Nintendo

20387 readers

162 users here now

A community for everything Nintendo. Games, news, discussions, stories etc.

Rules:

No NSFW content.
No hate speech or personal attacks.
No ads / spamming / self-promotion / low effort posts / memes etc.
No linking to, or sharing information about, hacks, ROMs or any illegal content. And no piracy talk. (Linking to emulators, or general mention / discussion of emulation topics is fine.)
No console wars or PC elitism.
Be a decent human (or a bot, we don't discriminate against bots... except in Point 7).
All bots must have mod permission prior to implementation and must follow instance-wide rules. For lemmy.world bot rules click here
Links to Twitter, X, or any alternative version such as Nitter, Xitter, Xcancel, etc. are no longer allowed. This includes any "connected-but-separate" web services such as pbs. twimg. com. The only exception will be screenshots in the event that the news cannot be sourced elsewhere.

Upcoming First Party Games (NA):

Game | Date

[S2] means Switch 2 only.

Other Gaming Communities

founded 2 years ago

MODERATORS

slimerancher@lemmy.world

CosmicSploogeDrizzle@lemmy.world

CosmicSploogeDrizzle@lemdro.id

picandocodigo@lemmy.world

Improved performance (?) from changing out thermal paste on Switch - and a question for anyone else who has tried (feddit.uk)

submitted 1 day ago by heythatsprettygood@feddit.uk to c/nintendo@lemmy.world

8 comments fedilink hide all child comments

I've been doing some experimentation lately with my Switch, and have replaced the thermal paste on all layers (under the copper heat spreader, under the heat sink, and on top of the heat pipe). I used a combination of Thermal Grizzly PhaseSheet (for heat spreader and heat sink) and K5 Pro (for heat pipe) if anyone is curious. I am running a Switch with the original SOC (higher heat and power) that I bought in about late 2018.

I wasn't expecting much if anything, but in games they seem to have had a bit of a performance improvement, especially in docked mode. The two games I especially noticed were Fire Emblem Three Houses and Pokemon Violet. In Three Houses, the loading sections in explore mode (so fully in game, with everything still showing on screen) took far less time than I remember (1-3 seconds compared to 5-10) along with a higher frame rate during the loading, and in Violet the frame rate seems a decent bit more stable. Unfortunately, since my Switch is not modded, I do not have any hard performance numbers to back up these perceived improvements. I have heard that the Switch will prioritise CPU clocks over GPU when loading due to thermal restrictions, but I cannot find anything about it being able to boost both if it has the thermal headroom. The outside of the Switch is definitely exhausting more heat, which could suggest more heat is being taken away from the SOC and consequently cooling it down.

Does anyone have any experience with doing this, and have you gotten a similar result? I am curious to find out if this is just placebo, or if this is genuinely a way to increase performance even just by a little bit.

you are viewing a single comment's thread
view the rest of the comments

[–] any1th3r3@lemmy.ca 2 points 1 day ago (1 children)

Interesting!
I could only find limited info about the SoC's TjMax / operating temperatures, although we could assume it's somewhat similar to the actual Tegra X1, in which case:

This datasheet lists a TjMax of 105°C (page 2) and several thresholds before that (page 17) at which the APU would downclock (or shut down), with no actual value unfortunately.
Gamers Nexus talks about the above, with a supposed throttling threshold at 70°C - and possibly 90°C - where frequency drops by 6%.
Digital Foundry made a video testing Switch overclocking years ago (apologies for the ResetEra link, but there's a good summary there), where they had operating temperatures in the 60s (though I'm not sure that's the junction temperature).

That being said, my (V1) Switch is modded so I could test that a bit more, and check operating temperatures and frequencies before/after a repaste.

[–] heythatsprettygood@feddit.uk 2 points 1 day ago

Thanks for the data sheet! I struggled to find anything even vaguely official about the X1, so this is quite helpful. I think the throttling at around 70 could be our culprit, as (assuming most of the docked power goes to the SOC) I've had processors in portable devices in that sort of power class with similar size coolers have junction temperatures go into the 80s and 90s regularly, so probably in heavier games the temperature goes past that threshold regularly on a lot of systems. This is not helped by the thermal compound both ageing and probably not being the best thing even on day 1, in comparison to something like PhaseSheet that is among the top performers.

Part of me is curious as to if Nintendo have a low temperature target set for the SOC that could see the boost stop early, and the lower temperatures from a repaste allow us to get past it. The data sheet suggests that the throttling thresholds are configurable considering there are no default numbers in the document at all, so Nintendo might have gone more conservative to increase system reliability. If you can, try and get some frequency data, as this might be the reason as to why I'm seeing an improvement.

I am excited to see your results, as it'll give some hard numbers to see if this is worth all the trouble for everyone to accomplish (the heat spreader is quite the pain to remove).