• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Question Who is the Rasberry PI expert ? And a recommended system.

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Endgame124

Senior member
Feb 11, 2008
391
228
116
update. I now I have OS installed, but the update failed for some reason (OS updates). I have rebooted. It was probably due to me uplugging the network cable. Its now on wifi. How do I start that again ?
I'm not sure where to do the update from the GUI, but from command line, this will work
:
sudo apt-get update
sudo apt-get upgrade -y
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
20,812
8,997
136
@Endgame124

OK, how to I...
Get CPU temps
Get CPU speed

Its running WCG now, I got that far, but the heatsink is so hot (in open air) I can't keep my finger on it.
 
Last edited:

Endgame124

Senior member
Feb 11, 2008
391
228
116
OK, how to I...
Get CPU temps
Get CPU speed

Its running WCG now, I got that far, but the heatsink is so hot (in open air) I can't keep my finger on it.
Here is a 1 line shell script to run from command prompt to give you temp, speed, voltage ever 30 seconds

while true; do date; vcgencmd measure_temp; vcgencmd measure_volts core; vcgencmd measure_clock arm; temp=`/opt/vc/bin/vcgencmd measure_temp | awk -F "=" '{print $2}' | awk -F "'" '{print int ( $1 )}'`; if [[ $temp -gt $maxtemp ]]; then maxtemp=$temp; fi; echo "Max temp: $maxtemp"; echo; sleep 30; done

Edit, with just a small heat sink, you will need a small amount of airflow to keep from thermally throttling. This is doubly true when using hdmi, as you are engaging the on chip video as well.
 
Last edited:

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
20,812
8,997
136
Here is a 1 line shell script to run from command prompt to give you temp, speed, voltage ever 30 seconds

while true; do date; vcgencmd measure_temp; vcgencmd measure_volts core; vcgencmd measure_clock arm; temp=`/opt/vc/bin/vcgencmd measure_temp | awk -F "=" '{print $2}' | awk -F "'" '{print int ( $1 )}'`; if [[ $temp -gt $maxtemp ]]; then maxtemp=$temp; fi; echo "Max temp: $maxtemp"; echo; sleep 30; done
So I put a 120mm fan pointing at it, and its now 65c and 1.5 ghz (pretty sure that it)
 
  • Like
Reactions: Endgame124

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
20,812
8,997
136
OK, very early numbers. It takes about twice as long to do a WCG covid task as a 3900x, and does 3 at a time instead of 22. So the 3900x is 14 times the horsepower at 142 watts. So 5 watts or so times 14 is 70 watts... hmmmm. Well, this is early, and I didn't take real out of the wall numbers. Well, if this hold up, 14 units would be about $1500, and a 3900x system is less than that (approx). Also, that many PI's is a lot to keep track of. Also the space of 14 of those vs one 3900x system is about equal.

So it looks like the PIs win in total power, but loose in maintenance, usability and so on. I have not even started on EPYC.....

Again, very preliminary results.
 

Endgame124

Senior member
Feb 11, 2008
391
228
116
OK, very early numbers. It takes about twice as long to do a WCG covid task as a 3900x, and does 3 at a time instead of 22. So the 3900x is 14 times the horsepower at 142 watts. So 5 watts or so times 14 is 70 watts... hmmmm. Well, this is early, and I didn't take real out of the wall numbers. Well, if this hold up, 14 units would be about $1500, and a 3900x system is less than that (approx). Also, that many PI's is a lot to keep track of. Also the space of 14 of those vs one 3900x system is about equal.

So it looks like the PIs win in total power, but loose in maintenance, usability and so on. I have not even started on EPYC.....

Again, very preliminary results.
This sounds about right - power wise the pi comes out ahead of almost all the desktop chips, especially if they are running stock. Management wise they can be a PITA, but that’s why I’m working on getting kubernetes running on them, to make management easier.

Note, each pi should be running 4 tasks, not 3 (not sure if typo). Also, You can further cut down the power of the pi by under volting, turning off hdmi, etc, but that doesn’t help with base management at all.
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
20,812
8,997
136
I changed CPU to 100/100 and now its doing 4 tasks. But it looks like it will take 5 hours a task, so that makes it even worse all round. vs 1:40 on the 3900x. Thats 18x difference or 90 watt vs 142. Not worth it for the amount of time it will take to manage them IMO.
 

Endgame124

Senior member
Feb 11, 2008
391
228
116
I changed CPU to 100/100 and now its doing 4 tasks. But it looks like it will take 5 hours a task, so that makes it even worse all round. vs 1:40 on the 3900x. Thats 18x difference or 90 watt vs 142. Not worth it for the amount of time it will take to manage them IMO.
So increasing to 4 processes made each process slower? Are you thermally throttling at 4 processes?

Edit - how much management are you planning on? My experience so far is that they are pretty fire and forget, but I’m also not trying to change projects with mine either. The biggest exception management was so far is when I had a UPS failure and I blew out a few SD cards, and when I was over clocking the pis and I was dealing with the occasional lock up.
 
Last edited:

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
20,812
8,997
136
So increasing to 4 processes made each process slower? Are you thermally throttling at 4 processes?

Edit - how much management are you planning on? My experience so far is that they are pretty fire and forget, but I’m also not trying to change projects with mine either. The biggest exception management was so far is when I had a UPS failure and I blew out a few SD cards, and when I was over clocking the pis and I was dealing with the occasional lock up.
Last I looked it was still at 1.5 ghz and 72c
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
20,812
8,997
136
93% in 7 hours.....

OK, now this is a very specific test, but as much as they are disparate, I don't think other scenarios would be that different.

So I got an 8 gig Raspberry PI (8 gig, 32 gig HD, A72 4 core 1.5 ghz). Its does the Open Pandenics covid 19 WCG unit, 4 cores times 7.5 hours

My EPYC 7742 does 128 of these same units at 2 ghz in 2.5 hours.

So it would take 32 PI's to do the same work, but 3 times slower. An 32 PIs at 5 watts each is 160 watts, vs the 7742 @250 watts. (motherboard, memory and all) So more power and 3 times longer for the same work is 2 times the power usage.

Cost... 32 PIs at about $60 each (including power supplies and cables, probably more) would be at least $1920.

The EPYC is about $4000+480+580 or $5060.

So the EPYC effectively less because it cost 2 times less electricity for the same work.

This seems WAY different than I am seeing when people talk about ARM and efficiency and power usage. Please tell me where I am mistaken in my math., But don't be a jerk if you find the errors of my ways. The run times are real.. The PI is at 7.5 hours and hasn't actually finished a unit yes, no other tasks running on the PI. And I am looking at several units on the 7742, one is at 97% in 2:21.

Edit: The first unit finished in 7:50, almost 8 hours. The next 2 units are at 8 hours and still running at 96 and 98%
 
Last edited:

StefanR5R

Diamond Member
Dec 10, 2016
3,641
3,910
106
This seems WAY different than I am seeing when people talk about ARM and efficiency and power usage.
Raspberry Pi 4 has got a Broadcom BCM2711 processor. That's a 28 nm chip.

There is just now way that this can ever be as power efficient as a 7 nm chip (provided the latter is properly configured for efficiency instead of for show), regardless of the instruction set architecture difference.
 
  • Like
Reactions: Endgame124

Endgame124

Senior member
Feb 11, 2008
391
228
116
Raspberry Pi 4 has got a Broadcom BCM2711 processor. That's a 28 nm chip.

There is just now way that this can ever be as power efficient as a 7 nm chip (provided the latter is properly configured for efficiency instead of for show), regardless of the instruction set architecture difference.
The true ARM competition to Epyc is amazon’s Gravitron or something based on the Neoverse N1 or maybe Neoverse V1. That kind of stuff doesn’t exist in the consumer market though, and you can’t really even order it as server parts or whole servers, even if you wanted to.

The benefit for raspberry pi using 28nm is that they can make the whole single board computer and sell it for $35. They expect to be on 28nm for at least 5 more years, so expect that raspberry pis are always going to be 3-6 nodes behind everyone else.
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
20,812
8,997
136
Well, I did not know it was 28nm. I thought it was the latest ARM for a consumer, and at least in the ballpark of Rome, or even NAples.

Live and learn the hard way.
 

Endgame124

Senior member
Feb 11, 2008
391
228
116
For what it’s worth, when you consider that Sandy Bridge was 32nm and ivy bridge was 22nm, the efficiency seen on a raspberry pi is pretty remarkable. It’s competing well on efficiency against 10nm intel and 7nm AMD consumer products.

As I see it, the primary reasons to use a raspberry pi are:

1) decommissioning old hardware. I replaced my old core2quad with 2x 4gb pi 4s. I get the same average credit in Rosetta, and use roughly 140w less to get it

2) low cost. I can setup a passively cooled 2GB pi 4 Using a Flirc or villaros case for $70 or less. I can’t incrementally add processing power to my other PCs for less than 3x that. I’m pretty much stuck with my A10-7870k until I’m willing to spend $500+ to upgrade it, and at this point there is no sense buying anything until Ryzen 4000 chips are available.

3) staying under a certain power envelope. Ex - let’s say I’m willing to commit 20 watts to DC - I can’t even idle any of my PCs for 20 watts, yet I could run 4-5 pi 4s for 20 watts.

Edit - if someone had the right connections, it would be really, really interesting to see what the 7nm, 160 core Ampere platform could do:

 
Last edited:
Thread starter Similar threads Forum Replies Date
E Distributed Computing 11

ASK THE COMMUNITY