Skip to content

RFC autonuma: Optimize memory placement in memory tiering system

Huang, Ying linux-mm lkml 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 RFC Latest10 patches 11 repliesIn review

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/10autonuma: Fix watermark checking in migrate_balanced_pgdat()mm/migrate.c+1 -11 reply
2/10autonuma: Reduce cache footprint when scanning page tablesmm/mprotect.c+4 -41 reply
3/10autonuma: Add NUMA_BALANCING_MEMORY_TIERING modeinclude/linux/sched/sysctl.h kernel/sched/core.c kernel/sysctl.c+12 -90 replies
4/10autonuma, memory tiering: Rate limit NUMA migration throughputinclude/linux/mmzone.h include/linux/sched/sysctl.h kernel/sched/fair.c +2 more+82 -00 replies
5/10autonuma, memory tiering: Use kswapd to demote cold pages to PMEMmm/migrate.c mm/vmscan.c+24 -90 replies
6/10autonuma, memory tiering: Skip to scan fastest memorymm/huge_memory.c mm/mprotect.c+34 -100 replies
7/10autonuma, memory tiering: Only promote page if accessed twicemm/huge_memory.c mm/memory.c mm/mprotect.c+37 -150 replies
8/10autonuma, memory tiering: Select hotter pages to promote to fast memory nodeinclude/linux/mempolicy.h include/linux/mm_types.h include/linux/sched/numa_balancing.h +6 more+109 -204 replies
9/10autonuma, memory tiering: Double hot threshold for write hint page faultinclude/linux/sched/numa_balancing.h kernel/sched/fair.c mm/memory.c+6 -00 replies
10/10autonuma, memory tiering: Adjust hot threshold automaticallyinclude/linux/mmzone.h kernel/sched/fair.c+47 -45 replies

Reviews trailers per patch

2 of 10 patches have a Reviewed-by or Acked-by.

PatchMel Gorman
1/10A-b
2/10A-b
3/10
4/10
5/10
6/10
7/10
8/10
9/10
10/10

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v1 thread

Cover letter [RFC 00/10] autonuma: Optimize memory placement in memory tiering system

From: Huang Ying <[email protected]>

With the advent of various new memory types, there may be multiple memory types in one machine, e.g. DRAM and PMEM (persistent memory). Because the performance and cost of the different types of memory may be different, the memory subsystem of the machine could be called memory tiering system.

After commit c221c0b0308f ("device-dax: "Hotplug" persistent memory for use like normal RAM"), the PMEM could be used as cost-effective volatile memory in separate NUMA nodes. In a typical memory tiering system, there are CPUs, DRAM and PMEM in each physical NUMA node. The CPUs and the DRAM will be put in one logical node, while the PMEM will be put in another (faked) logical node.

To optimize the system overall performance, the hot pages should be placed in DRAM node. To do that, we need to identify the hot pages in the PMEM node and migrate them to DRAM node via NUMA migration.

While in autonuma, there are a set of existing mechanisms to identify the pages recently accessed by the CPUs in a node and migrate the pages to the node. So we can reuse these mechanisms to build mechanisms to optimize page placement in the memory tiering system. This has been implemented in this patchset.

At the other hand, the cold pages should be placed in PMEM node. So, we also need to identify the cold pages in the DRAM node and migrate them to PMEM node.

In the following patchset,

[PATCH 0/4] [RFC] Migrate Pages in lieu of discard
https://lore.kernel.org/linux-mm/[email protected]/

A mechanism to demote the cold DRAM pages to PMEM node under memory pressure is implemented. Based on that, the cold DRAM pages can be demoted to PMEM node proactively to free some memory space on DRAM node, so that the hot PMEM pages can be migrated to the DRAM node. This has been implemented in this patchset too.

The patchset is based on the following not-yet-merged patchset:

[PATCH 0/4] [RFC] Migrate Pages in lieu of discard
https://lore.kernel.org/linux-mm/[email protected]/

This is part of a larger patch set. If you want to apply these or play with them, I'd suggest using the tree from here.

    http://lkml.kernel.org/r/[email protected]

With all above optimization, the score of pmbench memory accessing benchmark with 80:20 read/write ratio and normal access address distribution improves 116% on a 2 socket Intel server with Optane DC Persistent Memory.

Best Regards,
Huang, Ying