sfx2000
Part of the Furniture
Always thought the udoo x86 board was interesting with the arduino built in - which makes it interesting when doing things like robotics (the quark SE and ARC's are via USB to the N3710 host).
Asus TinkerBoard
- Thread starter sfx2000
- Start date
Always thought the udoo x86 board was interesting with the arduino built in - which makes it interesting when doing things like robotics (the quark SE and ARC's are via USB to the N3710 host).
sfx2000
Part of the Furniture
Fun stuff... chmod +x after creating the file - tinkerboard tends to throttle hard after 70 Celsius..
pi3 sticks around to around 80 before it slows down...
vi info.sh
As shown below... watch -n1 ./info.sh
pi3 sticks around to around 80 before it slows down...
vi info.sh
Code:
#!/bin/bash
celsius1=$(cat /sys/class/thermal/thermal_zone0/temp | sed 's/.\{3\}$/.&/')
clock1=$(cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_cur_freq | sed 's/.\{3\}$/.&/')
echo "Hostname => $(hostname)"
echo "SW Rev => $(uname -vr)"
echo "Uptime =>$(uptime)"
echo "=========================="
echo "Core Temp => ${celsius1} °C"
echo "CoreClock => ${clock1} MHz"As shown below... watch -n1 ./info.sh
System Error Message
Part of the Furniture
Both my pi 3 and thinkerboard share the same case actually. I mean they both have heatsink and fan on them. The thinkerboard is raspberry pi compatible so all the same cases and cooling fits.
sfx2000
Part of the Furniture
TinkerBoard would definitely benefit from a heatsink fan - it does include a largish heatsink in the box (large enough that it might interfere with some HAT's, so watch clearance there.
The Rockchip idles around 50c - and it runs up fairly quickly if all 4 cores are loaded up (for example, video playback with some codecs)
Microsoft had an interesting article regarding active cooling on a Pi 3, showing the benefits...
https://microsoft.github.io/ELL/tutorials/Active-cooling-your-Raspberry-Pi-3/
sfx2000
Part of the Furniture
To summarize things up with the TinkerBoard. It's an interesting board - not for the beginner - a Pi3 is a better choice there - mostly due to documentation, software, and community support.
I would consider Tinker to be an intermediate class board - it's serviceable, one has to roll up one's sleeves a bit and get into the tender bits of linux to make the most of it, but all the tools are there to make that happen.
I would consider Tinker to be an intermediate class board - it's serviceable, one has to roll up one's sleeves a bit and get into the tender bits of linux to make the most of it, but all the tools are there to make that happen.
sfx2000
Part of the Furniture
BTW - the Asus UK site seems to have the best factory support for this board...
https://www.asus.com/uk/Single-Board-Computer/Tinker-Board/HelpDesk_Download/
They recently pushed out a new Tinker OS 2.05 release with quite a few fixes... Tinkerboard also supports android, there are ROM's there for that side of the house.
The best Tinkerboard Community seems to be homed in the UK as well...
https://tinkerboarding.co.uk/forum/
And then I should add the armbian community.
https://www.armbian.com/tinkerboard/
https://www.asus.com/uk/Single-Board-Computer/Tinker-Board/HelpDesk_Download/
They recently pushed out a new Tinker OS 2.05 release with quite a few fixes... Tinkerboard also supports android, there are ROM's there for that side of the house.
The best Tinkerboard Community seems to be homed in the UK as well...
https://tinkerboarding.co.uk/forum/
And then I should add the armbian community.
https://www.armbian.com/tinkerboard/
Good to hear. Seems right time to get onboard then. I came across a benchmark TinkerBoard vs Rasp Pi 3 B+ vs a few other. TinkerBoard is still ahead of 3 B+ by miles on most items.
Is it true that WiFi performance on TinkerBoard is only 30Mbit/s?
Benchmark: https://www.phoronix.com/scan.php?page=article&item=raspberrypi-3-bplus&num=3
sfx2000
Part of the Furniture
WiFi client performance on the Tinker is not a strong point...
The full GbE port in TinkerBoard is way more useful than 300Mbps ethernet port on Pi 3 B+ for network intensive applications.
From this benchmark, leaving 64-bit/32-bit aside, A17 is much more efficient core than A53.
These results from unixbench are a bit counter climax. I would think swapping 890.8 and 605.7 will make more sense. What's going on there?
sfx2000
Part of the Furniture
The DVFS curves on the tinker with the SW in use at the time exposed a throttling issue. It gets very hot, very fast, and pulls back under heavy load.
The Rockchip can burst very fast up to 1.8GHz, and it throttles back down to 125MHz - the 2.0.5 release should address some of this, and yes, one should consider a robust cooling solution for Tinker if one wants to push all the cores hard.
sfx2000
Part of the Furniture
That's not really a benchmark - it's DMIPS/HZ for various ARM cores...
That being said - the A17 is a high-end core, and the RK3288 is a great implementation - we don't see a lot of other A17 chips as they were introduced about the same time as the big ARMv8 jump to 64-bit.
a53 essentially replaced A9, as it's small on the die, and supports more features - one can place 4 Cortex-a53 cores in less space than two Cortex-A9's, and get better performance and thermals at a given process node.
The Gigabit port on Tinker is actually the Rockchip GMAC, it's built into the SoC, and it is very good considering the price.
The Pi3 B+ -- it's GbE, but it's part of a USB2 combo controller, so while the PHY is GbE, effective throughput is going to be limited. Still nice to have, as it reduces some power, and it adds POE support, which for some is nice.
The guy ran tests and did the charting from his own data according to the article..
DMIPS/HZ wise, A9 appears to be a more efficient core than A53. The wonder ppl are seeing in A53 (such as RT-86U) is higher clock, and NEON (for OpenSSL).
Nice. Definitely preferred if applications are more network intensive. I heard TinkerBoard requires a good quality PSU. Also the TinkerBoard S is just released. I can't find any retailer nearby..
sfx2000
Part of the Furniture
Actually no - the post slide is direct from an ARM presentation - I know, I was there...
A9 is a good core - A17 is the logical end - it is an improvement on A12 (paper release) and developed outside of A15, and it's competitive and more power efficient. It's probably the best ARMv7 only core out there, just not very popular...
a53 - it's likely ARM's most successful core to date - it has warts and bugs, but it's an in-order core similar to a7 - that being said, in certain tasks, it's very good.
The Broadcom SoC's like the RT-86U - it's not a53, it's a architecture variant, and it's pretty good - echo's of Vulcan tweaks there...
similar to what QCOM did with Krait/Kyro, but in a different use-case.
I went through the article again..he didn't get the numbers himself, and he said so in the article. My bad..lol.
"...To do so, I’ve simply use DMIPS/Mhz (Dhrystone MIPS/Megahertz) values listed on Wikipedia."
TinkerBoard vs Pi 3 B+ i.e. A17 vs A53.
TinkerBoard wins over Pi 3 B+ categorically albeit cost more for TinkerBoard.
The newer TinkerBoard S cost $80 (but comes with on-board 16GB eMMC)
TinkerBoard also has 2GB RAM double the size of 3 B+.
Not sure about TinkerOS or Armbian features. Need to take a further look at the software.
edit:
The Rockship has extra 400MHz advantage over the Broadcom chip.
Given A17 is more efficient than A53, that also helps quite a bit I think.
Also that the Rockship is overclockable to 2.2Ghz - a further extra 400MHz advantage over standard TinkerBoard/S. That's sweet.
Last edited:
sfx2000
Part of the Furniture
One of the nice things for an advanced user - Tinker uses uBoot as a bootloader, so it opens up possibilities... it's a cleaner approach for an experienced linux developer, as Pi's boot is pretty odd there - Pi can use uBoot, and some distro's outside of Raspbian do, but it complicates things.
Tinker also supports Android, and one of the two images that Asus provides is an android image - if I recall, it might be Marshmallow, maybe Noughat, but it's there, and that being said, if one tires of mucking about with the linux stuff, the tinker makes for a great STB for some Kodi action over 4K.
One complaint about Tinker, and this is not really directed just at Tinker, but the ARM SoC vendors in general - video support - it's a real mess there - one can do dumb framebuffer and get something on screen, but pretty much anything _fast_ needs ugly blobs - the the RK3288 on Tinker is the same there - Mali is a PITA, so it pretty much every other vendors (IMG, Adreno, Vivante, even VC4 on the Pi for that matter) which locks one into the SoC vendor's kernel/driver package. There's been a bit of upside with Rockchip since google uses this SoC for many chromebooks, but google's stuff is odd with dependencies that are not mainline. Same goes with the other SoC vendors with Android as those are purpose specific and are hard, if not impossible, to port into a mainline kernel. Grrr...
Tinker also supports Android, and one of the two images that Asus provides is an android image - if I recall, it might be Marshmallow, maybe Noughat, but it's there, and that being said, if one tires of mucking about with the linux stuff, the tinker makes for a great STB for some Kodi action over 4K.
One complaint about Tinker, and this is not really directed just at Tinker, but the ARM SoC vendors in general - video support - it's a real mess there - one can do dumb framebuffer and get something on screen, but pretty much anything _fast_ needs ugly blobs - the the RK3288 on Tinker is the same there - Mali is a PITA, so it pretty much every other vendors (IMG, Adreno, Vivante, even VC4 on the Pi for that matter) which locks one into the SoC vendor's kernel/driver package. There's been a bit of upside with Rockchip since google uses this SoC for many chromebooks, but google's stuff is odd with dependencies that are not mainline. Same goes with the other SoC vendors with Android as those are purpose specific and are hard, if not impossible, to port into a mainline kernel. Grrr...
sfx2000
Part of the Furniture
FYI....
TinkerOS Debian 2.07 - link below - again, the UK site seems to have the best support for this board...
https://www.asus.com/uk/Single-Board-Computer/Tinker-Board/HelpDesk_Download/
Changes:
TinkerOS_Debian V2.0.7
TinkerOS_Debian V2.0.7
TinkerOS default username is “linaro”, password is “linaro”
Release Notes
*Changelog
1. Merge Upstream/Rockchip (commit: 2801809, tag: release-20171213, kernel: 4.4.103).
2. Update Qt video for Media Player to support repeat function.
3. Improve the HDMI Eye Pattern.
4. Fixed onboard USB audio module missing after resume from suspend.
5. Update U-Boot's UMS name to "TinkerBoard UMS disk".
6. Set the GPIO pins default mode to input mode.
7. Support CAN-Bus.
8. Fixed Tinker Board can not be used as Wi-Fi AP.
9. Add missing spidev nodes when SPI enable.
10. Support Google AIY Voice Kit (v2) HAT sound card.
11. Sync Rockchip package to support display ratio setting.
12. Improve the USB camera performance.
13. Support ADS7846 touch screen for 4"/3.5" HDMI LCD.
14. Set U-Boot only stop the autoboot when triggered the space key.
15. Update rtl8723bs driver.
16. Improve the WiFi roaming fail issue.
17. Fix the warning message when trying to build the Kernel source code.
18. Add back the missing comment of the rk3288-gpiomem.c file.
*ARM Security Update:
1. Meltdown and Spectre
Variant 1: bounds check bypass (CVE-2017-5753)
Variant 2: branch target injection (CVE-2017-5715)
*Tinker CLI and GUI's Configuration Tools (Experimental):
1. Support to enabled/disabled wlan verbose log.
MD5:
bbe7fe5b56c5bb4ce3ebd057b939ebda (zip)
1a723676d5e977096a8fbabf8ef06bc4 (img)
TinkerOS Debian 2.07 - link below - again, the UK site seems to have the best support for this board...
https://www.asus.com/uk/Single-Board-Computer/Tinker-Board/HelpDesk_Download/
Changes:
TinkerOS_Debian V2.0.7
TinkerOS_Debian V2.0.7
TinkerOS default username is “linaro”, password is “linaro”
Release Notes
*Changelog
1. Merge Upstream/Rockchip (commit: 2801809, tag: release-20171213, kernel: 4.4.103).
2. Update Qt video for Media Player to support repeat function.
3. Improve the HDMI Eye Pattern.
4. Fixed onboard USB audio module missing after resume from suspend.
5. Update U-Boot's UMS name to "TinkerBoard UMS disk".
6. Set the GPIO pins default mode to input mode.
7. Support CAN-Bus.
8. Fixed Tinker Board can not be used as Wi-Fi AP.
9. Add missing spidev nodes when SPI enable.
10. Support Google AIY Voice Kit (v2) HAT sound card.
11. Sync Rockchip package to support display ratio setting.
12. Improve the USB camera performance.
13. Support ADS7846 touch screen for 4"/3.5" HDMI LCD.
14. Set U-Boot only stop the autoboot when triggered the space key.
15. Update rtl8723bs driver.
16. Improve the WiFi roaming fail issue.
17. Fix the warning message when trying to build the Kernel source code.
18. Add back the missing comment of the rk3288-gpiomem.c file.
*ARM Security Update:
1. Meltdown and Spectre
Variant 1: bounds check bypass (CVE-2017-5753)
Variant 2: branch target injection (CVE-2017-5715)
*Tinker CLI and GUI's Configuration Tools (Experimental):
1. Support to enabled/disabled wlan verbose log.
MD5:
bbe7fe5b56c5bb4ce3ebd057b939ebda (zip)
1a723676d5e977096a8fbabf8ef06bc4 (img)
sfx2000
Part of the Furniture
Should add this...
https://tinkerboarding.co.uk/wiki/index.php/Main_Page
The Tinker Board community is working to build things up...
https://tinkerboarding.co.uk/wiki/index.php/Main_Page
The Tinker Board community is working to build things up...
Good to see the kernel gets updated.
Btw, Libre Computer is offering very appealing alternatives using the same and higher grade Rockchips!
sfx2000
Part of the Furniture
The Tinker is getting better SW support over time - Armbian stays fairly close to current (I think they're on 4.14). The upside with Tinker is that it is Asus, which may be important to some...
The Libre Computer Renegade boards use the more recent RK3328, and the Renegade Elite uses the RK3399
The biggest challenge with the Rockchip and Allwinner based boards is the Mali graphics drivers - some distro's are better than others...
The Libre Computer Renegade boards use the more recent RK3328, and the Renegade Elite uses the RK3399
The biggest challenge with the Rockchip and Allwinner based boards is the Mali graphics drivers - some distro's are better than others...
sfx2000
Part of the Furniture
Still something up with TinkerBoard and DVFS.... that's the only thing I can think of off the top of my head - and these numbers actually concur with other projects I've done - both are stock boards - the Tinker has the Heat Sink included with the board, the Pi 3B+ does not have a heat sink - both are running on SanDisk Extreme 32GB sd cards...
Both are headless, running the CLI only, configured as such via the OS utils.
Should already be there, but just to make sure...
Code:
sudo apt install build-essential gitNow, download UnixBench from GitHub...
Code:
git clone https://github.com/kdlucas/byte-unixbench.gitBuild it local - both platforms support what is needed....the ./Run script should work, but one can do a make clean, make, and then do the ./Run
Up first is TinkerOS 2.07
Code:
Benchmark Run: Mon Jul 09 2018 16:32:15 - 17:00:49
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 8712967.8 lps (10.0 s, 7 samples)
Double-Precision Whetstone 1144.7 MWIPS (12.1 s, 7 samples)
Execl Throughput 1240.1 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 132610.6 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 35006.1 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 334655.8 KBps (30.0 s, 2 samples)
Pipe Throughput 222101.9 lps (10.0 s, 7 samples)
Pipe-based Context Switching 32554.8 lps (10.0 s, 7 samples)
Process Creation 2247.1 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 2294.4 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 695.8 lpm (60.1 s, 2 samples)
System Call Overhead 505074.9 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 8712967.8 746.6
Double-Precision Whetstone 55.0 1144.7 208.1
Execl Throughput 43.0 1240.1 288.4
File Copy 1024 bufsize 2000 maxblocks 3960.0 132610.6 334.9
File Copy 256 bufsize 500 maxblocks 1655.0 35006.1 211.5
File Copy 4096 bufsize 8000 maxblocks 5800.0 334655.8 577.0
Pipe Throughput 12440.0 222101.9 178.5
Pipe-based Context Switching 4000.0 32554.8 81.4
Process Creation 126.0 2247.1 178.3
Shell Scripts (1 concurrent) 42.4 2294.4 541.1
Shell Scripts (8 concurrent) 6.0 695.8 1159.7
System Call Overhead 15000.0 505074.9 336.7
========
System Benchmarks Index Score 316.3
------------------------------------------------------------------------
Benchmark Run: Mon Jul 09 2018 17:00:49 - 17:29:40
4 CPUs in system; running 4 parallel copies of tests
Dhrystone 2 using register variables 22388261.9 lps (10.0 s, 7 samples)
Double-Precision Whetstone 3541.5 MWIPS (13.5 s, 7 samples)
Execl Throughput 3383.9 lps (29.5 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 170703.7 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 44959.4 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 485914.9 KBps (30.0 s, 2 samples)
Pipe Throughput 906476.6 lps (10.0 s, 7 samples)
Pipe-based Context Switching 81474.5 lps (10.0 s, 7 samples)
Process Creation 6417.8 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 5247.4 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 672.5 lpm (60.2 s, 2 samples)
System Call Overhead 1535317.7 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 22388261.9 1918.4
Double-Precision Whetstone 55.0 3541.5 643.9
Execl Throughput 43.0 3383.9 787.0
File Copy 1024 bufsize 2000 maxblocks 3960.0 170703.7 431.1
File Copy 256 bufsize 500 maxblocks 1655.0 44959.4 271.7
File Copy 4096 bufsize 8000 maxblocks 5800.0 485914.9 837.8
Pipe Throughput 12440.0 906476.6 728.7
Pipe-based Context Switching 4000.0 81474.5 203.7
Process Creation 126.0 6417.8 509.3
Shell Scripts (1 concurrent) 42.4 5247.4 1237.6
Shell Scripts (8 concurrent) 6.0 672.5 1120.8
System Call Overhead 15000.0 1535317.7 1023.5
========
System Benchmarks Index Score 682.6Now the Pi3 B+
Code:
Benchmark Run: Mon Jul 09 2018 16:32:59 - 17:01:08
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 4884587.5 lps (10.0 s, 7 samples)
Double-Precision Whetstone 1111.7 MWIPS (10.8 s, 7 samples)
Execl Throughput 917.0 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 144045.5 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 41659.0 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 361676.9 KBps (30.0 s, 2 samples)
Pipe Throughput 264765.2 lps (10.0 s, 7 samples)
Pipe-based Context Switching 53270.5 lps (10.0 s, 7 samples)
Process Creation 2231.0 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 1786.1 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 578.2 lpm (60.1 s, 2 samples)
System Call Overhead 604504.0 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 4884587.5 418.6
Double-Precision Whetstone 55.0 1111.7 202.1
Execl Throughput 43.0 917.0 213.3
File Copy 1024 bufsize 2000 maxblocks 3960.0 144045.5 363.8
File Copy 256 bufsize 500 maxblocks 1655.0 41659.0 251.7
File Copy 4096 bufsize 8000 maxblocks 5800.0 361676.9 623.6
Pipe Throughput 12440.0 264765.2 212.8
Pipe-based Context Switching 4000.0 53270.5 133.2
Process Creation 126.0 2231.0 177.1
Shell Scripts (1 concurrent) 42.4 1786.1 421.2
Shell Scripts (8 concurrent) 6.0 578.2 963.7
System Call Overhead 15000.0 604504.0 403.0
========
System Benchmarks Index Score 311.8
------------------------------------------------------------------------
Benchmark Run: Mon Jul 09 2018 17:01:08 - 17:29:13
4 CPUs in system; running 4 parallel copies of tests
Dhrystone 2 using register variables 17267337.2 lps (10.0 s, 7 samples)
Double-Precision Whetstone 4215.9 MWIPS (9.9 s, 7 samples)
Execl Throughput 2258.9 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 229944.0 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 62439.0 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 573495.6 KBps (30.0 s, 2 samples)
Pipe Throughput 1051084.1 lps (10.0 s, 7 samples)
Pipe-based Context Switching 234694.7 lps (10.0 s, 7 samples)
Process Creation 4731.8 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 4482.9 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 618.1 lpm (60.2 s, 2 samples)
System Call Overhead 2315401.8 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 17267337.2 1479.6
Double-Precision Whetstone 55.0 4215.9 766.5
Execl Throughput 43.0 2258.9 525.3
File Copy 1024 bufsize 2000 maxblocks 3960.0 229944.0 580.7
File Copy 256 bufsize 500 maxblocks 1655.0 62439.0 377.3
File Copy 4096 bufsize 8000 maxblocks 5800.0 573495.6 988.8
Pipe Throughput 12440.0 1051084.1 844.9
Pipe-based Context Switching 4000.0 234694.7 586.7
Process Creation 126.0 4731.8 375.5
Shell Scripts (1 concurrent) 42.4 4482.9 1057.3
Shell Scripts (8 concurrent) 6.0 618.1 1030.1
System Call Overhead 15000.0 2315401.8 1543.6
========
System Benchmarks Index Score 765.4Similar threads
Similar threads
| Thread starter | Title | Forum | Replies | Date |
|---|---|---|---|---|
| D | Forum bias toward Netgear and Asus? | Other Discussions | 33 |