The one game that can effectively use 12 cores (6 actual)!

Started by Mark_Kratzer, May 23, 2015, 09:03:06 AM

Previous topic - Next topic


The is Deep Fritz 12, a chess engine, I have it running on all processors.  I set it to idle, since it will get 97% of the machine anyway.

Now, of course I don't need this type of parallelism for playing chess.  I am just an amateur; not a grand master.  But this is great for post game analysis.  You can get a very detailed analysis of your game and the use of multiple cores vastly reduces the amount of time needed to carry this out.


On my old PC from 2007, it took about 10 minutes to analyze one of my games looking 15 moves ahead at each board position.

On this machine, such level analysis is only 15-20 seconds.  Of course, I could certainly go deeper, but at my level of play that is not necessary.

Jeremy Collake

That is a good app for benchmarking the CPU in a real-world scenario, and may make for another good real-world example of ProBalance in action -- if you hadn't already set it's priority class to Idle, causing ProBalance to ignore it.

Thanks for bringing it to our attention!
Software Engineer. Bitsum LLC.


Any chess engine with the word "Deep" in it is the multi-threaded version that will consume every cycle your machine has.  If it does not contain the word "Deep", then it is the single processor version.

You can demo your product with a chess engine in three areas:

(1)  Post game analysis.  This is very common and is a CPU hog.

(2)  Infinite position analysis.  Just looking on one board position but real detailed.  CPU hog.

(3)  Engine vs Engine game.  There are different engines and you can compare them, by having them play against each other.  CPU hog.

The various game engines go for around $80 USD a pop.  Expensive for a casual chess player, but peanuts for business doing marketing.


I am using an older version of Deep Fritz 12 which is 32 bit.  They now have it in 64 bit.

Chess engines can also suck memory.  They use the memory as a heuristic.  There are multiple ways to arrive at the same board position.  By keeping track of which board positions have already been analyzed (transposition tables), CPU load can be reduced by not repeating the same work.

So, I might be interesting to see how a hoard of these threads would gobble memory in a 64 bit environment.  However, you can in the engine setup define the amount of memory is should use.

Jeremy Collake

Thanks, we may indeed add this to the real-world demo list, something we need to better document.

One great example is Sony Vegas, or most any multimedia editing software. When it goes to save the changed video, it will consume 100% of available CPU cycles for at least 8 threads (probably can do more parallelism even). It is amazing to see the difference when ProBalance acts on it while it is saving, which can take hours on some PCs. It is literally the difference between having a usable PC, or not, while saving. With ProBalance, all is fluid. Without it, Sony Vegas will eat up too many CPU cycles for most systems to handle without mouse lags, application stalls, etc..
Software Engineer. Bitsum LLC.