cache affinity scheduling in new processors
cache affinity scheduling in new processors
- Subject: cache affinity scheduling in new processors
- From: Alejandro <email@hidden>
- Date: Thu, 13 May 2010 20:41:35 +0200
Hello,
I am using the thread affinity API.
I see that in the new processors (Nehalem, i7) there is a small L2
cache per core and a per processor shared L3 cache.
My app is not distributing threads among L2 caches by affinity in a
single processor machine (i7).
Is this because the cache affinity distribution policy does not work in
the new processors or because it considers affinity whithin the L3
cache instead? In the second case, I understand that a two processor
machine is needed, because the L3 is per processor.
Thank you,
Alejandro
--
|
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-kernel mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden