Author: Robert Hyatt
Date: 12:34:30 05/22/04
Go up one level in this thread
On May 22, 2004 at 14:18:54, Anthony Cozzie wrote: >>Who knows. I use threads on SMP. On NUMA. There is no significant reason not >>to. And since Eugene's EGTB code is specifically designed for threads as well, >>the gain by having a large shared LRU cache of table data is also worthwhile. > >Actually, I think there is a very good argument for using threads over >processes, namely the new chip-multiprocessors due in year or two which will >probably have a shared L2 cache. > >I don't know how smart the OS is in terms of consistently scheduling threads on >the same CPU. I know when I run things on my dual opteron at home it >consistently bounces processes from CPU to CPU, which seems stupid even on an >SMP system (still have to transfer all that data from cache to cache). > >anthony Yes, but then that is a minor problem on NUMA boxes, in that no matter what you do, a particular page of memory containing instructions will exist on one local memory bank, and be remote to all the other processors. Of course, based on my testing with Crafty, the 1M L2 cache eliminates this as much of a problem. However, and this is something that Vincent doesn't know and I _hate_ to tell him, but if you run the _same_ program twice on any decent O/S, the instructions do _not_ get duplicated even if the processes are not threads. The O/S is smart enough to catch that so you have the _same_ performance penalty without threads (with respect to nearness of instruction memory to L2 cache) as you do _with_ threads. But don't tell him, it gives him another argument to look stupid over. :) when he repeatedly tells me over and over "threads are no good on NUMA". :)
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.