Inconsistency in the performance of RPi 3 B+

[HawaiianPi]

HawaiianPi, thank you very much for sharing these results with us. They look cool in both senses of this word Unfortunately, Flirc Gen2 with shipping to my country that I found costed a bit too much, so I have ordered another aluminium case suggested in this thread by RaTTuS at aliexpress and if it will show similar performance I will be more than happy.

I hope that case actually fits the newer 3B+ model, because it has a taller SoC than the original 3B.

From what I have read, that case cools the model 3B a bit better than the Flirc, but it can cause wireless problems (it completely envelopes the Pi in metal, while the Flic has plastic on the bottom and near the wireless antenna).

Concerning RPi as a daily driver. A boot time of 11sec is almost a killer feature by itself.

My Lenovo Legion Y520 laptop boots Win10 faster than that (10 seconds, give or take a few decimals). Of course the NVMe SSD it boots from cost more than all my Pi computers combined, but it is very fast.

Another astonishing thing for a Windows user is an update of 234Mb that will use additionally only 1024b of disk space

Agreed, Windows updates are a mess. I did a clean install after the Spring 2018 update to get rid of all the leftover cruft from previous updates. It's also annoying how Windows needs to reboot multiple times to install updates (although that has improved slightly with Win10).

[jahboater]

HawaiianPi, thank you very much for your results.

There is a tiny advantage for the black finned heat sink, but for practical purposes the two cooling solutions have similar class leading performance - at least close enough not to make it worth while changing, especially if the case solution is preferred.

FLIRC case - ambient 20C

Code: Select all

Time      Temp    CPU        Throttle       Vcore
22:52:53 38.1'C  600MHz 0000000000000000000 1.2V
...............
22:54:15 47.8'C 1400MHz 0000000000000000000 1.3750V

Kintaro heatsink - ambient 24C

Code: Select all

Time      Temp    CPU        Throttle       Vcore
10:09:11 41.3'C  600MHz 0000000000000000000 1.2V
...............
10:10:34 47.2'C 1400MHz 0000000000000000000 1.3250V

[HawaiianPi]

Yea, the Flirc numbers could probably be improved with a better thermal pad, but I haven't gotten near throttling temps, even in the summer when it's much hotter here.

I may give one of those Kintaro heatsinks a try. Don't really need one, but I just love buying gadgets.

[jahboater]

Yea, the Flirc numbers could probably be improved with a better thermal pad, but I haven't gotten near throttling temps, even in the summer when it's much hotter here.

I wonder if some very thin (0.5mm or less) washers were placed under the board mounting holes, raising the height of the board very slightly might help, then a very thin layer of high performance TIM could be used, rather than a pad.
Of course too high and the sockets would be unusable

[bensimmo]

Or remove the heatspreader.

[HawaiianPi]

I wonder if some very thin (0.5mm or less) washers were placed under the board mounting holes, raising the height of the board very slightly might help, then a very thin layer of high performance TIM could be used, rather than a pad.
Of course too high and the sockets would be unusable

Interesting idea there...

According to Flirc, the thermal pad for the Pi3B+ is 0.4mm, so I guess that's how high you'd have to raise it. Do that and slap on some Thermal Grizzly and I'll bet it would run... well, a tiny bit cooler. We are talking about a Pi SoC here, not a 100+ watt TDP intel CPU. Also, the fit of the ports on the Flirc is pretty tight. Not sure how much you could raise it without putting pressure on the ports. But, the family is heading up to Seattle for a few days tomorrow, and I'm staying home to take care of the animals, so if I get bored I may just look into that.

On the other hand...
I have some 0.5mm thermal pads with a higher thermal conductivity than many pastes. One of these days I may try swapping the pad in one of my Flirc encased Pi3B+ computers and comparing it to the stock pad. I have no idea how good or bad the stock pad is, but my Flirc cases do radiate heat that I can feel, and the Pi stays under throttling temps, so they are probably not terrible.

In the end, though, I'm not sure there is much room for improvement, since we're only dealing with a few watts of thermal energy from a Pi. If you're already using a Kintaro heatsink or Flirc case you're probably not going to see a big drop in temps no matter what you do (unless you get crazy with something like water or sub-zero cooling).

[jahboater]

Interesting. 0.4mm. For Intel PC's that would be ridiculous. The heat sink contact pressure is very large (4 large bolts on my PC here).
Serious users might lap or polish the contact surface of the heat sink to maximize the metal to metal contact area. Great care goes into the choice of TIM.

For the Pi, it is, as you say, of academic interest.
My 3B+ will never throttle under any "real" load, is unconditionally stable, and I expect the lifetime is in decades.

It is fun though to explore the possibilities though