Run the same code 3 times, get the best result
Code | bb v0.10.163 | bb v0.10.164 | nbb | Lumo | Cherry | Squint | Shadow-CLJS Compiled |
---|---|---|---|---|---|---|---|
with-record | 11380 ms | 10480 ms | 26700 ms | 2410 ms | --- | --- | 438 ms |
with-vectors | 8960 ms | 6800 ms | 23150 ms | 3800 ms | 1680 ms | --- | 1320 ms |
with-js-vectors | --- | --- | 25910 ms | 3300 ms | 1070 ms | --- | 378 ms |
bb-optimized | 3670 ms | 3580 ms | 7200 ms | 3720 ms | 1760 ms | --- | 1390 ms |
For squint, it does compile but displays the wrong result; for Chery, it does not yet support defrecord
To have some comparisson, I changed the squint code in the last (bb-optimized
) example to replace bit_shit_left
with <<
and remove int$
so the code at least can run. I also compiled replacing doseq
's range
with (vec (range...))
- this gives me 352 ms