[pslam]

This article is actually describing how the Raspberry Pi 4 does NOT need a fan.

This is not the 1990s. It is perfectly acceptable and even advantageous to design for a high peak:normal load ratio, with thermal throttling. In this case, it allows for a compact, cheap, fanless design for the vast majority of users.

There is no evidence the heat dissipated will impact lifespan. It is common for the components picked out in particular (power supply, USB-C controllers) to be deliberately designed to run hot. They aren't made on the same process as the SoC.

I feel like there is a missing piece of the software/hardware design art here. There are many takes like this on the Raspberry Pi 4 design. Why only one ethernet? Why no fan? Why not more USB-C? Because it's $35 and because, perhaps, you aren't the target majority market. It's going to satisfy the vast majority of people, and those it doesn't have very simple and cheap ways to mod it so it does.

[morganvachon]

I'll respectfully disagree. Running my Pi 4 with no fan/heatsink reduces performance to an absolute crawl. It's on par with the desktop on the original Pi released in 2012, i.e. nigh unusable for anything beyond saying "yep, it boots to desktop". Thermal throttling effectively neuters the device.

I have it in an open-air acrylic "sandwich" style case now, with the same Pi-Fan as the author, and it feels performant enough to use as a daily driver for web browsing and other light duties (basically on par with any Chromebook I've come across the past few years). It's still not "desktop replacement" level due to the SD card performance hit, but it's finally good enough for its intended use case in education without being frustrating. Once boot-from-USB3 arrives it will likely be fast enough to use as a second Linux workstation in a serious capacity.

[Erlich_Bachman]

He said "no fan", why do you include "no heatsink" into that? Those are very different things to operate without.

[morganvachon]

The parent was claiming the Pi was designed with thermal throttling in mind so that no cooling solution was needed. I disagree based on my experience (I’ve owned at least one of every model B Pi since launch as well as three revisions of the Zero).

[floatingatoll]

The assumption implied here is that the Raspberry Pi should deliver continuous peak performance equal to its burst performance.

If the RPi4 specs indicated a peak performance of 10% of burst performance, adding a fan would be a material alteration of the published thermal envelope of the purchased RPi4.

Was this performance characteristic clearly indicated at time of sale?

If so, then the article title, and the assumption above, is wrong: RPi4 does not need a fan, as long as it adheres to the published burst and continuous specs. Clearly there’s room in the thermal envelope for improved continuous performance with a fan, but that in no way is a “need” if they are transparent about this spec.

If not, then they are in good company with Apple and other compact hardware manufacturers in failing to publish their thermal envelope burst/continuous details in clear specifics. If they comply enough to say “min speed / burst speed” then they comply with what modern users generally expect in marketing documentation.

Does the RPi community wish further detail to be included at time of sale regarding thermal envelope behavior and timings at room temperature using the passive cooling case provided? If so, that’s a fair request to make of RPi and one they can easily comply with.

[geerlingguy]

I am not directly the target market, no. And the throttling itself is an indication that the heat was designed/accounted for in the Pi's design.

However, to say that the majority of users don't need better cooling, and it won't cause problems most of the time, seems inaccurate. If you use a microSD card to boot/run the Pi, how many of them are rated for blisteringly-hot operating conditions 24x7 (many people using the Pi as a computer will have it booted pretty much all day and just turn off the monitor, or go to screensaver).

And I have already had one PoE HAT's PoE socket pop partly off the board when removing it from a Pi 3 B+. Partly due to stress of flexing the board, definitely... but that kind of tiny solder joint on a part that is stressed when making a connection is exactly the point of failure that may not be directly caused by, but is definitely not helped by thermal stress.

[cyanoacry]

> There is no evidence the heat dissipated will impact lifespan.

I have to nitpick at this a little bit as a professional EE who works in high-reliability electronics. Wearout rates absolutely do depend on temperature (and thus heat), and thus the chips used here will have a shorter lifetime than those with active cooling. Now, whether that lifetime will be long enough for the common user is another question (maybe it's 1 million hours of life that get reduced to 100,000 hours, so not normally noticable).

[pslam]

Yes, wearout rate definitely depends on temperature, but without reference to any actual data, this is what I mean by "no evidence".

Is it reducing 10 year lifespan to 9 years? 9.99 years? 5 years? Was it 50 year lifespan? It's pointless conjecture.

[tkiley]

If the pi is designed to throttle at a temperature where its lifespan is reduced from 50 to 49 years, then it is throttling at a gratuitously low temperature that affects quite a few use cases.

On the other hand, if it is designed to throttle at a temperature that materially reduces the lifespan, then it needs a fan to preserve its lifespan.

Either way, the hardware is starkly suboptimal for a pretty large set of advertised use cases until you add a fan.

[abainbridge]

> If the pi is designed to throttle at a temperature where its lifespan is reduced from 50 to 49 years, then it is throttling at a gratuitously low temperature that affects quite a few use cases.

That's another claim that needs evidence. It seems possible to me that the thermal throttling is designed to prevent logic errors.

[kingosticks]

It's way, way more likely this is the reason than anything to do with lifespan.

[ryanmercer]

To add to this, I wonder if a properly designed case could create a chimney effect (not unlike the crossdraft kiln Primitive Tech made in his most recent video) where escape heat is drawing in cooler air in such a fashion as to be adequate enough fo the vast majority of applications.

Not that fans need to be very loud, or even spin very fast for something of this size, I imagine relatively low RPMs (probably sub 100 RPM) could still achieve considerable cooling, especially with a crossdraft optimized case.

[geerlingguy]

This is why I suggest in the video to just stick some ventilation holes in the top cover. Instead of the heat pooling in that area, it could escape, and there would be some natural convection—in through the gaps around the ports, out through the top. It wouldn't be as good as a fan, but definitely better than slapping a heat sink inside the case.

[tracker1]

My plan was to just use a Dremel to notch some holes in the top of the case I ordered a kit with... I just ordered another case that has a fan and will try that... the new one will probably take over retro-gaming chores and I'll find something else for my 3B+ to do.

[guelo]

But then you would need a "This Side Up ↑" label and you would only be able to orient the case in one way.

[simongr3dal]

The case has small rubber feet on one side and the case is slightly dome shaped on the other, I think people will be able to place it correctly without an explicit instruction, plus there is only one “right way” to place it as it is already.

[darkpuma]

It would work for most people, wouldn't it? And the people putting their raspis into unusual scenarios for which the case was poorly suited could simply choose to not use that particular case. Encouraging thermal convection (ideally flowing over a proper finned passive heatsink) with a chimney design seems like a pretty good idea to me.

[neetdeth]

See also: the Power Mac G4 Cube

[codesushi42]

Fans fail too.

The heat dissipated will probably shorten the lifetime of the board. At $35 who cares though? And you have more to worry about from the SD storage.

The RPi needs an SSD. :)

[unixhero]

Just get an Intel NUC

[inetsee]

A quick look at Newegg shows an Intel NUC that includes a CPU, memory, and an SSD at over $500. I don't think so.

[ja27]

I put together a little Gigabyte Brix box (GB-BLCE-4105-BW) for $212 shipped. 2.5 GHz 4 core Celeron, 8GB RAM, 250GB SSD. Not a bad little PC.

[ekianjo]

A used NUC with a Celeron CPU can be had for 70 usd these days and is way faster than any Raspberry Pi.

[SkyPuncher]

Flip side, I almost pulled the trigger on one from Amazon yesterday for $118. They had $500 options, but I didn't need it.

[universenz]

I know right? People need to quit suggesting an Intel NUC as an alternative to $35 RPi.

[unixhero]

I am one of those people, and I do this whenever people want more performance, connectivity or anything else that the PI 3 or PI 4 doesn't have. Then in my view a logical alternative is to man up and pay the reasonable price for an Intel NUC to achieve the solution one is looking for.

[tuananh]

adding up case, charger, a decent sd card and the pi is way above 35$

[mlonkibjuyhv]

How's the reliability of those these days? At a previous gig I witnessed about a 15% failure rate within 3 years.

[amelius]

If you advertise a product with certain specs, and the user can only use the full capability during a fraction of the time, then that is misleading.

[throw20102010]

It might be misleading, but it's not uncommon. Apple MacBook Pros have thermal throttled at high loads for the past several years. It has throttled for so long that reviewers were surprised that the latest version seems to be using a decent thermal paste and throttles much less.

So if Apple got away with it for years on computers that mostly sold for over $2k, then I think the RPi foundation will probably get away with it on a $35 board.

[the_pwner224]

If Apple can get away with it then the RPi will also get away with it, but it is still a highly deceptive and fraudulent practice.

I spent $2000 on a XPS 15 and peripherals two years ago; if I had known about the thermal throttling I would not have purchased it. Dell literally robbed me of a thousand dollars - had I known that all the 'ultrabook'-style laptops had throttling issues, I would have bought a cheaper and sturdier and higher-specced gaming laptop which beats the XPS in every category except battery life and Thunderbolt. Instead, I spent more and got a substantially worse product.

Collectively laptop manufacturers have defrauded people to the tune of hundreds of millions (even billions?) of dollars. That's not OK.

[dangus]

All portable laptops throttle. What we are vaguely talking about is a firmware bug in the 2018 MacBook Pro i9 that was patched within the first week of release.

It's only a problem if throttling takes them below the advertised base clock under normal conditions. There is absolutely nothing deceptive about this.

A gaming laptop is not really a laptop at all by comparison. They still get less than 4 hours of battery life under load, they're an inch or more thick, they're heavy. Many of them do not fit in backpacks. Gaming laptops are essentially designed for plugged in operation.

I'm a little confused at what's "sturdier" about a gaming laptop as well. Did you break your XPS 15 physically? Gaming laptops have tons of flex and plastic-ness, see MSI.

What benchmark did Dell promise you exactly? Dell didn't rob you of anything. Your own unrealistic expectations did.

[the_pwner224]

> All portable laptops throttle.

> It's only a problem if throttling takes them below the advertised base clock under normal conditions.

Seems like a bit of a contradiction?

Dell said that my laptop would have an i7-7700HQ at 2.8 GHz + turbo to 3.something, and a GTX 1050m.

The laptop Dell sent me behaved like this under load: it goes to the max turbo speed, overheats within a few seconds to a minute, and then throttles to 800 MHz.

It's like if someone advertised a car as having a 280 HP engine, but the engine controller limited it to 100 HP because it had an inadequate cooling system and would overheat at any more load. It is deceptive.

With the laptop, underclocking can fix the CPU throttling - mine can now sustain the max turbo speed forever at 70 degrees. But that's like buying the 100 HP car and modding it to get to 280 - you were sold a faulty product. In the case of the laptop it's fixable in software so that's not as big of a deal, but how many normal people do you think would be willing to mess with their CPU voltages? Or even be aware of the throttling?

Even after underclocking the CPU, I cannot use it at the same time as the discrete graphics card. If I do, the combined heat makes the CPU once again go to 800 MHZ.

Expecting to be able to use the hardware my computer was advertised as having is not having unrealistic expectations.

> I'm a little confused at what's "sturdier" about a gaming laptop as well.

In my experience most gaming laptops are made of metal or thick plastic and seem more durable. I actually had a MSI laptop, it was a tank - like the revered Thinkpads, but better. Dropped it from a few feet and it only got a scratch on the surface.

> A gaming laptop is not really a laptop at all by comparison. They still get less than 4 hours of battery life under load, they're an inch or more thick, they're heavy. Many of them do not fit in backpacks. Gaming laptops are essentially designed for plugged in operation.

That's why I got the XPS 15. Had I known that its performance was a fraction of what was advertised I would have ignored this whole category of deceptive laptops and bought a gaming laptop, even though they have all these downsides.

[girvo]

Whereas I am aware of thermal throttling issues across most thin-and-light laptops, and kept that in mind when purchasing it. I was not defrauded. Perhaps though, I'm more sensitive to it, as I used to write my own throttling scripts for my weird AMD APU "netbook" years ago.

[dangus]

Just to clarify here, all laptops of reasonably small size throttle. Throttling below base clock is the only problem, and it was only a temporary one that was patched out with new firmware on 2018 MacBook Pros.

[qplex]

I guess the reasonable size is relative.

None of my laptops do this.

The thermal design is basically broken if the CPU overheats and has to throttle to prevent a fire.

[BubRoss]

That's basically modern CPUs for the modern consumer. Laptops often have cooling that can't keep up, same with intel NUCs. Same even with ryzens with integrated video cards. There are bios setting for relaxed thermal throttling that let it get hotter that are a huge boost to performance.

[maigret]

Tell Tesla then, they cannot make a full fast lap of Nurburgring.

Actually, most cars can break so much so often before they lose performance because of overheating or even burning brakes.

[rowanG077]

Did you actually try to run the Pi as a desktop replacement? That is what they are advertising. A desktop replacement that is useable for a couple of minutes and then throttles is not something you want.