On 2023-10-13, at 4:16 PM, Tim Rowledge tim@rowledge.org wrote:
The insta-benchmarks, as opposed to staid, careful, ones, report 70% of my 2012 iMac 3.9GHz i7 at 1.6billion bytecode/sec and 140 million sends/sec.
Running the Shootout benchmarks across a few machines is interesting. That's four tests named nbody, bintree, chredux & thread.
Pi5 times at 5.1, 3,4, 7,2 & 8,4 secs respectively. My iMac reports 3.8, 2.1, 4.2 & 6.2 which means Pi5 is so close to 70% we might as well use that as a round figure; my other i7 linux machine reports near identical results.
So Pi 5 is a bit over 2.5 times faster than a Pi 4overall which is interesting because it hold true for pretty much every generation step. Add in the jump from the original 2013 era interpreter that we had for the very first Pi to the current cog and we get ratios of (wait for it)
160, 147, 81 & 37
between 2013 and 2023 systems. This Pi is about double the dollar cost of the earliest ones with 32 times the memory and all the other improvements.
The good news is that the 'XWayland' app catches all the old X code and handles it without any obvious issue.
Whoops, turns out my Pi isn't configured to use wayland. I'll try it out sometime.
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Strange OpCodes: KFP: Kindle Fire in Printer