Quote:
Am I right in thinking that this is slower than it should be, and any idea why this might be the case?
Well those Nash distances doesn't mean much if you don't include a reference point like starting pot and bets. 0.12 is decent with starting point of 100 and sizeable bets, not so much with with starting pot of 10.
Quote:
Here's a benchmark: solving the sample-turn example included in the software for 119 seconds yielded a Nash distance of 0.067.
Yeah, this is useful. My dev version gets to 0.061 in 10 seconds. The free one should be more or less 2x slower so probably 20-30 secs accounting also for CPU difference (I have a 3.4GHz i7).
Usually the problems with "this works slower than expected" our users reported so far were connected to:
-Cooling system is not working well: CPU fan is dying, radiator is dirty, the thermal paste is old/not functioning. To check for that you can see what's CPU temperature (download some thermometer, like this one:
http://openhardwaremonitor.org/) at rest is. It should be 40-50C. Where you are at it you can see if the solver uses all four CPUs when running.
-Stuff is running in the background: bazillion tabs of the browser, HM creating notes or importing hands, teamviewer etc.
-CPU is just slow. This is unfortunately quite possible. i5 is a very good CPU but it may be some laptop specific version which is slower. Also with only 4GB RAM you won't comfortably solve flops unless those are very simple. For more see this:
https://piosolver.myshopify.com/pages/technical-details
To be honest I wouldn't advise you buying the flop version now. Maybe if you only are interested in CAP spots and are willing to wait a lot. It won't be smooth experience though.