I see same discrepancy in speeds here:I think those servers spin the fans at full when booting as a preventative measure to clear out the accumulated cat hair. For the two-socket server here I getI created new short "Jet sound in basement, 8-socket server starting" for those who do not:
https://www.youtube.com/shorts/KBpVzTYbr-g
Code:
# dmidecode -t memory | grep "Speed" | sort | uniq -c 8 Configured Memory Speed: 1866 MT/s 8 Speed: 2133 MT/s
" for those who do not:Code:
hermann@x3950-X6: ~$ sudo dmidecode -t memory | grep "Speed" | sort | uniq -c 32 Configured Memory Speed: 1600 MT/s 32 Speed: 2133 MT/shermann@x3950-X6:~ $https://www.intel.com/content/www/us/en ... tions.html
Memory Types : DDR4-1333/1600/1866, DDR3...
... as v4 has same restrictions:
https://www.intel.com/content/www/us/en ... tions.html
Memory Types: DDR4-1333/1600/1866 DDR3...
Even my favorite 24-core 8890v4:
https://www.intel.com/content/www/us/en ... tions.html
Memory Types: DDR4-1333/1600/1866 DDR3...
You cannot change distance function to float, that is likely to return different optimum and the rules of mona-lisa100K.tsp need to be followed. Also your own random generator does not allow to compare apples to apples. I will try to change both and get your code do exact same computation.
P.S:
Tool validate_all allows me to compare normalized and see any unwanted code difference:
https://github.com/Hermann-SW/RR/blob/m ... lidate_all
Code:
pi@raspberrypi5:~/RR/tsp/pthread/all $ ./validate_all nohup: ignoring input and appending output to 'nohup.out'pi@raspberrypi5:~/RR/tsp/pthread/all $ Statistics: Posted by HermannSW — Thu Aug 28, 2025 5:49 pm