Oberon Benchmarks

Current (since olr151219.tgz) release ships with good old Hennessy3.pm.Mod benchmark suit. Here are the results for MIPS ci20 and ARM Cortex A7 Orange Pi PC computers, both running at 1.2 GHz. No results of original MIPS compiler as old olr versions don't run on current ci20 Debian. Dim3 is the graphics benchmark of Dim3.Panel in Gadgets.

RPi3 was benchmarked using old version of Ceres based compiler which is about 10% slower than new version.

MIPS ci20 MIPS ci20, 15GPR OPi PC, Ceres compiler OPi PC, ROP2 compiler OPi PC, ROP2, MaxVarR=4 RPi 3, old Ceres compiler Haswell @ 3.2GHz Haswell @ 1.2GHz
Perm 23 41 31 30 24 28 4 9
Towers 40 49 34 39 38 33 3 9
Queens 22 34 22 23 20 22 2 5
Intmm 28 45 30 29 17 27 2 7
Mm 36 52 33 30 19 31 3 7
Puzzle 213 337 210 240 182 206 26 77
Quick 20 40 28 27 21 28 6 16
Bubble 50 77 57 55 43 59 7 14
Tree 27 36 29 25 23 28 7..21 37
FFT 55 98 60 55 42 60 7 32
NonFP: 5.10E+02 7.76E+02 5.39E+02 5.53E+02 4.52E+02 5.29E+02 7.13E+01 2.42E+02
FP: 8.59E+02 1.36E+03 9.02E+02 8.85E+02 6.94E+02 8.86E+02 1.14E+02 3.67E+02
Dim3 1.34s 1.76s 1.04s 0.95s 0.88s 0.14s 0.43s

The result clearly shows that MIPS and Oberon is a perfect match. This single issue MIPS CPU sometimes is nearly as fast as a highly complex Haswell CPU and nearly always faster than 1 1/2 issue ARM Cortex A7. Reducing register set from 31 to 15 significantly decreases performance. While moving from HardFload to VFP gives a significant speed increase, direct use of VFP registers surprisingly gives only a small speed gain.

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License