Actually, in this benchmark, Due uses more instantaneous power but less total energy than Mega, because Due is ~7.615x faster than Mega, and the Mega, therefore, uses ~6.87x more energy than Due.
Great! We want to buy a fast arduino board. Arduino DUE is retired and Arduino Mega 2560 not... So I thought Mega 2560 is probably a best choice but your video show me the opposite! Thanks
good video. I made another measurement: computation of position of moon and sun, 1000 times. TIme for due: 4.1 sec, for 2560 12.9 sec. This is a factor of 3.15 (many , many trigonometric formulas, many doubles)
Thank you so much for your video. I was been asking myself what's a difference between them, and it's pretty good tell - tell videos for idiots like me to see a real difference between them.
You cant compare it this way only. there are many differences in the way things work. f.ex.modulus works muuch more faster on arm that on avr per mhz of clock speed.
Yes of course. You are right. The architecture of an ARM processor is completely different. I am only comparing the for a quick visual represententation of the speed diference. It is not a complete benchmark, just to get an idea of how much faster Arduino Due is.
Good video, BUT! If we now place a DUE on a RAMPS 1.4 board Would a 3D printer work better with it? (I ask this because I shall than immediately order a DUE for my 3D printer!) Friendly greetings from The Netherlands! Rob.
what are you talking about ? Of course arduino DUE is faster it has a 32 bit ARM cpu VS 8 bit cpu of Mega .....and much more RAM for the stack and heap ....need to make a test ?
Let's say 100 W max. 100 folds more power, but 2000 folds faster! The ratio of performance to power for PC is much higher. Don't get me wrong though, I know they are two different things. I'm just saying. I love Arduino.
@@pomegranatechannel You forgot the display aka monitor and the power used to start the PC and that both systems are got shout down after all is done to be equal. If they are considered as a one purpose system of course.
Educ8s.com I agree with him. The algorithm used to compute the value of pi must have been logically flawed. The algorithm having the least complexity is the BPP formula. Which happens to be so simple that even handheld calculators can compute the few 9 digits of pi in a fraction of a second. Probably running on clock speeds of less than 1 MHz, graphing calculators use 6 MHz or less. I don't see computational limitations to be the cause of the error. You can calculate the first 15 digits of pi in less than 10,000 iterations.
greetings, your videos are great, I hope I can sorting the conflict that I have with the drivers of the Arduino Mega 2560 is a clone but not detected and windows (7.8) or Linux, only mac osx but when I reboot the computer has trouble detect it. thanks
Hi Nick, I tried duplicating your project but keep coming up with the compile error "undefined reference to pi", which I am presuming results from me not having the contents of what appears to be the second tab "pi(??) - can't read it properly in your video"? So how do I obtain the pi array data used in your ide and how do I create the second tab and install this data in it? Greatly appreciate your help in this regard - unfortunately arduino coding/programming is still a big mystery for me! Keep up the great work and Seasons Greetings for you and family. Cheers, Ian
Hello Ian, thanks! Seasons Greetings for you and your loved ones! You are right, in the same folder that you have placed the Arduino Sketch you need a .c file which contains the binary data for the splash screen icon. It is named pi.c and you can find it on the website that I share the code. Press complile and it will work! Cheers Nick
Well 1.14159 rounded up is 1.1416, which is similar to the 3.1416059. The small error is due to the insufficient number of illiterations. If the test involves more illiterations the time required will be exponentially longer but the result only marginally more accurate so in this benchmark there is no reason for it
talha saglam Educ8s.com hi you are explaining so nice thank you for that.i wonder which one is faster between arduino due and cortex m0?Which one ARM is the arduino due used?i mean arduino using cortex m3 or cortex mX which one?
Celal Afşar Arduino DUE runs an ARM processor at 84Mhz. Arduino Zero Pro which has an ARM m0 runs at 48Mhz. So I expect that Arduino DUE is at least 2 times faster from M0. But of course we have to compare them in real life scenario to see.
Thank you for your kind and nice message.i wonder what do you thinking about the programming M0 and arduino due, which one is easier?İs there a big difference about programming?and if you want to do a complex robot which one will be your choice?
Celal Afşar If I were you, I would use the DUE. It is faster, it is cheaper and it is more time to market. That means that there is more software available for it, and more shield would work! Good luck!
Yes, I forgot to mention it. It is this graphics library that I was using in the previous videos as well: www.henningkarlsen.com/electronics/library.php?id=47
+RPdigital it's more than that, 32-bit uses less cycles to compute than a 8-bit, plus other innumerable CPU capabilities (... not to discuss here), more than just the frequency.
+RPdigital in fact if you use DSP capabilities, you may compute numerous advanced operations per cycle... even more: programming the DMA controller performs realtime operations without using CPU overhead almost like another CPU, but I don't know if Arduino takes advantages of those (...and much more!!!) advanced evil capabilities. I'm in the PIC32 world, but suppose 32-bit AVR (arduino DUE) is like that also. [my english sucks]
sssserya1 Yes I plan to do that. Performance of ATMEGA328 is about 1Mips per MHz. So, increasing the frequency to 20Mhz from 16MHz we have 25% increase in performance. So I expect that benchmark to complete in around 5.5 sec. But we will see.
+ComputerIT This algorithm for a low number of iterations does not provide a good estimate of Pi. It was run for 100.000 iterations in this example. For better results, it needs at least 1.000.000 iterations.
OK all car card sets and tests are useless as there are different engines build in.
5 років тому+1
That is why it is interesting. One is much more expensive and some projects do not need that extra power. It is good to know what real-life performance gains are. As RPdigital noted, if you just look at the specs you might assume that the performance difference would be around 5.25x but in the test, the performance difference is closer to 7.5x.
Actually, in this benchmark, Due uses more instantaneous power but less total energy than Mega, because Due is ~7.615x faster than Mega, and the Mega, therefore, uses ~6.87x more energy than Due.
We always prefer faster boards anyway
ps.: the time for an UNO was 21.7 sec. same sketch.
Excelente. Muchas gracias, justamente estaba buscando una comparación como la que hiciste
Great!
We want to buy a fast arduino board. Arduino DUE is retired and Arduino Mega 2560 not... So I thought Mega 2560 is probably a best choice but your video show me the opposite!
Thanks
good video. I made another measurement: computation of position of moon and sun, 1000 times. TIme for due: 4.1 sec, for 2560 12.9 sec. This is a factor of 3.15 (many , many trigonometric formulas, many doubles)
The voltage is different, one is 4.1V and 4.7V for the other test.
If the regulator is lineal the power measurement is not valid.
ps the UNO had other interrupts , so the execution time was longer then for the 2560
Thank you so much for your video. I was been asking myself what's a difference between them, and it's pretty good tell - tell videos for idiots like me to see a real difference between them.
This was a really cool and interesting video
+Brad Sogn Thanks!
Why do they calculate π as 3.1416059 and not 3.14159265?
I have a question, (doesn't refer to this video) are all shields of arduino uno compatible with arduino mega?
xxMikegrinxx No, not all. Many of them are, but you have to see if the shield supports the Mega.
Okey, thank you ;D
Thank you!
You cant compare it this way only. there are many differences in the way things work. f.ex.modulus works muuch more faster on arm that on avr per mhz of clock speed.
Yes of course. You are right. The architecture of an ARM processor is completely different. I am only comparing the for a quick visual represententation of the speed diference. It is not a complete benchmark, just to get an idea of how much faster Arduino Due is.
Good video,
BUT!
If we now place a DUE on a RAMPS 1.4 board
Would a 3D printer work better with it?
(I ask this because I shall than immediately order a DUE for my 3D printer!)
Friendly greetings from The Netherlands!
Rob.
what are you talking about ? Of course arduino DUE is faster it has a 32 bit ARM cpu VS 8 bit cpu of Mega .....and much more RAM for the stack and heap ....need to make a test ?
Please suggest me the lcd display for arduino due
I cannot find the link of arduino due tutorial.
the due still need improvement i advice every body to use mega its better due still need many modifications
Congratulations. Very good explanation!!!
+Edison Gonçalves Soares Thanks!
Yes but speed is only one small factor, especially boot speed, what about capabilities? Do a video on that.
I rewrote your code in C# and ran on my Core i5-6200U PC and it was 0.5 ms :)
Cool! But how much power does your computer need? Arduino needs less than 1W!
Let's say 100 W max. 100 folds more power, but 2000 folds faster! The ratio of performance to power for PC is much higher. Don't get me wrong though, I know they are two different things. I'm just saying. I love Arduino.
@@pomegranatechannel You forgot the display aka monitor and the power used to start the PC and that both systems are got shout down after all is done to be equal. If they are considered as a one purpose system of course.
very interesting thank you
Can I use all the sensors from the mega and r3 uno on the faster Arduino Due ?
Most of them, just take into acount Due can only handle 3.3V voltage levels
Umm, am I the only one who noticed that the result was wrong, PI is actually: 3.14159265359...
But the result on those displays was 3.1416059... wtf!?
Of course it is wrong because the algorithm run for only 100.000 iterations. You need more and more to have a result closer to the real pi.
Educ8s.com I agree with him. The algorithm used to compute the value of pi must have been logically flawed. The algorithm having the least complexity is the BPP formula. Which happens to be so simple that even handheld calculators can compute the few 9 digits of pi in a fraction of a second. Probably running on clock speeds of less than 1 MHz, graphing calculators use 6 MHz or less. I don't see computational limitations to be the cause of the error. You can calculate the first 15 digits of pi in less than 10,000 iterations.
greetings, your videos are great, I hope I can sorting the conflict that I have with the drivers of the Arduino Mega 2560 is a clone but not detected and windows (7.8) or Linux, only mac osx but when I reboot the computer has trouble detect it. thanks
thank you for legend in portuguese.
But that's not PI.
This PI value isn't right
Hi Nick, I tried duplicating your project but keep coming up with the compile error "undefined reference to pi", which I am presuming results from me not having the contents of what appears to be the second tab "pi(??) - can't read it properly in your video"? So how do I obtain the pi array data used in your ide and how do I create the second tab and install this data in it? Greatly appreciate your help in this regard - unfortunately arduino coding/programming is still a big mystery for me! Keep up the great work and Seasons Greetings for you and family. Cheers, Ian
Hello Ian, thanks! Seasons Greetings for you and your loved ones!
You are right, in the same folder that you have placed the Arduino Sketch you need a .c file which contains the binary data for the splash screen icon. It is named pi.c and you can find it on the website that I share the code. Press complile and it will work!
Cheers Nick
3.14159 This is the PI I know. I would like to know what causes this miscalculation...
Well 1.14159 rounded up is 1.1416, which is similar to the 3.1416059. The small error is due to the insufficient number of illiterations. If the test involves more illiterations the time required will be exponentially longer but the result only marginally more accurate so in this benchmark there is no reason for it
Excellent
Hi Sir I'm perch Arduino Mega 2560 R3 I need a mega 2560 software
but, but mega has 1 more processor that still sucks!
talha saglam Educ8s.com hi you are explaining so nice thank you for that.i wonder which one is faster between arduino due and cortex m0?Which one ARM is the arduino due used?i mean arduino using cortex m3 or cortex mX which one?
Celal Afşar Arduino DUE runs an ARM processor at 84Mhz. Arduino Zero Pro which has an ARM m0 runs at 48Mhz. So I expect that Arduino DUE is at least 2 times faster from M0. But of course we have to compare them in real life scenario to see.
Thank you for your kind and nice message.i wonder what do you thinking about the programming M0 and arduino due, which one is easier?İs there a big difference about programming?and if you want to do a complex robot which one will be your choice?
Celal Afşar If I were you, I would use the DUE. It is faster, it is cheaper and it is more time to market. That means that there is more software available for it, and more shield would work! Good luck!
aTalha SAGLAM Educ8s.com thank you so much, i'm grateful ,i'll be follow you:)
could someone tell me please why arduino due has been retired?
I'm pretty sure when they cut productions in italy because of the arduino srl lawsuit, they cut some of the arduino boards manufactured in italy.
Can you share the code program of nokia 5110 LCD?
Yes, I forgot to mention it. It is this graphics library that I was using in the previous videos as well: www.henningkarlsen.com/electronics/library.php?id=47
Thanks
Вау!
84MHz/16MHz=5,25x faster in theory.
+RPdigital Yes! You are right!
+RPdigital it's more than that, 32-bit uses less cycles to compute than a 8-bit, plus other innumerable CPU capabilities (... not to discuss here), more than just the frequency.
rodstartube Good point!
+RPdigital in fact if you use DSP capabilities, you may compute numerous advanced operations per cycle... even more: programming the DMA controller performs realtime operations without using CPU overhead almost like another CPU, but I don't know if Arduino takes advantages of those (...and much more!!!) advanced evil capabilities. I'm in the PIC32 world, but suppose 32-bit AVR (arduino DUE) is like that also. [my english sucks]
rodstartube You can tell from the experience and I cannot add anything more to this. Input has been accepted and memorized! :-)
can check Arduino Uno ?
I will check. But I think Arduino Mega and Arduino Uno have similar performance.
Educ8s.com Try replacing quartz 20 MHz forum.arduino.cc/index.php?PHPSESSID=ntpgqaltsh1sihhodphgk6k041&topic=137902.15
sssserya1 Yes I plan to do that. Performance of ATMEGA328 is about 1Mips per MHz. So, increasing the frequency to 20Mhz from 16MHz we have 25% increase in performance. So I expect that benchmark to complete in around 5.5 sec. But we will see.
due ori, or due chinese clone ,.. 😎
3.14159265359!
+ComputerIT This algorithm for a low number of iterations does not provide a good estimate of Pi. It was run for 100.000 iterations in this example. For better results, it needs at least 1.000.000 iterations.
+educ8s.tv I know :) I have calculated PI myself, but not on arduino. :) By the way, your channel is my favorite! :)
primer like
Muchas gracias
I just clicked on this video to give a thumbs down. This are 2 completely different type of processors there is nothing to compare
OK all car card sets and tests are useless as there are different engines build in.
That is why it is interesting. One is much more expensive and some projects do not need that extra power. It is good to know what real-life performance gains are. As RPdigital noted, if you just look at the specs you might assume that the performance difference would be around 5.25x but in the test, the performance difference is closer to 7.5x.