djerius
djerius
> feel free to bikeshed the actual name. it doesn't matter to me as long as it can be accessed easily & without loss in precision. [If your bikeshed is...
> > feel free to bikeshed the actual name. it doesn't matter to me as long as it can be accessed easily & without loss in precision. > > [If...
Here's a plot of user time vs N, where the code run is `zeroes(6,N)->dog`. This was run on an AMD Epyc 7763 w/ 128 cores & 638Gb, using 100 parallel...
I've attached a zip file with NYTProf output for N= 1_000, 10_000, and 100_000 [profile.zip.txt](https://github.com/PDLPorters/pdl/files/11083544/profile.zip.txt)
> This is really cool - how did you make it? For future reference, it's been added to the plot comment above.
> Thank you. That will be at least twice as valuable if you also include the rest, i.e. specifically how you turned those times into the plot shown? Check out...
Updated the plot in the above comment, just for kicks.
> Updated the plot in the above comment, just for kicks. Final update of the plot. Now the relationship looks really odd at higher N. Maybe there was process contention...
Rerunning as: ``` seq 1000 1000 1000000 \ | parallel --shuf -t --jobs 25 --joblog joblog \ taskset --cpu-list '{%}' \ /bin/time --format "'%E %U %S'" \ perl -MPDL -e...
@mohawk2 new results finished; see plot in comment w/ revised code. Much cleaner.