Asymptote 0.7

[malu]

I just tagged v0.7 of Asymptote.

In comparison with the step from v0.5 and v0.6, the strength increase between v0.6.3 and v0.7 is rather modest.

Code: Select all

ELO   | 82.10 +- 13.61 (95%)
SPRT  | 10.0+0.1s Threads=1 Hash=8MB
Games | N: 1750 W: 789 L: 383 D: 578

ELO   | 91.18 +- 13.02 (95%)
SPRT  | 60.0+0.6s Threads=1 Hash=64MB
Games | N: 1750 W: 762 L: 313 D: 675

The most significant changes are improved time management and introduction of singular extension (which still has some inefficiencies). Most impressive (to me at least) is that the codebase shrunk by a whopping 60 lines between v0.6.3 and v0.7.

Many binaries are already up on Github. If there's a binary missing which you would like to use, either wait a moment and hope it does appear, or let me know

As always, thanks to anyone interested and testing Asymptote!
Let me know if you encounter any problems.

[Gabor Szots]

Thanks Max. On my Haswell PC Broadwell is a bit faster than popcnt. I'd use it if I were sure it won't crash owing to some instruction that Haswell does not recognize.

[malu]

Thanks Max. On my Haswell PC Broadwell is a bit faster than popcnt. I'd use it if I were sure it won't crash owing to some instruction that Haswell does not recognize.

Hi Gabor

Seeing that Broadwell is the "Tick" to Haswell's "Tock", I expect it to just work. I don't know for sure, but I don't think any of Broadwell's new instructions are used in the binaries, as they are mostly about video decoding. If you sucessfully ran a few test games and everything worked fine, I expect no problems.