Page 2 of 2

Re: Hyperthreading vs performance

Posted: Fri Mar 28, 2014 5:36 pm
by meteoadriatic
Interesting!

Re: Hyperthreading vs performance

Posted: Wed Dec 31, 2014 6:02 pm
by wilsors
So you've demonstrated multi-threading causes the WRF modeling to slow down, but have we tried to run two separate instances of WRF simultaeously, where the second instance is using the second available hyperthread core? IE, allocating adjacent processing with those same HT cores with two separate model runs, 4 on each. This way, we aren't wasting the HT capability.

Edit: I think this may have to be done in a VM environment