. You have to compare it to stock with the values disclosed though, otherwise measuring the effect is difficult. For instance, if you want to measure FPS increase, you must necessarily force the same resolution, that’s the only way to get a good measurement. You just have a common denominator.

I made a game rendered in 360p stock with up to 20 fps drawdown, render in 720p at 60fps with no drawdown, at the cost of only 2 watts added to 5 watts of stock. You said that this is not surprising, and you expected to see it.
By how many times did I increase the output per watt?
How many times did I increase the total performance?

Calculate in any way convenient for you, just give me a approx. number, and I will finally decide if I should stay here, or I shouldn't have left the Russian-speaking forum from the bottom internet

p.s. Thanks for the compliment. I appreciate that at least someone at the GBA said a kind word to me.

link to your telegram is still there in your main closed thread

I deleted the post, where did you find the link?

 

[Lamcza]

I deleted the post, where did you find the link?

no, no, no man iam from Poland we dont do that in here i will not sell you even if you ask for it i mean not for free at least

Spoiler: maybe

maybe for an english channel on Telegram ; )

 

[Foxi4]

I made a game rendered in 360p stock with up to 20 fps drawdown, render in 720p at 60fps with no drawdown, at the cost of only 2 watts added to 5 watts of stock.
By how many times did I increase the output per watt?
How many times did I increase the total performance?

Calculate in any way convenient for you, just give me a approx. number, and I will finally decide if I should stay here, or I shouldn't have left the Russian-speaking forum from the bottom internet

I don’t know why you insist on calculating it in this way, or why you’re mixing in power efficiency. We’ve had this exact conversation before.

If you want to measure compute performance change, just run the game at *fixed resolution* without any overclock as it would run on a non-modified Switch, then run it with your OC and check how much your framerate improved. Unless the resolution is fixed, you *can’t* measure performance easily because you’re comparing two renders that are different - they have to be the same.

If you want to measure performance per watt, run the game in *the exact same conditions* on both and check how your consumption changed.

Those are two different measurements that you should keep separate. Do you not have access to a regular installation of Atmo with no mods besides the monitor? It seems like something that should be easily achievable. Additionally, your problem seems to be that you’ve fundamentally changed how the clocks are managed in 4FIR, so it’s de facto not an accurate comparison. You should be comparing 4FIR to non-4FIR. I say this because you’ve told me your Switch “doesn’t do” what the average Switch is supposed to do depending on Docked/Undocked status.

 

[Cooler3D]

Do you not have access to a regular installation of Atmo with no mods besides the monitor?

Do you doubt the accuracy of the telemetry in my screenshots?

no, no, no man iam from Poland we dont do that in here i will not sell you even if you ask for it i mean not for free at least ^^

Okay, looks like I'm ripe for this self-humiliating act. Listen, since you're my multi-account... Have you been banned from the RetroNX discord yet? I've been banned there for a year and a half. Ask CtCaer if there is a place for me on their Discord? "4IFIR and OC Suite - RetroNX exclusive", he could not cancel them, but he will be able to lead, which is much better. I even recently started repositories for both projects. He still won't find a better offer.

I will, together with him and his secarii, cancel other promising projects. I'm tired of being decent, no one believes me at the GBA anyway.

 

[hippy dave]

More claims of repositories, still no links, shocker.

 

[Foxi4]

Do you doubt the accuracy of the telemetry in my screenshots?

I’m taking them at face value and telling you how to measure the difference between stock and non-stock in a semi-accurate fashion. Between the two of us, you don’t trust the monitor more so than me, since you literally told me that “in reality the framerate was lower” than displayed. I doubt that - I think the framerate was likely exactly what the monitor said, but the human brain is very sensitive to framerate fluctuation and wild swings between 12 and 30 will appear as very noticeable stutter, almost a slide show.

In any case, your project is interesting enough. I hope you’ll put it on GIT just like your fork of Switch OC Suite, so that it can be linked and examined without licensing concerns. That should be your next step, at least in my opinion.

More claims of repositories, still no links, shocker.

Cooler now has a repository at:

https://github.com/Cooler3D/Switch-OC-Suite

In fact, he’s had this account for quite some time, so I don’t know why his other projects aren’t on the same account. They should be. Once 4FIR is properly documented, we can have a closer look at it and draw our own conclusions. Until then we’ll have silly arguments about numbers.

 

[hippy dave]

Cooler now has a repository at:

https://github.com/Cooler3D/Switch-OC-Suite

In fact, he’s had this account for quite some time, so I don’t know why his other projects aren’t on the same account. They should be.

Maybe he's lying.

 

[Cooler3D]

In fact, he’s had this account for quite some time, so I don’t know why his other projects aren’t on the same account. They should be.

Translation, from RU dev diary:

QUESTION:
Why doesn't 4IFIR have a public code repository?

ANSWER:
Shortly after 4IFIR became available to regular users, the project's repositories were canceled by members of the largest Switch CFW developer community. The most vocal developers did everything in their power to prevent the project from appearing and spreading. They openly stated that 4IFIR's capabilities should not be available to every random fool without deep skills in assembly and overclocking. I listened to them.

Since then, the project's source code has only been available in my cloud without documentation, pointers, or comments. If developers need it, they'll have to figure it out themselves.

By the way, 4IFIR has become more accessible to regular users since then.

However, this is no longer relevant, since I started the repository on your advice. Now the question is, where should I return if I feel like an outcast at the GBA. You guys were even ready to give the unfortunate thread of OC Suite to anyone, but decided to refuse me. Trust level - Kuller3D.