From mboxrd@z Thu Jan 1 00:00:00 1970 From: broonie@kernel.org (Mark Brown) Date: Tue, 17 Dec 2013 12:02:01 +0000 Subject: [PATCH 4/6] arm64: topology: Implement basic CPU topology support In-Reply-To: <52AFFB0B.3010505@linaro.org> References: <1386767606-6391-1-git-send-email-broonie@kernel.org> <1386767606-6391-4-git-send-email-broonie@kernel.org> <52AF11FE.70000@linaro.org> <20131216152235.GI3185@sirena.org.uk> <52AFFB0B.3010505@linaro.org> Message-ID: <20131217120201.GR3185@sirena.org.uk> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Tue, Dec 17, 2013 at 03:19:39PM +0800, Alex Shi wrote: > For topology meaning, it may be better to have cluster concept there. > And don't know if there will has a real ARM NUMA system for 64bit > server. If so, socket_id is good in a NUMA system. The scheduler does have a separate NUMA node ID that we're not currently using which I'd have expected to be a better fit for actual NUMA stuff; looking at the options we've got it's not clear to me that with what the scheduler has at the minute sockets aren't a good mapping for what the hardware has. There doesn't seem to be anything between threaded cores and physical packages. -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 836 bytes Desc: Digital signature URL: