I think that heatsink actually helped a lot. Without it the Pi probably throttled to keep the temperature in check, so while the heatsink only improved the temperature by 2C, the performance might have increased a lot.
I love your videos. I have been working with computers since I was a kid and discovered the pi about a year ago. Your videos have been a huge help in learning everything I needed to know.
Wow, I just bought my first Raspberry Pi 3 yesterday to use in a project which would involve cramming it in a small case with no airflow. I forgot to get a heatsink and figured it would be fine. I'm going to order one right now, thanks so much for this video!
Absolutely Essential Viewing for any Pi owner. Thank you enormously, Chris, for your time and diligence. Certainly, I think, active cooling is now a must and I would love to see another update to this video, *please*! Even though I say it a lot (because it matters) thanks for your high quality content, creation and production values. I would seriously love to see you on the tv. I hope you and yours have a great week!
Another excellent Raspberry Pi related video. These tests were very interesting and also very useful to know. I intend doing some projects on my Raspberry Pi 3 and keeping the CPU temperature as low as possible is essential as it will effect performance. I would love to see some results with some active cooling such as adding a fan on top of the heat sink, the passive cooling did very little to reduce the heat it would seem. Kind of reminds me of when the Sinclair ZX Spectrum 128K came out and it had that huge heat sink on it's right hand side which did get hot also. Looking forward to the next tests.
Great instructional video! Most other videos I watched , on other channels, were really vague about what the cord actually was and flashed it on screen pre-wrote for a split second.
Thanks for the heads-up! About a week ago, Peter Oakes published a video on UA-cam, called "Using a FLIR AX8 to test a PI 3 full load temperature". His conclusion was that the internal temperature sensor isn't very accurate. In his case the actual SoC temperature was something like 30C higher than indicated by vcgencmd measure_temp. So, adding a heatsink (and fan) might be even more important for a heavily loaded Pi3.
+Kees Nuyt If that is the case, I would be happy to lend Chris and Infra-red thermometer gun which has an accuracy of 0.1C. Then it would be able to determine just how accurate the temp sensor on the SOC is. ;-)
Yes please, I can only imagine what one would look like. I guess you'd use a couple of the GPIO pins to turn the fan on when it got to a certain temperature.
+ExplainingComputers here's a follow up on how good your work on this is indaparliament.blogspot.com/2016/05/pi3-raspberry-pi-3-cpu-heat-sink-test.html And
pokies100 You have not installed sysbench! :) You need to be online, and then open a terminal and type the command "sudo apt-get install sysbench" (do not enter the quote marks). Then all will run OK. :)
Great video, however there is some more information I would love to see in these tests! I would love to see the time it takes to complete the CPU test. Using time to lunch the test would display the run time quite well. As the CPU throttling takes place due to heat triggers being surpassed it would be great to know what kind of performance hit we see. Doing that with no heatsink, heatsink, and heatsink with active cooling would provide all sorts of useful information about how I would build out a raspberry pi system. Also measuring power consumption during the tests would be pretty great too so I can see how much more power the CPU can consume when heat dissipation goes up and throttling happens at different performance levels because of that dissipation. Thanks again! Looking forward to your next videos.
Very interesting! Thank you for the information. Here in my country my raspberry pi overheats so much that it freezes right from the start (room temp 32~34ºC and humidity around 70%). I'll going to buy another Pi and put a heatsink on it. Thank you again.
+Marcio Maia let me guess..... do you are in brazil? here in my country (panama) is a bit colder (25-30°C), sometimes up to 35°C and humidity from 60-70% up to over 90%.... my router and modem are overheating ALL THE TIME, so i think i will need a delta fan to solve that without A/C....
An active cooling video would be great. Very interested in what the db level, and over all noise effect, would be for a fan on the RPI. Thanks for all you do, very interesting stuff!
Запуск CPU (и GPU) температурных тестов на Raspberry Pi 3, а затем установка радиатора и тестирования снова! Включает некоторых скриптах bash.Вы можете смотреть мне подходят вентилятор и попробовать активное охлаждение в этом последующем видео: www.youtube.com/watch?v=5Ud-g...И подходит очень большой радиатор в этом видео: www.youtube.com/watch?v=1AYGn...Вы также можете, как мой” Малина Pi 3: Обзор & скорость тесты " видео: www.youtube.com/watch?v=T_WlF...Некоторые из моих других Raspberry Pi видео включают в себя этот, где я провожу неделю, используя Raspberry Pi как мой единственный компьютер: www.youtube.com/watch?v=7QH6R...И этот обзор Малины Pi ноль: www.youtube.com/watch?v=WJdQ4...Больше видео на вычислять-родственные темы можно найти на: ua-cam.com/users/explainingcomp...Вам также может понравиться мой канал ExplainingTheFuture по адресу: ua-cam.com/users/explainingthef...
Thanks Chris. A nice video as usual but what about power consumption of the Pi 3 under different loads? Hope to see a follow up on this video to cover the power aspect.
excellent video as always! I have a theory about why the heatsink didn't show a significant decrease in temperature. the purpose of a heatsink is to increase the exposed area of something with a good heat conductive material to allow a higher heat transfer with the surrounding environment. the heatsink you used for the test is a really tiny heatsink that almost make no difference for the CPU. besides, it seems that the glue used to fit the heatsink on the CPU wasn't the right jelly one that has a very high thermal conductivity. I recommend a bigger heatsink with taller blades (maybe see the one used on the odroid c1) to effectively increase the exposed area and paying attention to the glue, to get a better and more satisfactory result on the thermal test. anyway, thank you very much for your time and knowledge!
Very interesting. A thing to consider though is that the system on a chip will slow throttle at more than 82° regardless if there is a cooler or not. Cooling shouldn't have an effect on temperature at full charge but rather on execution time of the test. I like your Raspberry Pi benchmark videos. Keep it up.
+Guillaume Laurent Thanks. Though ideally a cooler will reduce temperature at full processor utilization to below the throttling threshold. Look out for this Sunday's video! :)
Good instructions! Thanks. I tested my new (first) PI4 with it. Passive cooling "Amor Case" (CPU contact surface polished) with EC 360 gold thermal pad. At no time more than 56 °C. Nearly perfect
another great video I'd love to see how the pi temperature differs with active cooling. I use my pi3 for the retropie software and it seems to stay rather cool even in a case with high graphic games running such an amazing little machine
It's mostly a shut off thing or losing the RPi eventually. I like my new RPi 3B department with a bunch of terminal commands and DIY and tinkering. The worst is to block the holes and it gives the temp/heat warning. A heat sink can also go to a limit. Just take care of it. With the many USB ports and pins we can mount a fan from somewhere and run scripts on various sensors.
It looks like you've already decided to make the active-cooling video, but it gets my vote too. Other commentators have already said that thermal paste, and maybe a slightly larger heatsink, would probably yield better cooling. The more surface area there is to dissipate heat to the air, the better, and the paste should be a better interface than the tape.
+MarkTheMorose Yes, the interest in the active cooling video was clear -- so it will upload on Sunday! :) As you asy, bigger heatsinks and better heat transfer to them would improve things -- but I wanted to start with a product on the market. But Sunday's video is very much DIY cooling!
@ExplainingComputers please experiment with fans and possibly a networked parallel processing configuration. ie. two raspberry's connected and running a task
I tested a similar heatsink and got similar results (only a few degrees C at best). This was all I expected, because these days surface mount PCBs like the Pi derive most their passive cooling through thermal coupling to the large PCB ground and power planes, so these small stick-on heatsinks are only minimally effective. You can get as much (or more) improvement in cooling for free, by simply turning the Pi 90 degrees on edge, which improves the natural convective cooling over the surfaces of the PCB.
+ExplainingComputers Yep, this is also why your fan works so well - because it cools the whole surface of the board, nicely cooling down the internal copper ground and power planes that the Pi SOC and RAM are soldered to. Of course, a passive heatsink does help a little, even without a fan, and this advantage also improves if we turn the Pi 90 degrees so convection can better flow air upwards over the heatsink - and this would probably be fine for most folks. For a nice sealed case passive cooling solution, I have found nothing that comes even close to the FLIRC raspberry Pi case. This case was originally intended for the 4 USB port 512k Pi B but fits the newer P2 just fine. The case includes an aluminum thermal link pedestal which couples the outer aluminum case with a thermal pad directly on top of the Pi CPU SOC. This gives very good heat coupling out to the sides of the aluminum case, which has plenty of surface area for cooling. With the standard Pi, heat flows out the bottom of the chip into the circuit boards large ground and power planes, and the FLIRC case arrangement basically doubles the heat flow out of the chip by also coupling the TOP of the SOC chip to a large external metal surface. When installing a Pi 2 in the FLIRC case, be sure to check the alignment of the thermal pad, because the case seems to have bee designed to also support the earlier Pi where the CPU was slightly offset from the newer Pi models. For best alignment with the Pi 2 I had to offset the thermal pad slightly to the upper left corner of the square thermal pedestal so that it would be centered on top of the Pi 2's CPU when assembled. The FLIRC case, because of its efficent thermal coupling to the top of the SOC, does better than any of the other heatsinks I have seen for the Pi (except perhaps those also using a fan), and it just flat out looks BEAUTIFUL The only hesitation I would have in recommending the FLIRC case concerns the Pi 3's. Although the FLIRC case will work with the PI 3 there could be issues with the WiFi, because of it's metal construction. There is some hope for the WiFi and Bluetooth because the side of the FLIRC case where the Pi 3's ceramic antenna is located has a plastic insert which should allow WiFi and Bluetooth RF to exit, but there are other issues with the Pi 3 though with this case - such as needing to create a new opening for viewing the LEDs, and having to either enlarged the opening for the the MicroSD slot, or attach a tab of thin tape to each MicroSD card so you can grab it and pull it out. These became issues when the Pi foundation (in its infinite wisdom) changed the LED locations on the Pi3 vs the Pi and Pi2, and also changed from a push-to-release, to a non-spring loaded pull-to-remove MicroSD type socket. ... but for the Pi 2 non of these issues arise, and, seriously, this case is just flat out BEAUTIFUL. So, if you aren't hacking your Pi's GPIO every ten minutes, and just want a nice clean fully enclosed case with excellent thermal performance, you can't beat the FLIRC case. And nothing I have found comes even close, when it comes to making the Pi look like a solid industrial, real professional Micro-PC.
Great video, and my pi 3b was running hot for some reason it's is pretty cold where I am. It was giving me the thermometer warning during regular use. So I checked the temps at idle. The pi ran at about 45.5 c average. So I decided to get a heatsink from a old soundcard I had, and cut it so that it was fit on the pi. Then I applied thermal paste, and my pi was at 36 c average.
+clangerbasher Agreed ! Would love to see some company make a SUPER small fan for the Pi. Like a fan/heatsink that has the same size profile as the heatsink shown in this vid. Would be comically small but awesome !
GruntW0rk I think there are a couple of cases out there that are also heat sinks. It would be nice to see a proper little copper one with vanes etc. BTW the Pi isn't small, we are just big!!!!!!!! :) ;)
+clangerbasher Intel has released a small x86 computer known as the "Compute Stick"; and it has an adorable little heat-sink and fan :3 www.legitreviews.com/wp-content/uploads/2015/04/intel-compute-stick-fan.jpg
An active cooling test would be great, especially if there was a variety of active cooling options to compare against each other, i.e. comparing between different active cooling options, not just active vs passive in general.
Thank you for this informative video, though, I do have a suggestion (which should not be mistaken as criticism, just something I personally would have liked to see), it would be more like informative and maybe even more accurate if you would've put a second terminal where you ran the command: "watch -n1 vcgencmd measure_temp" which will provide the temperature every second forever, and play the video while overlaying this terminal. Then you would get somewhat real time temperature info, not seconds, milli, micro or nano after the video stopped playing (which could theoretically make a difference in temperature, though the difference you would see with the inaccurate sensor is indeed neglect-able), it would at least provide some visual feedback (and I personally find that rather pleasing x) ).
I would be curious to see active cooling results compared to no cooling and passive cooling. I currently use a heatsink on my media center Pi and heatsink+fan on my web server Pi.
Love your videos, used this series as a guide for adding my own heatsinks and a fan along with the associated testing. My question though, is about your test script. I know it's almost totally a matter of style, but why do you bother to make your .sh script executable rather than just running it from the sh command?
Thanks for your kind feedback. On the script, it is just a personal preference -- I guess I am used to the old DOS days of files that execute with no additional syntax, hence go that route. :)
Professor, good day and great video. Hypothetically, could it be possible to apply thermal paste between the proc and the tiny heatsink? Maybe a tiny drop? This video was very interesting to me because made me wonder how creative people could get (for example, a very tiny water cooling solution, or custom made copper heatsinks and exhausts) in order to take the Pi3 to its limits. Already a very flexible and cheap computing solution already.
+Luis Mercado I think thermal paste will be more than hypothetical in my forthcoming tests. :) I had not considered water cooling! I think there will be two follow up videos now . . .
The video test wasn't very comprehensive. I saw your videos were 25fps. The situation would be different if playing 60fps. I assume those videos were encoded using h.264 codec? The CPU test was pretty straight forward. Thanks for the tests. I'll plan my RPi3 cooling solutions based on your numbers. The rest of this comment goes bit more over the normal use-case video playing but: The situation would be a whole lot different if the video codec was something the gpu cannot decode (VP9 for example). Although everyone should re-encode their videos to a codec that RPi3 GPU can decode. Then there's all the possible layers that can display the video. Of course that's not dependent on the video being played so the lightest one should be preferred (if possible). I assume the program you used to play the video used fully GPU accelerated layer to display the video however.
how can i adjust my usb mouse sensibility in linux..? (retropie emulator)..? is emulation considered like stressing the cpu..? thx for your videos they are so helpful..!
I just used thermal paste , no tape , and the flat copper heat sinks and so far they do nicely and the paste holds them on ok , not that I transport it often
Great info! I am in a hot area where ambient is in low 90s F. My pi was running in the mid 40s idling, and a little higher with medium tasking. That was with no case and heat sink. put in case and jumped up to high fifties on idle, 60s with lite load. That is with the covers off the case. My plan was to make a couple of pis for my children's study area. Just attach to back of 14 inch monitor with keyboard and mouse. I will have to rethink and set up active cooling. Now need to consider noise from up to four fans running in study area. Would like to see more tests with active cooling please. bob
even with a fan, my cpu underload got to the high 60's. This new cpu while more powerful than the previous cpu gets way hotter. Still, there are a handful of cases with fans out there now that will do you well
You might run a whetstone (FP performance) or Dhrystone (Integer erformance) test, and run several copies simultaneously to exercise all of the processors.
Great video I have a case with heatsinks and a fan with kodi while being connected to vpn and get 50.5 c after one hour of usage so I am within the safe temperature I see thanks!
ive seen ur cooling vids etc. . .nice . .will u be making a overclocking vid. . . i try to follow wot eta prime duz but i have no luck . . i dont have a config follder showing like in his vid. . .im not nuts on overclocking, i just want to run 1250Cpu- 1275Cpu & 425Gpu . .a mild tweek + run it cooler
Although the action of the heat sink seems to be minimal, what wasn't clear from this test is how much more work the CPU was doing. I agree that the 82° test looks like it was throttling. It may have been doing significantly less work at that point due to this. It would be very interesting to quantify this.
+albedo0point39 It continued to show 100 per cent utlilization in the monitor, as would be expected in he Sysbench test. So I think it was still running at full pelt, just more slowly.
Active Cooling test please. Also how about connecting the fan to the GPIO pins so the fan can be throttled by temperature so that power is saved when the fan is not needed.
I'd love to see an active cooling solution for a Rasperry Pi! I've often wondered if there were fans that were made a certain size with the Pi in mind, and if it would run off of the GPIO pins. I've been wanting to use my Pi 2 as a NIDS, using an old Cisco 2950 with Port Mirroring turned on to copy all traffic to it, but I'm sure that with all my systems running, that would be a very big task for the little Pi, and I'm worried about how hot it may get while doing that. I've also considered running a small Minecraft server using a Pi, so that it could be up 24/7 without using a lot of power, but again, I'm worried about the CPU usage that Minecraft would require, and wonder if that would need active cooling.
I've got a Xbox Kinect camera working with an RPi3 but the temp gets very high very quickly (around 90 degrees) and I'm a little worried I might be causing damage. I'd love to see some guidance on active cooling and also more importantly whether its worth it ?
I was finally able to stress test my raspberry pi 3. It seems that I somehow missed a couple of lines of code before running the test as I had followed exactly what you did in this great video tutorial. I forgot the "chmod +x (filename)" and the "./" lines before running the test. My Raspberry PI 3 is enclosed in an acrylic case with two heat sinks and a little fan on top of the main CPU. The ambient temperature in the room was 76* and during the test the Raspberry never got over 42.9*C. Do you think that I could over clock the CPU after such good results from the stress test? Thank you so very much for your cool videos. I'm learning a lot about not only my Raspberry, but also about code writing and computers in general. I thought I knew a lot about computers until I started watching your video tutorials. Now I know how dumb I really am about computers. Thanks again. Rusty (Wyatt Earp).
Pls do a active cooling test ! In theory , active cooling should be much more effective as the heat gets drawn away by moving air quickly instead of being slowly transferring into the surrounding 'stagnant' air when using a heatsink.
Active cooling is a no go in my use case, but maybe a bigger heatsink might work. Maybe a bigger heatsink made of copper will do the job. And thanks for the test.
Thanks for the video, but I would like to see you retest with heatsinks on the CPU, ethernet/usb controller and the GPU (on the bottom) and using (additional) thermal cooling adhesive. I'm curious as to what would be the results with the best possible cooling using heatsinks only. Cheers.
+Dave The chip on the bottom is the memory -- the GPU is integrated with the CPU within the SoC that I've stuck the heat sink on. :) I've yet to try heat sinks on the memory and controller chips -- though I don't think they have temperature sensors(?).
Right, I guess additional heatsinks wouldn't make a difference for this test. Re the GPU, I was following the steps for cooling on github: github.com/retropie/retropie-setup/wiki/Overclocking I guess they have it wrong then. I have some heatsinks for my pi I'm considering on putting on, not sure if it would be worthwhile waiting to apply some additional thermal paste or not, hence why I asked. Would be interested in seeing what a difference a small fan would make also. Anyway, thanks for the video!
Dave I have posted another video in which I added a fan, and another with a much larger heatsink. :) There is no doubt the GPU is on the SoC -- that's why it is a system on a chip!
I think that heatsink actually helped a lot. Without it the Pi probably throttled to keep the temperature in check, so while the heatsink only improved the temperature by 2C, the performance might have increased a lot.
I love your videos. I have been working with computers since I was a kid and discovered the pi about a year ago. Your videos have been a huge help in learning everything I needed to know.
Thank you for the walkthrough. Managed to replicate your active cooling setup and got my CPU temps down in the 37-38*C range under full load!
Excellent. :)
Wow, I just bought my first Raspberry Pi 3 yesterday to use in a project which would involve cramming it in a small case with no airflow. I forgot to get a heatsink and figured it would be fine. I'm going to order one right now, thanks so much for this video!
Absolutely Essential Viewing for any Pi owner. Thank you enormously, Chris, for your time and diligence. Certainly, I think, active cooling is now a must and I would love to see another update to this video, *please*!
Even though I say it a lot (because it matters) thanks for your high quality content, creation and production values. I would seriously love to see you on the tv. I hope you and yours have a great week!
+Kevin H Thanks Kevin.
Another excellent Raspberry Pi related video. These tests were very interesting and also very useful to know. I intend doing some projects on my Raspberry Pi 3 and keeping the CPU temperature as low as possible is essential as it will effect performance.
I would love to see some results with some active cooling such as adding a fan on top of the heat sink, the passive cooling did very little to reduce the heat it would seem. Kind of reminds me of when the Sinclair ZX Spectrum 128K came out and it had that huge heat sink on it's right hand side which did get hot also.
Looking forward to the next tests.
Very useful as always. Love your production. Got myself rasbery pi 3 two weeks ago, will try to build quadracopter (y)
i swear explaining computers is one of the best Channels on youtube
:)
Outstanding presentation and approach to the testing. Best video I've seen for a while. Thanks!
Thanks! :) I am uploading a video trying all kinds of Pi 3 B+ cooling solutions a week on Sunday.
I want to see the active cooling and much more Pi related. Thanks!!
Great instructional video! Most other videos I watched , on other channels, were really vague about what the cord actually was and flashed it on screen pre-wrote for a split second.
Thanks for the heads-up! About a week ago, Peter Oakes published a video on UA-cam, called "Using a FLIR AX8 to test a PI 3 full load temperature". His conclusion was that the internal temperature sensor isn't very accurate. In his case the actual SoC temperature was something like 30C higher than indicated by vcgencmd measure_temp. So, adding a heatsink (and fan) might be even more important for a heavily loaded Pi3.
+Kees Nuyt If that is the case, I would be happy to lend Chris and Infra-red thermometer gun which has an accuracy of 0.1C. Then it would be able to determine just how accurate the temp sensor on the SOC is. ;-)
Yes please, I can only imagine what one would look like. I guess you'd use a couple of the GPIO pins to turn the fan on when it got to a certain temperature.
What a lovely video to intriduce bash script. Well done. Very good video to introduce all a pi3 can do with maximum range of useful commands.
+pokies100 Many thanks. :)
+ExplainingComputers here's a follow up on how good your work on this is
indaparliament.blogspot.com/2016/05/pi3-raspberry-pi-3-cpu-heat-sink-test.html
And
pokies100 Briliant! :) Many thanks.
+ExplainingComputers I got "bash: sysbench: command not found
So new to this do not know what to do...
pokies100 You have not installed sysbench! :) You need to be online, and then open a terminal and type the command "sudo apt-get install sysbench" (do not enter the quote marks). Then all will run OK. :)
I would be very interested in active cooling, great video by the way,
damn I love this channel
Awesome, isn't it? :0)
a trick: watch movies on Kaldrostream. Me and my gf have been using them for watching a lot of movies during the lockdown.
@Micah Kairo Yup, I have been watching on Kaldrostream for since november myself =)
You are always superb with always new topic on technology
Yes, looking fwd to the active cooling solution video. Thanks.
Yes. Do active cooling video please.
Great video, however there is some more information I would love to see in these tests! I would love to see the time it takes to complete the CPU test. Using time to lunch the test would display the run time quite well. As the CPU throttling takes place due to heat triggers being surpassed it would be great to know what kind of performance hit we see. Doing that with no heatsink, heatsink, and heatsink with active cooling would provide all sorts of useful information about how I would build out a raspberry pi system. Also measuring power consumption during the tests would be pretty great too so I can see how much more power the CPU can consume when heat dissipation goes up and throttling happens at different performance levels because of that dissipation. Thanks again! Looking forward to your next videos.
Thanks for the video tutorial. I'm learning a lot as I go.
Very interesting! Thank you for the information. Here in my country my raspberry pi overheats so much that it freezes right from the start (room temp 32~34ºC and humidity around 70%). I'll going to buy another Pi and put a heatsink on it. Thank you again.
+Marcio Maia Very interesting. I think I will experiment with fans . . .
+Marcio Maia let me guess..... do you are in brazil? here in my country (panama) is a bit colder (25-30°C), sometimes up to 35°C and humidity from 60-70% up to over 90%.... my router and modem are overheating ALL THE TIME, so i think i will need a delta fan to solve that without A/C....
An active cooling video would be great. Very interested in what the db level, and over all noise effect, would be for a fan on the RPI. Thanks for all you do, very interesting stuff!
Thanks . Always worth watching .
Запуск CPU (и GPU) температурных тестов на Raspberry Pi 3, а затем установка радиатора и тестирования снова! Включает некоторых скриптах bash.Вы можете смотреть мне подходят вентилятор и попробовать активное охлаждение в этом последующем видео: www.youtube.com/watch?v=5Ud-g...И подходит очень большой радиатор в этом видео:
www.youtube.com/watch?v=1AYGn...Вы также можете, как мой” Малина Pi 3: Обзор & скорость тесты " видео: www.youtube.com/watch?v=T_WlF...Некоторые из моих других Raspberry Pi видео включают в себя этот, где я провожу неделю, используя Raspberry Pi как мой единственный компьютер:
www.youtube.com/watch?v=7QH6R...И этот обзор Малины Pi ноль:
www.youtube.com/watch?v=WJdQ4...Больше видео на вычислять-родственные темы можно найти на: ua-cam.com/users/explainingcomp...Вам также может понравиться мой канал ExplainingTheFuture по адресу: ua-cam.com/users/explainingthef...
YES! For the active cooling test!
Informative, enjoyed the testing of the cpu
Can this little thing handle to make a broadcast?
As usual, great video Chris. yes please do an active cooling video on the P i3, thanks..
Thanks Chris. A nice video as usual but what about power consumption of the Pi 3 under different loads? Hope to see a follow up on this video to cover the power aspect.
Would love to see the active cooling options. I've not seen many options for active cooling so will be interesting to see what you come up with!
+LJH08Ralith My final solution is quite a practical bit of diy. :)
+ExplainingComputers can't wait to see it then !!
Great video.. Would love to see an active cooling test, especially since I may pick one up myself to use as a hi-def media center!!
excellent video as always! I have a theory about why the heatsink didn't show a significant decrease in temperature. the purpose of a heatsink is to increase the exposed area of something with a good heat conductive material to allow a higher heat transfer with the surrounding environment. the heatsink you used for the test is a really tiny heatsink that almost make no difference for the CPU. besides, it seems that the glue used to fit the heatsink on the CPU wasn't the right jelly one that has a very high thermal conductivity. I recommend a bigger heatsink with taller blades (maybe see the one used on the odroid c1) to effectively increase the exposed area and paying attention to the glue, to get a better and more satisfactory result on the thermal test. anyway, thank you very much for your time and knowledge!
Very interesting. A thing to consider though is that the system on a chip will slow throttle at more than 82° regardless if there is a cooler or not. Cooling shouldn't have an effect on temperature at full charge but rather on execution time of the test.
I like your Raspberry Pi benchmark videos. Keep it up.
+Guillaume Laurent Thanks. Though ideally a cooler will reduce temperature at full processor utilization to below the throttling threshold. Look out for this Sunday's video! :)
+ExplainingComputers Oh you were way ahead of me weren't you?
Guillaume Laurent :)
Good instructions! Thanks. I tested my new (first) PI4 with it. Passive cooling "Amor Case" (CPU contact surface polished) with EC 360 gold thermal pad. At no time more than 56 °C. Nearly perfect
That is a good case.
Love your videos - thanks very much
thank you for the wisdom ! Sensei
another great video I'd love to see how the pi temperature differs with active cooling.
I use my pi3 for the retropie software and it seems to stay rather cool even in a case with high graphic games running such an amazing little machine
Yes! I'm a fan of the fan video. :)
+Infinite loop Nice. :)
Thank you, that's what I was looking for! :)
Active cooling, yes please Chris.
It's mostly a shut off thing or losing the RPi eventually. I like my new RPi 3B department with a bunch of terminal commands and DIY and tinkering. The worst is to block the holes and it gives the temp/heat warning. A heat sink can also go to a limit. Just take care of it. With the many USB ports and pins we can mount a fan from somewhere and run scripts on various sensors.
Great video! I would like to see some active cooling action!
Thank you very much enjoyed this one very much
excellent video. I would love to see an active test!
Excellent video, I would love to see a practical review of any active cooling solution you can come up with for the pi3
I quite enjoy your videos. Where can I find the bash script you use in this video?
Yes, active cooling video 😊
That movie clip is robot porn, hardcore.
Thanks for this video, and would like to watch Active Heatsink test on RP3.
Raspberry pi temperature monitor software TWIC :
github.com/gavinlyonsrepo/raspberrypi_tempmon
It looks like you've already decided to make the active-cooling video, but it gets my vote too. Other commentators have already said that thermal paste, and maybe a slightly larger heatsink, would probably yield better cooling. The more surface area there is to dissipate heat to the air, the better, and the paste should be a better interface than the tape.
+MarkTheMorose Yes, the interest in the active cooling video was clear -- so it will upload on Sunday! :) As you asy, bigger heatsinks and better heat transfer to them would improve things -- but I wanted to start with a product on the market. But Sunday's video is very much DIY cooling!
@ExplainingComputers please experiment with fans and possibly a networked parallel processing configuration. ie. two raspberry's connected and running a task
I tested a similar heatsink and got similar results (only a few degrees C at best). This was all I expected, because these days surface mount PCBs like the Pi derive most their passive cooling through thermal coupling to the large PCB ground and power planes, so these small stick-on heatsinks are only minimally effective. You can get as much (or more) improvement in cooling for free, by simply turning the Pi 90 degrees on edge, which improves the natural convective cooling over the surfaces of the PCB.
+Steve B. Very interesting. :)
+ExplainingComputers
Yep, this is also why your fan works so well - because it cools the whole surface of the board, nicely cooling down the internal copper ground and power planes that the Pi SOC and RAM are soldered to.
Of course, a passive heatsink does help a little, even without a fan, and this advantage also improves if we turn the Pi 90 degrees so convection can better flow air upwards over the heatsink - and this would probably be fine for most folks.
For a nice sealed case passive cooling solution, I have found nothing that comes even close to the FLIRC raspberry Pi case. This case was originally intended for the 4 USB port 512k Pi B but fits the newer P2 just fine. The case includes an aluminum thermal link pedestal which couples the outer aluminum case with a thermal pad directly on top of the Pi CPU SOC. This gives very good heat coupling out to the sides of the aluminum case, which has plenty of surface area for cooling. With the standard Pi, heat flows out the bottom of the chip into the circuit boards large ground and power planes, and the FLIRC case arrangement basically doubles the heat flow out of the chip by also coupling the TOP of the SOC chip to a large external metal surface.
When installing a Pi 2 in the FLIRC case, be sure to check the alignment of the thermal pad, because the case seems to have bee designed to also support the earlier Pi where the CPU was slightly offset from the newer Pi models. For best alignment with the Pi 2 I had to offset the thermal pad slightly to the upper left corner of the square thermal pedestal so that it would be centered on top of the Pi 2's CPU when assembled.
The FLIRC case, because of its efficent thermal coupling to the top of the SOC, does better than any of the other heatsinks I have seen for the Pi (except perhaps those also using a fan), and it just flat out looks BEAUTIFUL
The only hesitation I would have in recommending the FLIRC case concerns the Pi 3's. Although the FLIRC case will work with the PI 3 there could be issues with the WiFi, because of it's metal construction. There is some hope for the WiFi and Bluetooth because the side of the FLIRC case where the Pi 3's ceramic antenna is located has a plastic insert which should allow WiFi and Bluetooth RF to exit, but there are other issues with the Pi 3 though with this case - such as needing to create a new opening for viewing the LEDs, and having to either enlarged the opening for the the MicroSD slot, or attach a tab of thin tape to each MicroSD card so you can grab it and pull it out. These became issues when the Pi foundation (in its infinite wisdom) changed the LED locations on the Pi3 vs the Pi and Pi2, and also changed from a push-to-release, to a non-spring loaded pull-to-remove MicroSD type socket.
... but for the Pi 2 non of these issues arise, and, seriously, this case is just flat out BEAUTIFUL.
So, if you aren't hacking your Pi's GPIO every ten minutes, and just want a nice clean fully enclosed case with excellent thermal performance, you can't beat the FLIRC case. And nothing I have found comes even close, when it comes to making the Pi look like a solid industrial, real professional Micro-PC.
Great video, and my pi 3b was running hot for some reason it's is pretty cold where I am. It was giving me the thermometer warning during regular use. So I checked the temps at idle. The pi ran at about 45.5 c average. So I decided to get a heatsink from a old soundcard I had, and cut it so that it was fit on the pi. Then I applied thermal paste, and my pi was at 36 c average.
I've seen a custom liquid cooling unit for the pi on youtube. It was pretty neat, and funny at the same time.
Yes do the active test.
This new intro is so much better
Thanks. Not everybody liked the change! :)
That's usually the case. I certainly appreciated it, it tends to balance out the 70's BBC feel haha
Yes is much much better!And thanks ExplainingComputers you helped me so much with my raspberry pi!!
That is interesting. I wonder what it will mean for the next iteration of Pi?
I would like to see active cooling tested please. :)
+clangerbasher Agreed ! Would love to see some company make a SUPER small fan for the Pi. Like a fan/heatsink that has the same size profile as the heatsink shown in this vid. Would be comically small but awesome !
GruntW0rk I think there are a couple of cases out there that are also heat sinks. It would be nice to see a proper little copper one with vanes etc. BTW the Pi isn't small, we are just big!!!!!!!! :) ;)
+GruntW0rk actually there is fans in that size but they're quite pricey. Google pi micro fan. It's on Amazon for 8 dollars.
+clangerbasher
Intel has released a small x86 computer known as the "Compute Stick"; and it has an adorable little heat-sink and fan :3
www.legitreviews.com/wp-content/uploads/2015/04/intel-compute-stick-fan.jpg
Have they released a Linux one this time?
Very usefull video. I would like to learn more about cooling fans and how to increase overall performance of a raspberry pi 3
There are follow-up videos on that! :)
An active cooling test would be great, especially if there was a variety of active cooling options to compare against each other, i.e. comparing between different active cooling options, not just active vs passive in general.
you look like bill gate in 1980-1990
It is Bill Gates but after he learned how to camouflage his voice as an Englishman!!
I can't help but think of a mix of Bill Gates and Paul McCartney each time I see him on one of his (incredibly instructive) videos...
yes, by all means, please make a raspi 3 active heatsink video.
Thank you for this informative video, though, I do have a suggestion (which should not be mistaken as criticism, just something I personally would have liked to see), it would be more like informative and maybe even more accurate if you would've put a second terminal where you ran the command: "watch -n1 vcgencmd measure_temp" which will provide the temperature every second forever, and play the video while overlaying this terminal. Then you would get somewhat real time temperature info, not seconds, milli, micro or nano after the video stopped playing (which could theoretically make a difference in temperature, though the difference you would see with the inaccurate sensor is indeed neglect-able), it would at least provide some visual feedback (and I personally find that rather pleasing x) ).
Was going to make this exact comment :)
Yes, fan test please.
Thanks.
+Paul Rautenbach I have shot it this morning! :)
Yes, I'd be interested in seeing the results from active cooling the Pi
amazing Video! thanks
I loved this video! Thats how I like explanations :)
More pi please, including cooling. Thanks.
I would be curious to see active cooling results compared to no cooling and passive cooling.
I currently use a heatsink on my media center Pi and heatsink+fan on my web server Pi.
Love your videos, used this series as a guide for adding my own heatsinks and a fan along with the associated testing.
My question though, is about your test script.
I know it's almost totally a matter of style, but why do you bother to make your .sh script executable rather than just running it from the sh command?
Thanks for your kind feedback. On the script, it is just a personal preference -- I guess I am used to the old DOS days of files that execute with no additional syntax, hence go that route. :)
Professor, good day and great video. Hypothetically, could it be possible to apply thermal paste between the proc and the tiny heatsink? Maybe a tiny drop? This video was very interesting to me because made me wonder how creative people could get (for example, a very tiny water cooling solution, or custom made copper heatsinks and exhausts) in order to take the Pi3 to its limits. Already a very flexible and cheap computing solution already.
+Luis Mercado I think thermal paste will be more than hypothetical in my forthcoming tests. :) I had not considered water cooling! I think there will be two follow up videos now . . .
The video test wasn't very comprehensive.
I saw your videos were 25fps. The situation would be different if playing 60fps. I assume those videos were encoded using h.264 codec?
The CPU test was pretty straight forward. Thanks for the tests. I'll plan my RPi3 cooling solutions based on your numbers.
The rest of this comment goes bit more over the normal use-case video playing but:
The situation would be a whole lot different if the video codec was something the gpu cannot decode (VP9 for example). Although everyone should re-encode their videos to a codec that RPi3 GPU can decode.
Then there's all the possible layers that can display the video. Of course that's not dependent on the video being played so the lightest one should be preferred (if possible). I assume the program you used to play the video used fully GPU accelerated layer to display the video however.
Very informative. As I was kind of expecting these results, I´m now confident to not have to order this heatsink with the Pi 3. Thank you.
Very helpful!
how can i adjust my usb mouse sensibility in linux..? (retropie emulator)..? is emulation considered like stressing the cpu..?
thx for your videos they are so helpful..!
I just used thermal paste , no tape , and the flat copper heat sinks and so far they do nicely and the paste holds them on ok , not that I transport it often
Wow! very nice tests as well as good scripting...by the way how can I learn it? thank you very much.:)
very helpful thank you
I would also be interested in seeing a follow up video.
Great video - how do you remove the heatsink from the cpu , when thermal tape is used ???
Theres a gadget for the taskbar which shows the temp... just saying ^^
Great info!
I am in a hot area where ambient is in low 90s F. My pi was running in the mid 40s idling, and a little higher with medium tasking. That was with no case and heat sink. put in case and jumped up to high fifties on idle, 60s with lite load. That is with the covers off the case.
My plan was to make a couple of pis for my children's study area. Just attach to back of 14 inch monitor with keyboard and mouse. I will have to rethink and set up active cooling. Now need to consider noise from up to four fans running in study area. Would like to see more tests with active cooling please.
bob
+Saffron brooks There is clearly an interest in this! I hope to shoot an active cooling video this week and upload next Sunday.
Would have been nice to have a time taken for the cpu stress test. Maybe it was slightly faster with the heatsink as it did not throttle so much ?
even with a fan, my cpu underload got to the high 60's. This new cpu while more powerful than the previous cpu gets way hotter. Still, there are a handful of cases with fans out there now that will do you well
You might run a whetstone (FP performance) or Dhrystone (Integer erformance) test, and run several copies simultaneously to exercise all of the processors.
Great video I have a case with heatsinks and a fan with kodi while being connected to vpn and get 50.5 c after one hour of usage so I am within the safe temperature I see thanks!
ive seen ur cooling vids etc. . .nice . .will u be making a overclocking vid. . . i try to follow wot eta prime duz but i have no luck . . i dont have a config follder showing like in his vid. . .im not nuts on overclocking, i just want to run 1250Cpu- 1275Cpu & 425Gpu . .a mild tweek + run it cooler
Although the action of the heat sink seems to be minimal, what wasn't clear from this test is how much more work the CPU was doing.
I agree that the 82° test looks like it was throttling. It may have been doing significantly less work at that point due to this. It would be very interesting to quantify this.
+albedo0point39 It continued to show 100 per cent utlilization in the monitor, as would be expected in he Sysbench test. So I think it was still running at full pelt, just more slowly.
Active Cooling test please.
Also how about connecting the fan to the GPIO pins so the fan can be throttled by temperature so that power is saved when the fan is not needed.
I'd love to see an active cooling solution for a Rasperry Pi! I've often wondered if there were fans that were made a certain size with the Pi in mind, and if it would run off of the GPIO pins. I've been wanting to use my Pi 2 as a NIDS, using an old Cisco 2950 with Port Mirroring turned on to copy all traffic to it, but I'm sure that with all my systems running, that would be a very big task for the little Pi, and I'm worried about how hot it may get while doing that. I've also considered running a small Minecraft server using a Pi, so that it could be up 24/7 without using a lot of power, but again, I'm worried about the CPU usage that Minecraft would require, and wonder if that would need active cooling.
I'm thinking that too with my MC Server.
explaining computers dot kom 👍
I've got a Xbox Kinect camera working with an RPi3 but the temp gets very high very quickly (around 90 degrees) and I'm a little worried I might be causing damage. I'd love to see some guidance on active cooling and also more importantly whether its worth it ?
+Kevin Hainsworth I now hope to post an active cooling video next Sunday . . .
I was finally able to stress test my raspberry pi 3. It seems that I somehow missed a couple of lines of code before running the test as I had followed exactly what you did in this great video tutorial. I forgot the "chmod +x (filename)" and the "./" lines before running the test. My Raspberry PI 3 is enclosed in an acrylic case with two heat sinks and a little fan on top of the main CPU. The ambient temperature in the room was 76* and during the test the Raspberry never got over 42.9*C. Do you think that I could over clock the CPU after such good results from the stress test? Thank you so very much for your cool videos. I'm learning a lot about not only my Raspberry, but also about code writing and computers in general. I thought I knew a lot about computers until I started watching your video tutorials. Now I know how dumb I really am about computers. Thanks again. Rusty (Wyatt Earp).
I'm glad things are progressing well. And I think you are safe at that temperature to try overclocking.
Thanks, Chris, for your prompt reply. I may try overclocking it and see what happens.
Pls do a active cooling test ! In theory , active cooling should be much more effective as the heat gets drawn away by moving air quickly instead of being slowly transferring into the surrounding 'stagnant' air when using a heatsink.
+VinsG1379ITtronics Now shot -- will upload on Sunday.
Active cooling is a no go in my use case, but maybe a bigger heatsink might work.
Maybe a bigger heatsink made of copper will do the job.
And thanks for the test.
See my 'Raspberry Pi 3: Extreme Passive Cooling' video in which I fit a much larger heatsink, which works very well. :)
Thanks for the information.
Thanks for the video, but I would like to see you retest with heatsinks on the CPU, ethernet/usb controller and the GPU (on the bottom) and using (additional) thermal cooling adhesive. I'm curious as to what would be the results with the best possible cooling using heatsinks only. Cheers.
+Dave The chip on the bottom is the memory -- the GPU is integrated with the CPU within the SoC that I've stuck the heat sink on. :) I've yet to try heat sinks on the memory and controller chips -- though I don't think they have temperature sensors(?).
Right, I guess additional heatsinks wouldn't make a difference for this test. Re the GPU, I was following the steps for cooling on github: github.com/retropie/retropie-setup/wiki/Overclocking I guess they have it wrong then. I have some heatsinks for my pi I'm considering on putting on, not sure if it would be worthwhile waiting to apply some additional thermal paste or not, hence why I asked. Would be interested in seeing what a difference a small fan would make also. Anyway, thanks for the video!
Dave
I have posted another video in which I added a fan, and another with a much larger heatsink. :) There is no doubt the GPU is on the SoC -- that's why it is a system on a chip!
See how hot you can make it get before it fails beyond repair! Active cooling would be good too.
It would be great if you could teach us how the pi can turn on the active cooling only when the temperature exceed a certain value....
Which is better copper or aluminium?