Quote:
Originally Posted by BitchIAmAMartian
Oh indeed it does. that helps a lot thx. I also wanted to ask how well gto+ scales with multiple cores/thread? im thinking of buying either a 5900x or 5950x and the price difference is 46% for 25% more cores and multicore cinebench performance. Im just wondering if this scales similarily in gto +? because if it doesn't 1 to 1 it might not be woth the money for me.
Your performance metrics are based on first-part results. You ought to wait for the likes of Gamers Nexus to release their own independent results and analysis.
I however can speak to the 3900x scaling well with GTO+ however even though cinebench is the best synthetic benchmark to compare when considering a CPU for poker solving, it is not linear comparison because cinebench is very synthetic and loads up the CPU residency up to 100% for the entire duration of the benchmark. GTO+ does not and this is more clear with the more cores and threads you have. Also, GTO+ has a lot of variance which can be seen if you solve the same tree over and over, you will get slightly different solve times.
But consider this, IF you can save 25% off of your solve times, calculate what 25% off of 2 hours is only half an hour, with perfect scaling which you wont get. This will scale as your databases get larger and the savings in time might be more valuable but you can also just wait a little longer the results (solutions) are the exactly same. It really depends on how big and complicated your trees and databases get and how often you are solving. If you are like me and start off with GTO ranges and solve a subset of flops to work with, then you aren't going to be re-running solutions often, and then the core count doesn't matter so much as even a quad core with no multi-threading will work, given enough time. Just how the world record for solving Pi was done on old hardware and beat out google with their multi-million dollar data-center. All it took was more time (and a bunch of memory).
Last edited by MadCat; 10-21-2020 at 11:13 AM.