I decided to keep the J4105 laptop, and I really like it — I did even 3Dprint holder and shoes for stationary use with higher display:...
I tried to run passmark benchmark, but was not able to install necessary libncurses5 (only in "Try Ubuntu").
Then I remembered pichart and downloaded and unpacked that.
After sudo apt install of gcc and make I did build without issues.
And what I got is in middle (43.019) between Pi4B (31.42) and Pi5 (58.3), which is OK for me.
Of course the laptop cannot be compared with my AMD 7950X (709.8) headless PC at top of pichart.
https://www.printables.com/model/845234-laptop-stand
Last night I finished PARI/GP script that determines "sum of two squares" for 323,228,467 decimal digits composite factor of Fermat number F30(=2^2^30+1), with 2 known prime factors, in 3.85s on AMD 7950X. Then I used that test and just reading Fermat.gp and running the init code while reading as "little benchmark" and compared running on 7 CPUs at my home:
https://gist.github.com/Hermann-SW/df18 ... nt-5048272
I have to analyze why the "F30 comp" time for Pi5 is so much worse compared to "read&init" time.
Since I installed Ubuntu on the SSD, I was now able to install Passmark benchmark (for Arm on Pi5 and x86_64 on J4105).
https://www.passmark.com/products/pt_linux/download.php
On the left side are the results for Pi5, on the right for J4105. I had forced the Pi5 to 3GHz (that is my default since months). The J4105 has 1.5GHZ default frequency, but as reported by Passmark can do 2.5GHz burst as well.
My "little benchmark" was single core, but Passmark "CPU Single Threaded" numbers say that 3 GHz Pi5 is better than J4105, but "little benchmark" shows that this is not true for the PARI/GP computations?!?!?
Statistics: Posted by HermannSW — Tue May 07, 2024 2:31 pm