JavaScript port of Sjeng

Gian-Carlo Pascutto · Post by **Gian-Carlo Pascutto** » Thu May 12, 2011 1:18 pm

Maybe one can parallelize the JavaScript with Web Workers.

uaf · Post by **uaf** » Thu May 12, 2011 1:28 pm

Werner · Post by **Werner** » Thu May 12, 2011 1:53 pm

Gian-Carlo Pascutto wrote:Maybe one can parallelize the JavaScript with Web Workers.

Ok, I will never ask so "silly" questions!

Shaun · Post by **Shaun** » Thu May 12, 2011 2:18 pm

If you don't want to sell 'Sjeng ct 2010' how about a donation to charity?

However a new commercial offering would be 'MOST' welcome

Shaun

mcostalba · Post by **mcostalba** » Sat May 14, 2011 8:22 am

Gian-Carlo Pascutto wrote:
yanquis1972 wrote:very pretty interface, what version of sjeng is this?
Sjeng 11.2, it seems.

After googling for some sensible script content:

Code: Select all

GCP&#58; this is straigth from Faile but converted to Sjeng internals

I've found this:

http://www.opensource.apple.com/source/ ... 0.3/sjeng/

Gian-Carlo Pascutto · Post by **Gian-Carlo Pascutto** » Sat May 14, 2011 9:59 am

mcostalba wrote: I've found this:

http://www.opensource.apple.com/source/ ... 0.3/sjeng/

Yes, Chess.App on every Mac has been using Sjeng for a long time. You can easily check if it you have a Mac by looking in the About menu.

gladius · Post by **gladius** » Sat May 14, 2011 4:08 pm

You joke about the webworkers, but I've considered doing that with GarboChess JS

. Currently they are just used to allow long thinking times without blocking the UI though.

Any idea what NPS the Sjeng JS implementation is getting?

Gian-Carlo Pascutto · Post by **Gian-Carlo Pascutto** » Sat May 14, 2011 6:26 pm

Do you actually get real multithreading with Web Workers or are they just emulated in the single JavaScript thread?

As for NPS: that's very browser dependent, of course. Firefox 4 and Chrome should give the best results.

gladius · Post by **gladius** » Sat May 14, 2011 7:46 pm

They are real threads, and have a decent cross-thread communication mechanism as well.

I've optimized on Chrome, so it's the best for me currently. Around 100k nps, with a very basic mobility eval. Of course nps doesn't mean a whole lot, but it's pretty cool that javascript can go that fast!

alpha123 · Post by **alpha123** » Mon May 16, 2011 12:16 am

Gian-Carlo Pascutto wrote:Do you actually get real multithreading with Web Workers or are they just emulated in the single JavaScript thread?

As for NPS: that's very browser dependent, of course. Firefox 4 and Chrome should give the best results.

According to MDN they are real OS-level threads, and section 1.2.6 of the spec seems to suggest that.

Opera's JavaScript engine is at least as fast as FF4.

Peter

JavaScript port of Sjeng

Re: JavaScript port of Sjeng

Re: JavaScript port of Sjeng

Re: JavaScript port of Sjeng

Re: JavaScript port of Sjeng

Re: JavaScript port of Sjeng

Re: JavaScript port of Sjeng

Re: JavaScript port of Sjeng

Re: JavaScript port of Sjeng

Re: JavaScript port of Sjeng

Re: JavaScript port of Sjeng