From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1758313Ab3KZWSW (ORCPT ); Tue, 26 Nov 2013 17:18:22 -0500 Received: from mx1.redhat.com ([209.132.183.28]:36165 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757423Ab3KZWSS (ORCPT ); Tue, 26 Nov 2013 17:18:18 -0500 From: riel@redhat.com To: linux-mm@kvack.org Cc: linux-kernel@vger.kernel.org, mgorman@suse.de, chegu_vinod@hp.com, peterz@infradead.org Subject: [RFC PATCH 0/4] pseudo-interleaving NUMA placement Date: Tue, 26 Nov 2013 17:03:24 -0500 Message-Id: <1385503408-30041-1-git-send-email-riel@redhat.com> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org This patch set attempts to implement a pseudo-interleaving policy for workloads that do not fit in one NUMA node. For each NUMA group, we track the NUMA nodes on which the workload is actively running, and try to concentrate the memory on those NUMA nodes. Unfortunately, the scheduler appears to move tasks around quite a bit, leading to nodes being dropped from the "active nodes" mask, and re-added a little later, causing excessive memory migration. I am not sure how to solve that. Hopefully somebody will have an idea :)