netdev - Re: [PATCH] ipv4: move rt_genid to different cache line

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <1388779483.12212.129.camel@edumazet-glaptop2.roam.corp.google.com>
Date:	Fri, 03 Jan 2014 12:04:43 -0800
From:	Eric Dumazet <eric.dumazet@...il.com>
To:	Tom Herbert <therbert@...gle.com>
Cc:	David Miller <davem@...emloft.net>,
	Linux Netdev List <netdev@...r.kernel.org>
Subject: Re: [PATCH] ipv4: move rt_genid to different cache line

On Fri, 2014-01-03 at 10:01 -0800, Tom Herbert wrote:

> It seems to me you have more of an issue with prequeue itself. Maybe
> it is worth considering removing it to eliminate the disparate
> behavior between read and poll+read, but then removing a long lived
> optimization has is own pitfalls.

The prequeue stuff made sense in the old model of a few concurrent flows
and one thread per flow, where the gain was worth it. It is still
interesting if you have a single flow trying to fully utilize the
bandwidth of a 20Gbp or 40Gbp link (no dst false sharing in this case)

Nowadays, the hit ratio of prequeue mode is almost 0 on servers where
network performance of concurrent flows is an issue.

https://code.google.com/p/sockperf/ is a good step using realistic
benchmark, trying to mimic the behavior of current high performance
programs.

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html