Jump to content

Pine H64 Model B

Humberg

By Humberg
in Pine A64

 Share

Recommended Posts

Da Alchemist

Da Alchemist

Posted
Posted

I have build a new image and feed it to my pineh64 (A). It runs up to 1.8GHz now with temperatures going up to 80°C. I think the real improvement is concerning USB3. my attached SSD is showing up UAS mode now! :thumbup:

BTW is it possible to get a fixed MAC Adress?

Link to comment
Share on other sites
NicoD

NicoD

Posted
Posted

I've got almost the same with the PineH64 at 1.8Ghz.
Using 7zip benchmark without heatsink it's 75°C after 3seconds. With the heatsink this takes longer, but after a minute full load it's also throttling heavily. 7zip isn't the heaviest load, it doesn't use full 100%.

Here Pine64 at 1.5Ghz no heatsink : CPU miner is a little faster than 1.8Ghz with heatsink, 7-zip is a little slower(CPU miner uses more of the cpu, so it throttles more than 7zip)
http://ix.io/1Ffr
Here Pine64 at 1.8Ghz with heatsink no fan
http://ix.io/1Flu
Here Orange Pi 3 at 1.8Ghz with heatsink no fan
http://ix.io/1EOn

I'm now doing SBC-bench @ 1.8Ghz no heatsink no fan. I'm sure it's worse than 1.5Ghz no heatsink no fan.

The base temperature does seem a bit lower on the PineH64. 45°C vs +50°C. The temperatures at full load seem the same. Both throttling.
I do these tests after running in the heatsink, and then let it cool to it's idle temp.


@TonyMac32 Could you try sbc-bench on it please?
I'd wished they'd sold a nice flat wide and long heatsink for it, from emmc to ethernet, and from hdmi to gpio's. There's enough room, and nothing's in the way. Only Hardkernel and FriendlyElec seem to care about that.

P.S.:
1.8Ghz no heatsink/fan
http://ix.io/1FlD
7-zip is as fast as 1.5Ghz no heatsink/fan, cpu miner is quite a bit slower.

A bit a pitty. But with small 5V fan at 3.3V you get max performance. So it's not like the Tinker Board/XU4, throttling with heatsink and fan.

Link to comment
Share on other sites
TonyMac32

TonyMac32

Posted
Posted
2 hours ago, NicoD said:

Could you try sbc-bench on it please?

I'll try that tomorrow, was working on other projects tonight. 

 

I do have a question for the general public:  Why is there not an opp between 1.488 and 1.8?  To be honest the spacing of the frequencies seems a bit strange in general, some are 240 MHz spaced, others are completely seemingly random numbers.  Is there a 1.6 GHz hiding in there somewhere?  I honestly haven't done any researching, just wondering if maybe this got overlooked, or if the clockspeeds are controlled by schizophrenic multipliers

Link to comment
Share on other sites
NicoD

NicoD

Posted
Posted
8 hours ago, TonyMac32 said:

I do have a question for the general public:  Why is there not an opp between 1.488 and 1.8?  To be honest the spacing of the frequencies seems a bit strange in general, some are 240 MHz spaced, others are completely seemingly random numbers.  Is there a 1.6 GHz hiding in there somewhere?  I honestly haven't done any researching, just wondering if maybe this got overlooked, or if the clockspeeds are controlled by schizophrenic multipliers

Indeed a bit strange 1.8Ghz 1.49Ghz 1.32Ghz 1.08Ghz 888Mhz 816Mhz 720Mhz 480Mhz

Those 2 in 800??? What the point of that? And then too much change with the higher numbers.

 

 

1 hour ago, constantius said:

I have not received yet the board but i'm prepared for 1.8 GHZ.

That will be more than sufficient. It's not that hard to cool it. I'm thinking of the people who will run it without heatsink/fan that will just throttle the board constantly without even knowing. And not having good results what again brings complaints...

Link to comment
Share on other sites
NicoD

NicoD

Posted
Posted
29 minutes ago, TonyMac32 said:

Perhaps a menu utility to set governor's and maximum/minimum clockspeeds for noobs?

Yes. Exactly! Just like with the Odroid's
https://github.com/mad-ady/odroid-cpu-control
 

I also notice the H64 has a lower base temperature than the OPi3. In idle this is a difference of 10°C. Maxed out this difference is smaller, but still 5°C less. This is so without heatsink, the same with theatsink and also with heatsink + 3.3V.
  


Here the H64 at the start of SBC-bench, so starts with idle temps. With heatsink +3.3V
Time        CPU    load %cpu %sys %usr %nice %io %irq   Temp
13:30:50: 1800MHz  0.38   6%   1%   4%   0%   0%   0%  30.8°C
13:32:50: 1800MHz  1.22  32%   1%  31%   0%   0%   0%  44.8°C
13:34:50: 1800MHz  1.35  33%   1%  32%   0%   0%   0%  37.6°C
13:36:50: 1800MHz  1.20  29%   0%  29%   0%   0%   0%  38.2°C
13:38:50: 1800MHz  1.30  29%   0%  28%   0%   0%   0%  37.9°C
13:40:50: 1800MHz  1.19  29%   0%  29%   0%   0%   0%  38.5°C
13:42:50: 1800MHz  1.12  29%   0%  29%   0%   0%   0%  38.5°C

Here the same with the OPi3
Time        CPU    load %cpu %sys %usr %nice %io %irq   Temp
17:09:32: 1800MHz  0.30  30%   1%  29%   0%   0%   0%  45.4°C
17:11:32: 1800MHz  1.16  32%   1%  31%   0%   0%   0%  50.1°C
17:13:32: 1800MHz  1.22  32%   1%  31%   0%   0%   0%  48.4°C
17:15:32: 1800MHz  1.14  29%   0%  29%   0%   0%   0%  44.6°C
17:17:32: 1800MHz  1.18  29%   0%  28%   0%   0%   0%  44.4°C
17:19:32: 1800MHz  1.37  29%   0%  28%   0%   0%   0%  44.7°C
17:21:32: 1800MHz  1.15  29%   0%  28%   0%   0%   0%  44.7°C

The OPi3 throttled a little bit with 3.3V at CPUMiner. The H64 with everything the same doesn't throttle a bit.

H64

System health while running cpuminer:

Time        CPU    load %cpu %sys %usr %nice %io %irq   Temp
13:53:23: 1800MHz  4.16  17%   1%  16%   0%   0%   0%  48.7°C
13:53:44: 1800MHz  4.35  99%   0%  99%   0%   0%   0%  68.7°C
13:54:06: 1800MHz  4.25 100%   0%  99%   0%   0%   0%  70.0°C
13:54:27: 1800MHz  4.26 100%   0%  99%   0%   0%   0%  70.6°C
13:54:49: 1800MHz  4.55 100%   0%  99%   0%   0%   0%  70.9°C
13:55:11: 1800MHz  4.73 100%   0%  99%   0%   0%   0%  71.2°C
13:55:32: 1800MHz  4.61 100%   0%  99%   0%   0%   0%  71.6°C
13:55:54: 1800MHz  4.56 100%   0%  99%   0%   0%   0%  71.6°C
13:56:16: 1800MHz  4.53 100%   0%  99%   0%   0%   0%  71.6°C
13:56:38: 1800MHz  4.51 100%   0%  99%   0%   0%   0%  71.7°C
13:56:59: 1800MHz  4.39 100%   0%  99%   0%   0%   0%  71.9°C
13:57:21: 1800MHz  4.48 100%   0%  99%   0%   0%   0%  71.8°C
13:57:43: 1800MHz  4.40 100%   0%  99%   0%   0%   0%  72.3°C
13:58:05: 1800MHz  4.32 100%   0%  99%   0%   0%   0%  71.8°C

OPi3
Time        CPU    load %cpu %sys %usr %nice %io %irq   Temp
00:08:38: 1800MHz  4.60  29%   1%  28%   0%   0%   0%  57.3°C
00:08:59: 1800MHz  4.50  99%   0%  99%   0%   0%   0%  74.3°C
00:09:21: 1800MHz  4.64 100%   0%  99%   0%   0%   0%  65.5°C
00:09:42: 1800MHz  4.46 100%   0%  99%   0%   0%   0%  70.7°C
00:10:04: 1800MHz  4.43 100%   0%  99%   0%   0%   0%  74.5°C
00:10:26: 1800MHz  4.31 100%   0%  99%   0%   0%   0%  75.1°C
00:10:47: 1800MHz  4.22 100%   0%  99%   0%   0%   0%  75.2°C
00:11:09: 1800MHz  4.26 100%   0%  99%   0%   0%   0%  74.7°C
00:11:31: 1800MHz  4.19 100%   0%  99%   0%   0%   0%  71.3°C
00:11:52: 1800MHz  4.13 100%   0%  99%   0%   0%   0%  71.2°C
00:12:14: 1800MHz  4.23 100%   0%  99%   0%   0%   0%  74.9°C
00:12:36: 1800MHz  4.16 100%   0%  99%   0%   0%   0%  65.7°C
00:12:57: 1800MHz  4.11 100%   0%  99%   0%   0%   0%  74.7°C
00:13:19: 1800MHz  4.25 100%   0%  99%   0%   0%   0%  74.8°C

Frequency doesn't show it here, but it is throttling, but only a little.

 

Link to comment
Share on other sites
TonyMac32

TonyMac32

Posted
Posted

A couple possibilities, one is a better thermal coupling to the power plane, a thicker copper layer somewhere, or, if these run the thermal sensors like an H3, a difference in the Vref for the circuit. Do you have a temperature sensor you could use to compare the two with external measurements?

Sent from my Pixel using Tapatalk

Link to comment
Share on other sites
NicoD

NicoD

Posted
Posted
3 minutes ago, TonyMac32 said:

A couple possibilities, one is a better thermal coupling to the power plane, a thicker copper layer somewhere, or, if these run the thermal sensors like an H3, a difference in the Vref for the circuit. Do you have a temperature sensor you could use to compare the two with external measurements?

I'll see if I can use my Voltage meter it's thermometer by sticking it onto the heatsink. Don't think it's very scientific, but well :)

 

Link to comment
Share on other sites
NicoD

NicoD

Posted
Posted
29 minutes ago, TonyMac32 said:

Do you have a temperature sensor you could use to compare the two with external measurements?

I've already learned that the heatsink heats up very slowly. Started at 35°C, always takes about 30seconds to rise a °C.  Now at 50°C while the cpu is throttling from 75°C to 70°C.
So a very small temperature difference in the heatsink makes for a large difference in the chip.
Also takes longer to cool down than to heat up. (the heatsink that is) In the cooldown the temperature shown is almost the same as shown on my volt meter. 48°C on the H64, 47°C on the volt meter. And while it goes down, it also shows the same on the H64.

I know my heatsink isn't very good, but it's the kind of heatsink everybody uses. It does show that the energy has trouble getting out of the chip.
When you let the heatsink warm up to about 60°C then it will be in constant throttle mode.

Can you do a SBC-bench now with your heatsink and without a fan?
(I let it cool down, and started 7z b again, now again 50°C and throttling)
Pictures here :
 

Spoiler

Cooldown after benchmark. 52°C on the meter, 53°C on the H64.

DSCN5897.thumb.JPG.3d358f4d8e934fc5ccff66074f1a942f.JPG

During benchmark, 51°C on the heatsink, throttling at the chip. (here 73°C, but it jumps between 77°C and 70°C constantly. The hotter the heatsink the faster the frequency of throttling.
DSCN5895.thumb.JPG.8112b0812c5de9828d8c2f14d4a5b188.JPG

 



 

Link to comment
Share on other sites
Dark26

Dark26

Posted
Posted

Thanks a lot for the image. My little test :

 

I try the debian desktop version in my rev A board.

 

Boot just fine.

 

when i try the last image from last year, problems appears for me :

 

- Lan was block à 100 mbits

- Usb3 not working

- pci express is a joke

- no graphical display

 

In this version, 

 

- Usb3 works fine. it's a very good news but not with a usb3 hub.

- Still the same for the lan, block at 100 mbits. But i think it's hardware problem, because with android image, i have only 100 mbits too.

 

If i  put a usb3 hub to have a usb 1000 mbits adaptateur, with a usb3 hard drive, then the performance shut down, like everything is in usb2. Too bad.

 

But it's working.

 

Thanks again for the great jobs.

 

Dark26

 

 

 

 

 

 

Link to comment
Share on other sites
NicoD

NicoD

Posted
Posted
8 minutes ago, Dark26 said:

when i try the last image from last year, problems appears for me :

Hi. That's a very old image. They are working on perfecting Armbian for the Pine H64 model B. Only on-board sound doesn't work. Otherwise it's very good.
You can try these images.
https://dl.armbian.com/pineh64/nightly/
 

They've been build last night. Should be a lot better than what you've tried now.
Otherwise you can build your own image.
https://docs.armbian.com/Developer-Guide_Build-Preparation/

Link to comment
Share on other sites
Dark26

Dark26

Posted
Posted
Just now, NicoD said:

Hi. That's a very old image. They are working on perfecting Armbian for the Pine H64 model B. Only on-board sound doesn't work. Otherwise it's very good.
You can try these images.
https://dl.armbian.com/pineh64/nightly/
 

They've been build last night. Should be a lot better than what you've tried now.
Otherwise you can build your own image.
https://docs.armbian.com/Developer-Guide_Build-Preparation/

 

 

That's the one i try today. :wub:

Link to comment
Share on other sites
NicoD

NicoD

Posted
Posted
15 hours ago, TonyMac32 said:

So 1.8, 1.64, 1.49

Great job. It's a lot better at 1.64Ghz without a fan.
For normal use this is perfect. Doesn't overheat quickly, and it's quite a bit faster than 1.5Ghz.
It now also throttles to 1.5Ghz instead of 1.3Ghz. This seems fine.

Even without a heatsink it's a lot better maxed out.
It is faster without a heatsink/fan than 1.5Ghz, and it's faster than 1.8Ghz.
And only when truly 100% maxed out it throttles. At 1.8Ghz even with only 1 core maxed it was throttling.

When throttling it's always inbetween 72°C - 82°C. It's not going to 85°C, but I do think above 80°C it goes to 1.3Ghz since once it reaches 80°C it immediatly goes down to 72°C. But it will be only a very small time at 1.3Ghz since my monitors don't show it.


I see you've kept the same voltage as 1.8Ghz. Can this be lowered to get an even better result?

I also tried the OPi+2 with the H3. That one has improved a lot since I last tried it. It now clocks to 1.37Ghz. But still performs the worst of all my sbc's( a lot worse than other armhf's at the same clockspeed).
It also throttles at 65°C to 1.1Ghz, but then the temperature keeps rising, so that can't be improved much this way. I wanted to know if HDMI sound really doesn't work on H3, it doesn't. 3.5mm audio jack is super quiet on it.

 

Link to comment
Share on other sites
NicoD

NicoD

Posted
Posted

The answer is probably just no.
My Pine H64 model B has got 4GB ram, but only 3GB is useable by the system.

Is there any way of accessing that extra GB? Maybe for a ram disk to run the OS on.

I think it will not have all the pins connected of the 2nd chip, but I don't know.

 

Link to comment
Share on other sites
renky

renky

Posted
Posted

https://youtu.be/HUsQaSZ0KhE?t=199

 

What is the significance of a battery connector?

 

Does it make it easier to make a battery notebook which you can connect and disconnect a power supply

unit to while notebook is running and charge while running?

 

Free software is software you can use, share, modify and redistribute. Does the wifi work

on free software?

 

Quote

ROCK64 4GB board designated as LTS

 

This computer is not?

 

Does the rock64 have a battery port?

 

Thanks.

 

 

 

 

 

 

 

Link to comment
Share on other sites
NicoD

NicoD

Posted
Posted
2 hours ago, renky said:

What is the significance of a battery connector?

 

Does it make it easier to make a battery notebook which you can connect and disconnect a power supply

unit to while notebook is running and charge while running?

As they said. That's the real time clock connector(RTC). That's for a module with a battery that keeps the time. I should not have said battery connector.

You can power the PineH64 with a power bank that can give stable 5V at 2A.
I use the Odroid C2 as a laptop with huge powerbanks of 26 800 mAh and a 7" or a 13" display. I can use it while maxed out for about 10 hours. I've got 2 of those powerbanks.
Some can be charged while in use. Mine can.

Link to comment
Share on other sites
andrw

andrw

Posted
Posted

Hi guys
Sorry for a noob question, but I'm trying to boot latest nightly on this board and see nothing in UART.
On the first page here it was mentioned that GND/TX/RX pins from EXT 6-pin connector should be used.
Is that enough?

I'm using CH340G USB-serial console from Pine store. 5V and 3.3V pins not connected. Tried both positions of S1 jumper on it.
Image: Armbian_5.83.190502_Pineh64_Debian_stretch_dev_5.0.10_desktop, connecting from laptop by `minicom -D /dev/ttyUSB1 -o -w -c on -b 115200`

Link to comment
Share on other sites
Igor

Igor

Posted
Posted
9 hours ago, andrw said:

but I'm trying to boot latest nightly on this board and see nothing in UART.


To my knowing they are all broken. I was too late to salvage last known good working, but soon there will be new builds with K5.1.y ... which will probably work.

Link to comment
Share on other sites
Igor

Igor

Posted
Posted
44 minutes ago, martinayotte said:

PineH64 should have their own nightly images, but it seems that build servers still sleeping there recent days ...


Small maintenance. Later/perhaps tomorrow. This service doesn't have a backup ATM but its planned.

Link to comment
Share on other sites
Guest
This topic is now closed to further replies.
 Share
Go to topic listing
×
×
  • Create New...

Important Information

Terms of Use - Privacy Policy - Guidelines

  I accept