Thank you for making the ODROID-C4 available. I am very interested in this machine, because I've been thinking about building a cluster and ample 12V power is much more available on a standard ATX power supply than the 5V used by many other single-board computers. I ran two programs, the Pi Pie Chart benchmark and John McCalpin's stream memory test.
The pie chart benchmark I created. It, hopefully, consists of reasonably well written OpenMP parallel C codes to solve four standard problems: prime sieve, merge sort, fast Fourier transform and the Lorenz 96 dynamical simulation. More information may be found at
https://www.raspberrypi.org/forums/view ... 3&t=227177
including results for the ODROID-N2
https://www.raspberrypi.org/forums/view ... 5#p1469792
The results I obtained for the C4 were
Code: Select all
$ ./pichart-openmp -t ODROID-C4
pichart -- Raspberry Pi Performance OPENMP version 34
Prime Sieve P=14630843 Workers=4 Sec=0.881419 Mops=1060.03
Merge Sort N=16777216 Workers=8 Sec=0.853603 Mops=471.71
Fourier Transform N=4194304 Workers=4 Sec=1.4914 Mflops=309.356
Lorenz 96 N=32768 K=16384 Workers=4 Sec=1.12385 Mflops=2866.25
The ODROID-C4 has Raspberry Pi ratio=22.9131
Making pie charts...done.
with a corresponding pie chart given as
I think I was the only one using the C4 during my tests. I ran the pie chart twice and obtained similar results both times. The Fourier transform was faster than the Raspberry Pi 4B. This algorithm is very memory intensive and likely implies greater memory bandwidth.
After making the pie charts, I then ran John McCalpin's stream program for measuring memory bandwidth. The parallel version of stream was executed multiple times using taskset to specify the number of cores. The results were
This suggests the memory bandwidth of the ODROID-C4 is nearly double the memory bandwidth of the Raspberry Pi 4B. In my cartoon world this might be because the 4B gets by with only one RAM chip while the C4 employs two.
Again, thanks for making the ODROID-C4 available for testing.
