For me 730 mA for a routine I wrote in C is pretty close to the theoretical maximum of 900. Since that program was designed to perform a realistic computation, a code specifically engineered to draw more power would likely do so.64-bit Pi OS of late 2024 isn't the same as the 2021 version, and I guess that a difference in the stress test itself or even the USB power check hardware can cause some variation in results. For the power check I used this: https://www.amazon.com/Eversame-Multime ... B07JYVPLLJ. And I explicitly ran sysbench on all 4 cores, so no holding back on power draw there:by ejolson » Mon Dec 23, 2024 1:28 am
On a test I wrote the Zero 2 consumes 730 mA in 64-bit mode but only 580 mA in 32-bit mode.
https://forums.raspberrypi.com/viewtopic.php?p=1930593#p1930593
It seems both of these are more than sysbench.Still puzzled though... why would you like a USB hub with more power if none of the nodes in your cluster would ever draw anywhere near the 900 mA this 3.0 hub can provide, assuming no hardware added.Code:
#!/bin/bashclearecho '** Starting at:'vcgencmd measure_tempfor (( i=1; i<=10; i++ ))doTIMEFORMAT='** Sysbench CPU test took %R seconds.'time { echo '** Run '$i sysbench --test=cpu --cpu-max-prime=20000 --num-threads=4 run >/dev/null 2>&1 vcgencmd measure_temp}done
I explained to the dog developer that a possible attack during the first cyber war might involve simultaneously running malware designed to draw maximal power on all the usually idle home and office computers. Fido growled in reply that half those machines would crash and shut off due to inadequate power supplies and cooling and thus save the country's electrical grid. On the other paw, malware coordinating the electronically controlled flush toilets in a city could spell disaster.
Not wanting to dwell on spelling, I quickly changed the topic back to Raspberry Pi clusters.
Statistics: Posted by ejolson — Mon Dec 23, 2024 2:46 pm