Not for me I think that was on a older OC suite, I'm using latest 1.4.0.1 and miscellaneous doesn't show voltage but I saw them on a screenshot from a older version of it for sure...
Homebrew Homebrew app Switch-OC-Suite
- Thread starter ChanseyIsTheBest
- Start date
- Views 82,386
- Replies 300
- Likes 13
- Status
For some reason Kazushime commented it out for v1 Switches only allowing v2 Switches to use that feature apparently. I'm not sure why. Maybe something doesn't work properly with it but I don't know.
Post automatically merged:
Do you get higher benchmarks in tinymembench with the overvolt compared to without it on that frequency you changed? If you don't I'd properly revert the change because it might be giving you little benefit. If you don't I'd properly revert the change because it might be giving you little benefit. You shouldn't have to manually change the overvolt values as its the one Nintendo sets for 1785 CPU because its stable on all Switches and used officially in boost mode is good.
Last edited by ChanseyIsTheBest,
Oh damn didn't know it was disabled for Erista, I thought it was removed from latest versions
, for now then I recommend InfoNX it's working good until he can add it to Erista on OC suite.About the voltage I think the score was about the same with 1.1V Vs default, I took photos with my phone and was very similar, I will leave at 1.1V for now since I finished 3 games and it's working fine but it's not much different than default because the max voltage from default really only is maxed out on the highest CPU clocks from "allow unsafe frequencies" which I don't use or care anyway...
This is from a previous post of mine on other thread (loader.kip dfault vs 1100\1.1V on Erista)
Can you please request the following on Github??
1- CPU voltage per each frequency on their online configurator for Erista\Mariko;
2- GPU voltage per each frequency on their online configurator for Erista\Mariko;
3- Mem voltage per each frequency on their online configurator for Erista\Mariko;
4- Mem timings per each frequency on their online configurator for Erista\Mariko.
Extra- Make KazushiMe to create an account here at GBA if he doesn't already
EDIT: Added some color...
Last edited by guily6669,
I'm not sure what the point of overvolting if the frequency is already stable because it shouldn't give you any extra performance but instead give you extra heat and power use. I don't get why some things are undervolted at random either. Overvolting should only be for if the frequency isn't stable already. I don't understand why someone would make this new loader.kip if they are distributing it to everyone. I would do multiple 3 thread benchmarks using tinymembench with and without the overvolting to see if there is actually any benefit (which I doubt)Oh damn didn't know it was disabled for Erista, I thought it was removed from latest versions
About the voltage I think the score was about the same with 1.1V Vs default, I took photos with my phone and was very similar, I will leave at 1.1V for now since I finished 3 games and it's working fine but it's not much different than default because the max voltage from default really only is maxed out on the highest CPU clocks from "allow unsafe frequencies" which I don't use or care anyway...
This is from a previous post of mine on other thread (loader.kip dfault vs 1100\1.1V)
Can you please request the following on Github??
1- CPU voltage per each frequency on their online configurator for Erista\Mariko;
2- GPU voltage per each frequency on their online configurator for Erista\Mariko;
3- Mem voltage per each frequency on their online configurator for Erista\Mariko;
4- Mem timings per each frequency on their online configurator for Erista\Mariko.
Extra- Make KazushiMe to create an account here at GBA if he doesn't already
Each unit is unique when it comes to undervolting and if you undervolt incorrectly status monitor will 'lie' about the clockspeed and you'll get worse performance. That's why 2397 CPU on Switch-OC-Suite gives worse performance then 2295 CPU because the voltage is too low. I would individually test each of these underclocked frequencies with a benchmark tool like tinymem compared to it without the undervolt and be aware that undervolting is unique to each unit because of binnings.
I can put in your request on Github because I have an account there.
I'm not overvolting, but rather undervolting from the default to 1.1 but like I said on memtest at least at 1785mhz the score is very similar between default and 1.1v but the difference is around 31.2mv at that frequency.
I understand the OC being different for each console I'm used to that on PC like going from my nice I7 2600K at 4.9ghz below 1.4v with 0 voltage compensation needed to a damn nasty upgrade (downgrade) I7 3770K that I could only do 4.5ghz above 1.4v already and with voltage competition almost maxed out and getting even worse performance on some stuff than my older 2600K lol (lucky me that the silicon lottery of that 3770K screwed me so badly when I was aiming for 5ghz, oh and yes I already had delided the 3770K which is even extra step that 2600K never needed) 
I would really love if you could request that features cause I would totally enjoy getting the lowest voltage possible per each frequency on CPU\GPU\MEM while testing on memtest, I know it won't save much power but I really enjoy spending time tweaking
Edit: forgot to say which memtest is best? I used mostly memtesterNX but I have both that I got from a older OC suite and did both long and stress test...
I understand the OC being different for each console I'm used to that on PC like going from my nice I7 2600K at 4.9ghz below 1.4v with 0 voltage compensation needed to a damn nasty upgrade (downgrade) I7 3770K that I could only do 4.5ghz above 1.4v already and with voltage competition almost maxed out and getting even worse performance on some stuff than my older 2600K lol (lucky me that the silicon lottery of that 3770K screwed me so badly when I was aiming for 5ghz, oh and yes I already had delided the 3770K which is even extra step that 2600K never needed
) I would really love if you could request that features cause I would totally enjoy getting the lowest voltage possible per each frequency on CPU\GPU\MEM while testing on memtest, I know it won't save much power but I really enjoy spending time tweaking
Edit: forgot to say which memtest is best? I used mostly memtesterNX but I have both that I got from a older OC suite and did both long and stress test...
Last edited by guily6669,
I'm not overvolting, but rather undervolting from the default to 1.1 but like I said on memtest at least at 1785mhz the score is very similar between default and 1.1v but the difference is around 31.2mv at that frequency.
I understand the OC being different for each console I'm used to that on PC like going from my nice I7 2600K at 4.9ghz below 1.4v with 0 voltage compensation needed to a damn nasty upgrade (downgrade) I7 3770K that I could only do 4.5ghz above 1.4v already and with voltage competition almost maxed out and getting even worse performance on some stuff than my older 2600K lol (lucky me that the silicon lottery of that 3770K screwed me so badly when I was aiming for 5ghz, oh and yes I already had delided the 3770K which is even extra step that 2600K never needed
I would really love if you could request that features cause I would totally enjoy getting the lowest voltage possible per each frequency on CPU\GPU\MEM while testing on memtest, I know it won't save much power but I really enjoy spending time tweaking
I misread your numbers I thought you were mainly overvolting and only undervolting some while you are actually just undervolting some and keeping many the same. Yeah you're doing the right thing with undervolting. I just lack reading comprehension lolhttps://github.com/KazushiMe/Switch-OC-Suite/issues/63
I summited your suggestion on GitHub.
Nice, Thank You!
I swear I was waiting for you to finally make a OC Suite thread (actually someone) so I could request that features to be requested on its github
Really hope KazushiMe ever makes a account here...
Ps: cheers
I swear I was waiting for you to finally make a OC Suite thread (actually someone
) so I could request that features to be requested on its githubReally hope KazushiMe ever makes a account here...
Ps: cheers
That was your master plan all along lol I don't mind hahaI swear I was waiting for you to finally make a OC Suite thread (actually someone
Did some RAM performance comparisons in some games going to add them to the first post.
v1 Switch RAM Tests - Games using 60FPS cheats to unlock framerate
Crash Bandicoot N. Sane Trilogy - Docked
12.48% increase in FPS with 1862 RAM vs 1600 RAM
No Man's Sky - Handheld
10.52% increase in FPS with 1862 RAM vs 1600 RAM
Hot Wheels Unleashed - Docked
8.12% increase in FPS with 1862 RAM vs 1600 RAM
Attachments
Also some games could maybe still be misleading at least some certain MHz in the memories will also increase visual quality depending on the game though not anywhere close to how the GPU clocks affect the visual quality with dynamic upscaling, it's much harder to spot.
Though some games I tested above 1600mhz didn't seem to change anything visually to my eyes but sometimes could be so small that could be really unnoticed easily but could still maybe steal a couple FPS on the higher frequency making the difference to look smaller.
Anyway about downvolting what exactly is happening when going below certain voltage at certain frequency that will give less performance while status-monitor keep showing the chosen frequency? Is it like really something within the tegra SOC that is like downclocking at certain frequency because it detects the voltage is lower than should for it or is there other reason (like a bug in sys_clk)?
Without having full voltage control for each frequency we can never sadly optimize it 100%, really hope they add it later cause I can't wait for it
Though some games I tested above 1600mhz didn't seem to change anything visually to my eyes but sometimes could be so small that could be really unnoticed easily but could still maybe steal a couple FPS on the higher frequency making the difference to look smaller.
Anyway about downvolting what exactly is happening when going below certain voltage at certain frequency that will give less performance while status-monitor keep showing the chosen frequency? Is it like really something within the tegra SOC that is like downclocking at certain frequency because it detects the voltage is lower than should for it or is there other reason (like a bug in sys_clk)?
Without having full voltage control for each frequency we can never sadly optimize it 100%, really hope they add it later cause I can't wait for it
Crash Trilogy doesn't have dynamic res its always 480p in handheld and 720p in docked, and I seriously doubt Hot Wheels Menu uses dynamic res.Also some games could maybe still be misleading at least some certain MHz in the memories will also increase visual quality depending on the game though not anywhere close to how the GPU clocks affect the visual quality with dynamic upscaling, it's much harder to spot.
Though some games I tested above 1600mhz didn't seem to change anything visually to my eyes but sometimes could be so small that could be really unnoticed easily but could still maybe steal a couple FPS on the higher frequency making the difference to look smaller.
Anyway about downvolting what exactly is happening when going below certain voltage at certain frequency that will give less performance while status-monitor keep showing the chosen frequency? Is it like really something within the tegra SOC that is like downclocking at certain frequency because it detects the voltage is lower than should for it or is there other reason (like a bug in sys_clk)?
Without having full voltage control for each frequency we can never sadly optimize it 100%, really hope they add it later cause I can't wait for it
About undervolting quoting directly from ZachyCatGames (the original creator of the 1862 RAM patch)
'the system will claim it's running at X frequency but calculated voltages will be too low and the GPU will drop frequency, hang, or spew errors'. It's not a bug with status monitor its just how the Switch's operating system Horizon is. Status Monitor is just reporting what's in the PCV which is modified by the loader.kip. You couldn't do this same thing with L4T (Linux) for Switch because it will get rid of these 'fake' clocks.
I would highly recommend reading his entire post about the pitfalls of undervolting and why Cooler's stuff doesn't work the way he think it does: https://gbatemp.net/threads/sys-clk-under-overclocking-sysmodule.531372/post-10054079 (TLDR it's crazy undervolted so the clocks in status monitor are lying and you get worse performance then you would at stock voltages.
Also if you make an 'optimized undervolted' loader.kip please don't share it as it's only going to work on your Switch and you know already but thoroughly check using benchmarks. If you have any more questions about this undervolting stuff I'd probably ask CTCaer. Honestly I'm a bit unsure if you should do it at all.
Also read this
and this
Last edited by ChanseyIsTheBest,
Is the increase in FPS only with very high GPU frequencies? because i checked mario rabbids sparks of hope in handheld mode with 460 mhz gpu and there was no change in fps between "normal" ram and "maximum" ram.
That's quite a lot of info, thank you, I read it all.
About the Dynamic Res I was more talking in general, it's not directly related to the games you posted results...
I'm using OC suite, will gonna wait to see if they ever add voltage per each frequency cause I can't do much with the Erista CPU max voltage option for undervolting for low frequencies as there's not much control.
But if we had full voltage control on a AVG switch I still find very helpful having someone's fully optimised values for a baseline, I always like some reference on PC to start from unless I think the values are too crazy
I don't have that game but first make sure you have the .kip file or 1600mhz to maximum will keep at 1600mhz as you can check in reported clocks if that's the case...
Or if the game is locked to 30FPS and it's already there well by default OC won't do much until you use a 60fps patch but still will probably improve visual quality if it has dynamic Res like OC on the GPU...
And for last the game FPS increase with OC all depends on the game and what's more capped in it, memory sure is a huge problem in switch games but some could be CPU bound maxed out and some are heavily GPU maxed out so OC to any of the 3 and the FPS boost will depend a lot on the game.
Ps: usually a big ram OC will lower GPU usage massively making it require less GPU clocks for the performance you want out of it saving a lot of power...
Last edited by guily6669,
I'm using latest 1.4.0.1 and the default 1862mhz for it without changing anything and is working fine, I tested like 4 hours or more on memtest that is on their 1.4.0 to download with 0 errors...
I also finished 3 games, only one was crashing from hours to hours, Red Faction but it was also the only one I used cheats so most likely will be the cheats fault, no idea as everything else is working super fine for now.
ps: Mine is a ERISTA V1, not Mariko...
Oh didn't realize there's a new beta, nice and btw new atmosphere too
Nice!!! Try new beta and tell me if RAM OC don't work for you too pleaseOh didn't realize there's a new beta, nice and btw new atmosphere too
I will on latest atmosphere too, I'm updating...
but just now trying their latest .KIP on their online configurator page it doesn't even load, it gives a error, so probably there's something wrong with it unless they are going back to the old python scripts or something to edit it??
New beta don't work with it for now, it's in changelog, maybe RAM OC is disable by default, will wait, I don't know if there is a way to modify kip without online configurator
It's exactly like yours and also Tinfoil went puff which never happened crashing when after using max setting but was still reading 1600mhz like yours (can be new atmosphere too).
I'm now using all their latest OC suite beta files but with the old default loader.kip, after reboot tinfoil is working...
I think in the past they had some phyton scripts to edit the ".kip" from what I read from their older versions, but no idea either...
- Status
Similar threads
- No one is chatting at the moment.
-
-
-
-
-
-
-
-
@
The Real Jdbye:
loved nes remix, but hated having to play Baseball, Tennis and Golf in order to progress+1 -
-
@
Psionic Roshambo:
Not a big fan of NES kid Icarus either... I know it was popular but I didn't care for it. -
-
@
cearp:
What stuff were you burning?
I used to burn tons of dvds (data files) many years ago, but haven't burnt a disk in ages. -
-
-
-
-
-
-
-
-
@
Psionic Roshambo:
BakerMan yeah me too.... maybe that's gonna be dinner for me there is this place near me I have been dying to try and they have them lol
-
-
@
Psionic Roshambo:
@The Real Jdbye, I thought about that and I have blank DVD's and Blurays but honestly after like 90 songs I just listen to it a couple of hundred times toss it out and burn something else or have like 2-3 CD's. Florida here the heat in the car tends to ruin the CD's after like a few months even commercial pressed ones don't last too long.
-
-
