The bottom of the board is probably quite bare and should fit the Thermalright bracket probably.
But yeah it’s a bit trial and error unless the official fan goes for sale.
The bottom of the board is probably quite bare and should fit the Thermalright bracket probably.
But yeah it’s a bit trial and error unless the official fan goes for sale.
that thermalright one looks like it would fit the holes just fine, but the other problem is that the heat spreader on the processor is much lower than an intel CPU, might need to use a real thick thermal pad.
can you take a picture of that? or measure the diameter of the cooler screw?
Fair point about the height issue on that Thermalright. There are some super slim fans which aren’t too expensive on Ali Express that might work.
Something like https://www.aliexpress.com/item/1005006216861796.html
Might be worth a gamble, or something like that, knowing it’s LGA 115x
the intel stock cooler is maybe 1-2mm too big to fit the mounting holes, i wouldn’t worry about that too much as it doesn’t make contact with the processor anyway. it uses plastic push pins instead of screws and isn’t adjustable so not ideal anyway,
Why exactly?
You’re not dealing with some inefficient x86 thingy but RK3588 which doesn’t even throttle running demanding benchmarks that represent a load scenario you would’ve a hard time to see with any real-world workloads.
As for (problems with) dimensions: ROCK 5 in ITX form factor
Since fans are so annoying (wasting energy, making noise and if you use a crappy thermal pad instead of an advanced thermal compound do more harm than good) I might end up glueing a passive heatsink to RK3588 once testing is done (as can be seen in the upper area inside the enclosure on this picture):
For what it’s worth, I tested another Thermalright cooler and the back plate’s mounting posts were also too big for the 5 ITX’s holes. I bought some cheap Xilence one on Amazon that did then fit so I’m playing with that instead. It didn’t make contact with thermal paste alone, so I used a 1mm thermal pad and it doesn’t go above 34-35c in a room with a 23c ambient room temp whilst under an all-core, full synthetic load.
To echo what tkaiser said though, you really don’t have to for actual thermals unless you’re sticking it in a super cramped case with minimal airflow. I tried it on mine for the meme basically
And these things can ruin everything since usually these pads are cheap and crappy. See this insane monstrosity of a ‘cooler’ not being able to dissipate heat efficiently away since the most important part (transfer heat away from the SoC) hasn’t been paid attention to:
Well I was thinking it could run fanless but by my experience of other ARM PCs, they can easily climb to over 80C when using the GPU for half an hour. That is why I was thinking of a fan, but agreed, it’s premature without having received the unit.
In general if the CPU is mostly idling, sure, but GPU use can make ARM chips get quite hot by experience.
This will mostly depend on the process node the SoC is on. RK3588 is made in a somewhat modern process (Samsung’s 8LPP) and the SoC of the ARM laptop I’m typing this on is on TSMC’s N3B process and also passively cooled. No problem with GPU/VPU whatsoever.
Experiences made with older SoCs on ancient processes can’t be transferred to more modern designs.
Super late reply, sorry!
Luckily I did have some somewhat decent pads here as they’re not crippling performance. I think my ambient temp was 23c and it was idling around 26 or so. Certainly overkill to strap a large chunk of metal and a fan on this, as we all knew
This is what I got. You can try either one.
or
With the passive, 25x25mm heatsink the CPU temperatures were about 40C. With the Thermaltake, it is 25-27C.
My average temps were around 40-50c and with some constant video playback or 3D using Panfrost around 85c but it was stable. Even though the case had plenty of holes I believed this was due to ambient temperature and when I introduced a 50mm extractor fan in the case the max temperature was more like 70-75c so a drop of about 10c but still nothing 23c or anything low like that.
What are we measuring? the outside of the CPU as I don’t really have a way to measure that. I’m just grabbing the readings from the boards internal sensors.
I am definitly convinced it is OK to run the CPU bare from my testing, but I don’t see temperatures that low either.
It is also possible that the readings I was getting are way off and too high, as the case was only only slightly warm, it wasn’t like radiating lots of heat at all so 85c seems hard pressed.
That size 25x25mm isn’t that a bit big?
I Googled the size of the RK3588 and it says 146 mm x 102 mm (5.75” x 4.02”)
So I would probably get these 14x14x6 mm? or is the larger one OK?
I guess a slightly oversized heatsink is fine considering there is nothing but empty space around the SoC.
No, it’s not too big. The size of the RK3588 is 23x23mm (I checked on the Rockchip website), if you are planning on getting the passive heatsink, get one that is 25x25mm. But in my trials of the passive heatsink, RK3588 was getting to about 45-50C. I switched to this one, https://www.amazon.com/dp/B00O4UIZ86?ref=ppx_yo2ov_dt_b_product_details&th=1, and it is consistently running at 25-28C.
146mm x 102 mm is the size of the whole mini-ITX board, not the chip.
50 °C is perfectly usable and indeed quite cool for a CPU.
50C is light use, 85C is Youtubing. It’s still OK though but you misread that. 50C is only like basic usage, but once you do a bit of Youtube, a bit of 3D it can still go up to 85C BUT it is OK on those temps.
No doubt, 50C is perfectly fine but 27C is even better
Yeah 50C is perfectly fine but 27C sounds unrealistic, like the device is off Still like I mentioned it did gets to 85C when doing YouTube so tiny heatsinks should help it a little.