search

100trillionUSD / bitcoin

258 stars 65 forks source link

Add 4 improved versions #10

Open juestr opened 3 years ago

juestr commented 3 years ago

This most likely requires a bigger data file to evaluate the speedup, and will depend a lot on the data access pattern to d.

When stitching data_test to itself 32 times this cuts run time approximately in half, but obviously this is not a good test. YMMV.

FrancisBehnen commented 3 years ago

Nice! I was playing a bit with a pandas implementation. Could you maybe make a PR with just the benchmark?

juestr commented 3 years ago

Nice! I was playing a bit with a pandas implementation. Could you maybe make a PR with just the benchmark?

It's just a shell script (benchmark.sh) creating a larger data_test file and time'ing 4 script executions.