Cooling The Raspberry Pi 4 With The Fan SHIM & FLIRC For Better Performance

[coder]

I don’t mind that RPI 4 comes without a heat sink. What is problematic is that Raspberry PI did not provide for a proper fan header or other solution for users to leverage.

Like salt on assault, there are no provisions on the PCB to physically attach a heatsink.

Good point.

Hence I found my own way ;
Plucked a brass heatsink from old &faulty Motherboard and attached it to the CPU using cheap white heat sink compound. The system is kept open.

Yes, exactly. Use a decent heat sink and decent thermal compound.

Perhaps you know this, but try to use the least amount of thermal compound possible, as long as it covers the entire intersection between the SoC and heat sink. Using more thermal compound than necessary will actually reduce thermal conductivity, since it's not as conductive as the metal of the heat sink.

BTW, perhaps the heat sink you're using is copper? Copper is the best (solid) heat sink material, while aluminum is second best and more common (since it's cheaper).

[coder]

Cortex A73 is only slightly more efficient than Cortex A72, but the 200Mhz difference in bigger cores matters..
I am comparing RAM speeds because they matter, in heavy workloads..

But your comparison of RAM speed to core speed is unjustifiable.

Look at the ODROID benchmarks vs. their N1. Real benchmarks beat your facile analysis.

https://www.hardkernel.com/shop/odroid-n2-with-4gbyte-ram/

You can't compare it to Rockpro64, simply because it doesn't have the features Rockpro64 has..
Rockpro64 has more peripherals, faster CPU, faster Ram, and a full pcie x4..

It's funny how you started out trying to compare performance and now you're saying it's not valid?

Of course it's valid, for someone who doesn't need the RockPro64's extra capabilities. However, for someone who does need them, then there's not much point in it, because it's a given which one they'll buy.

N2 is half the node size, that helps, but sooner or later it will throttle, because of self heating, no matter what..

I'm not sure you understand thermodynamics. Their temperature data clearly shows a plateau, which means the heatsink has reached steady state and is able to dissipate all the additional heat it's adsorbing from the SoC.

AmLogic is knows to cheat, pushing up some cores until they achieve a degree of temperature, then throttling them down..

This is not cheating. This is what all modern SoCs do, in order to avoid burning out. That's why you also need to look at the clock speed graph that ODROID posted right below their temperature graph, which shows no throttling.

This behaviour added to the fact that, it has relatively slower Ram, makes the idea, not useful..

There are a lot of factors that affect performance, besides RAM speed. Not all tasks have the same amount of memory I/O, and cache varies between SoCs. That's why I say you need to just look at the benchmarks.

Also, with a Soc beneath, the heat doesn't spread..

Again, this is you not understanding thermodynamics. Heat conduction is not dependent on orientation, and that's how heat is transferred into the N2's heatsink. But you don't even need to understand that much - just look at the data. Regardless of why or how, the data clearly shows the N2 doesn't throttle.

[coder]

Is nobody bothered by the fact that these diagrams don't start at 0? I completely misread the diagrams at first. Maybe it's me

The bar graphs all seem to start at 0, though they could be labelled to confirm this.

As for temperature, which zero? Zero F, C, ...Kelvin? I'd suggest the temperature graphs use current room temperature as the baseline, which I'd hope was a bit below where Michael seems to have put it.

[danmcgrew]

@ #40--

"...I don’t mind that RPI 4 comes without a heat sink. What is problematic is that Raspberry PI did not provide for a proper fan header or other solution for users to leverage."
*******************************
No; what is really problematic is the fact that the Raspberry Pi Group cannot do good, thorough, basic electrical/electronic engineering design. They are--demonstrably--NOT engineers, nor do they employ true hardware-design engineers.They turn out poor designs, and then are not open and transparent about the problems which they create with their poor designs.

Has everyone forgotten that the Raspberry Pi 4 has had, since it was introduced, a basic, BASIC engineering problem with the USB-C function of the board* because a Raspberry Pi "engineer" decided to get 'cute' with that design; when all that was needed was to directly copy the implementation from the USB-C specification? **

Has everyone forgotten that the highly-touted "PoE Hat" was an unmitigated disaster because some "engineer" DID DIRECTLY COPY the "suggested design", or "reference design" in the power-supply chip's data sheet, provided there simply for checking feasibity?

It goes without saying that two of the most powerful weapons--along with hard-won, high competency in electrical / electronic design--in the engineer's arsenal are validation testing, and Q-A testing; two crucial items of the design process about which The Raspberry Pi Group cannot and will not be bothered with.
The Raspberry Pi Group has proven, for the world to see, that they have no concept of what it takes to perform good, solid electrical and electronic engineering design.

Time to resort solely to what they apparently do best, now: printing magazines.
++++++++++++++++++++++++++++++++++++++++++++++++++ ++++++++++++++++
*
"Pi4 not working with some chargers (or why you need two cc resistors)"
https://www.scorpia.co.uk/2019/06/28...-cc-resistors/

"Your new Raspberry Pi 4 won't power on? USB-C cable problem now officially confirmed"
TechRepublic; Nick Heath, on July 8, 2019, 4:42 AM PST
“... I expect this will be fixed in a future board revision [says Eben Upton]...It's surprising this didn't show up in our (quite extensive) field testing program."
https://www.techrepublic.com/article...lly-confirmed/
++++++++++++++++++++++++++++++++++++++++++++++++++ ++++++++++++++++++++++++++
**
How to design a proper USB-C™ power sink (hint, not the way Raspberry Pi 4 did it)
Benson Leung
https://medium.com/@leung.benson/how...it-f470d7a5910

Pi4 not working with some chargers (or why you need two cc resistors)
https://www.scorpia.co.uk/2019/06/28...-cc-resistors/

[andreano]

Let's not forget to appreciate that some level of power management is actually implemented, which is why it can work without a heatsink at all.

Decent drivers, hereunder thermal throttling, is a luxury you don't always get with even expensive SBCs. My WandPi managed to overheat when idling, despite the generous heatsink that came with it. I measured 107C (heatsink temperature) with IR thermometer.

[cobratbq]

Talking about temperatures and zero is always difficult besides absolute zero i.e. 0K of course. I think it's reasonable for such measurements to start at room temperature in the diagrams.

Well, that makes sense, except that the starting value is arbitrary (32, 37, 38, 39) and seemingly dependent on the measurements. On the other hand, I see your point of starting from 0 doesn't mean that much. I'm mostly bothered about the the skewed impressions of relative difference between the lines. Anyways, good point, though.

[mvdvarrier]

QUOTE=wizard69;n1119577]I don’t mind that RPI 4 comes without a heat sink. What is problematic is that Raspberry PI did not provide for a proper fan header or other solution for users to leverage. [/QUOTE]

Like salt on assault, there are no provisions on the PCB to physically attach a heatsink. The best and most effective way to attach a heatsink to the chip is mounting it using clips or screws. that is not possible here. I am from India. In summer the atmospheric temperature will climb to 40-42 Degree Celsius here. It is not practical to avoid some kind of cooling. But the FLIRC case and ICE Tower cooling set is very expensive. The 16$ FLIRC case will cost here about 70$ after shipping, VAT, GST and retailer's cut. Hence I found my own way ;
Plucked a brass heatsink from old &faulty Motherboard and attached it to the CPU using cheap white heat sink compound. The system is kept open. When idle after about 20 Min usage the temp settled around 50-60 and then I stretched the cpu with a script

#!/bin/bash
clear
#nice little loop

for f in {1..7}
do
vcgencmd measure_temp
sysbench --test=cpu --cpu-max-prime=25000 --num-threads=4 run>/dev/null 2>&1
done

vcgencmd measure_temp

Without any fan the temperature readings as follows;
56, 65, 69, 73, 74, 75, 76, 78.
No throttling occurred.
With a ceiling fan with average speed, the readings are;
52, 58, 57, 59, 60, 59, 59, 58.
I am now using the pi 4 as my Desktop las 2 weeks. It found quiet stable with this set up
So I am not going to buy any expensive things for this.

[George99]

Is nobody bothered by the fact that these diagrams don't start at 0? I completely misread the diagrams at first. Maybe it's me

Talking about temperatures and zero is always difficult besides absolute zero i.e. 0K of course. I think it's reasonable for such measurements to start at room temperature in the diagrams.

[tuxd3v]

But the N2 is faster, being based on A73 cores, while the RockPi64 uses A72 cores. ODROID made a N1 that was based on the same SoC as the RockPi64, but they cancelled it.

Also, I don't know why you're comparing DDR speeds vs. CPU clock speeds. The N2 runs its A73 cores up to 1.8 GHz and its A53 cores at about 1.9 GHz.

Cortex A73 is only slightly more efficient than Cortex A72, but the 200Mhz difference in bigger cores matters..
I am comparing RAM speeds because they matter, in heavy workloads..

The N2 simply doesn't need a fan. It's manufactured at 12 nm. When running all 6 cores at sustained max load, using the stock, passive heatsink, it plateaus around 73 C, which seems to be below the point of throttling.

Have a good read through their benchmarks (including their N1 comparisons, should you want to compare it with RockPi64).

You can't compare it to Rockpro64, simply because it doesn't have the features Rockpro64 has..
Rockpro64 has more peripherals, faster CPU, faster Ram, and a full pcie x4..

N2 is half the node size, that helps, but sooner or later it will throttle, because of self heating, no matter what..
I ran a Rockpi64, for 14 months @2Ghz, full throtle, at around 55C, completely stable..
Try do that with the N2..

Clearly, I don't agree with your conclusion.

Well,
You have the option to disagree,
But even tough that s922x is half the node of RK3399,
AmLogic is knows to cheat, pushing up some cores until they achieve a degree of temperature, then throttling them down..

This behaviour added to the fact that, it has relatively slower Ram, makes the idea, not useful..
Also, with a Soc beneath, the heat doesn't spread.. and Rockpro64 has a full size pcie x4 slot, with a lot more connectivity..

They are different boards..

Even RockPi4 has pcie M2,
But it also suffers from the same problem N2 suffer( CPU in Bottom of the Board.. )..

[sykobee]

FLIRC case looks good, if you get the RPi4. I'd like to see if it helps with overclocking.

But let's admit it, most uses of a Pi don't push it often, so the default configuration is fine. But those cheap stick-on heatsinks are clearly pointless.

Note there isn't much difference, performance-wise, between the A72 and the A73, at the same clock. The A72 is wider, the A73 is more optimised (smaller core).

But the N2 is a good board if you need its features.