BPi-M3 performance

Since we’ve never seen a single benchmark provided by SinoVoip (that must happen for a reason, normally devs are proud that their hardware is fast and try to proove it) I’m really curious how fast the board is.

I had a look into one now available fex file and started laughing:

; B_LV1: core vdd is 1.08v if cpu frequency is (1800Mhz,  2016Mhz]
; B_LV2: core vdd is 1.00v if cpu frequency is (1608Mhz,  1800Mhz]
; B_LV3: core vdd is 0.92v if cpu frequency is (1200Mhz,  1608Mhz]
; B_LV4: core vdd is 0.84v if cpu frequency is (   0Mhz,  1200Mhz]
; B_LV5: core vdd is 0.84v if cpu frequency is (   0Mhz,  1200Mhz]
; B_LV6: core vdd is 0.84v if cpu frequency is (   0Mhz,  1200Mhz]
; B_LV7: core vdd is 0.84v if cpu frequency is (   0Mhz,  1200Mhz]
; B_LV8: core vdd is 0.84v if cpu frequency is (   0Mhz,  1200Mhz]

2.0 Ghz @ 1.08v – LOL. Can someone please post the output of

cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors
cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_frequencies
dmesg | egrep "ddrfreq|dsm|cpu_freq"

and then set the stuff to maximum, switch off all CPU cores but 4, do an “apt-get install sysbench p7zip-full mbw” and do a little benchmarking:

echo performance >/sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
echo 2016000000 >/sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq
for i in 7 6 5 4 ; do echo 0 >/sys/devices/system/cpu/cpu${i}/online ; done
sysbench --test=cpu --cpu-max-prime=20000 run --num-threads=4
7za b
mbw -t0 256 && mbw -t1 256 && mbw -t2 256

I’m really curious what to expect from a Cortex-A7 running above 2.0 GHz with these little tests (maybe just to discover that there’s an internal divider working and that the cores are running with 1 GHz in reality?)

Anyone? I would do it myself but my M3 hasn’t arrived yet.

To be more precise: @sinovoip, that’s your job, isn’t it? You’ve plenty of boards lying around and experienced developers that have already done this sort of tests since you can’t deliver an untested product, right?

When do you get back to us with the performance figures and output of the commands I posted above? Today? Tomorrow?

And please disable 4 CPUs as outlined above so that throttling might not jump in. Thx!

Ok, 2 hours later still no time to spend 5 minutes on providing us with some performance numbers? So tomorrow you will provide these short measurements, thx! It’s just copy&paste and you’re done in less than 5 minutes.

Since the A83T is such a fast 2 GHz beast both sysbench and 7-zip should finish within a minute :clap: :clap: :clap:

Really looking forward to the results.

And while we’re already talking about you the manufacturer being so kind and open regarding your information policy: For you as the ‘open source company’ it must be an honour to demonstrate your openness by providing a complete boot log on pastebin.com, yes? Just let one of your devs attach a serial console and log the whole stuff until the Ubuntu image shows the login prompt and put it online so that we (your beloved customers) can already have a look what to expect. This will look then similiar to eg. this here: http://pastebin.com/X2ppDEwS

1 Like

Good morning @sinovoip!

Still no time to spend the 5 minutes on providing us with some performance numbers? Or is performance that bad that you fear publishing results?

And regarding boot log: where is it? Or is this a confirmation that the M3 isn’t able to boot? Or do you lack a serial adapter to connect to the UART header? Should I send you one?