Actually not, at least for SETI apps. Maybe CUDA MB in some special circumstances will have some gain from setted affinity but it still not clearly shown. Affinity needed for hybrid AP indeed , just to optimize GPU feeding. But all our experiments with CPU-only SETI opt apps showed speed degradation with setted affinity. Probably OS really "knows better" when migrates thread to another core. Affinity needed for multithreaded apps, but AFAIK only Aqua has such app (and probably they use affinity lock already at own app level).
----- Original Message ----- From: "Charles Elliott" <[email protected]> To: <[email protected]> Sent: Thursday, March 11, 2010 8:06 PM Subject: [boinc_dev] BOINC Applications Speed-Up by Setting the ProcessorAffinity > If anyone is interested there is a good explanation for why BOINC > applications experience a speed-up > > when each application is assigned an affinity for one CPU (and by > implication why it does not work for > > single-CPU hyperthreading) here: > > > > Fedorova, A., S. Blagodurov, et al. (2010). "Managing contention for > shared > resources on multicore > > processors." Commun. ACM 53(2): 49-57. > > > > Essentially, one can infer from the article that the speed-up is due to > fewer cache misses and > > reduced contention for the front-side bus and the memory controller. > > > > > > Charles Elliott > > _______________________________________________ > boinc_dev mailing list > [email protected] > http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev > To unsubscribe, visit the above URL and > (near bottom of page) enter your email address. > _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
