[Beowulf] [EXTERNAL] Re: anyone have modern interconnect metrics?
Lux, Jim (US 3370)
james.p.lux at jpl.nasa.gov
Wed Jan 17 23:16:20 UTC 2024
To a certain extent, faster Ethernet is more likely to be a commodity – and at rates above 1 Gbps, there’s substantial “art” in making a PHY that works reliably. At the 10G speed, there’s things like RapidIO and SRIO, but they
1. Only work for short distances (<<1 meter)
2. Are *very* board layout and other implementation sensitive. Fine for getting in and out of a package, but not great for running any distance.
Then there’s XAUI (pronounced Zowie!) which is a multiwire wire interface between logic and 10G (or whatever) PHY. But it’s got the same problems as SRIO/RapidIO (or for that matter, the venerable (now) TLK2711 SERDES).
10G and 40G Ethernet do actually work over distances of meters, and over some moderate range of temperatures, and are likely to meet EMI/EMC requirements.
It is interesting that there doesn’t seem to be the same commercial pressure for optical versions. They all exist, but typically as modules you’d slide into your switch, not components you’d solder to a board. And there are plenty of XAUI->optical kinds of interfaces. And optical cables are cheap and relatively rugged.
From: Beowulf <beowulf-bounces at beowulf.org> On Behalf Of Scott Atchley
Sent: Wednesday, January 17, 2024 7:18 AM
To: Larry Stewart <stewart at serissa.com>
Cc: Mark Hahn <hahn at mcmaster.ca>; beowulf at beowulf.org
Subject: [EXTERNAL] Re: [Beowulf] anyone have modern interconnect metrics?
While I was at Myricom, the founder, Chuck Seitz, used to say that there was Ethernet and Ethernot. He tied Myricom's fate to Ethernet's 10G PHYs.
On Wed, Jan 17, 2024 at 9:08 AM Larry Stewart <stewart at serissa.com<mailto:stewart at serissa.com>> wrote:
I don't know what the networking technology of the future will be like, but it will be called Ethernet.
- unknown (to me)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://beowulf.org/pipermail/beowulf/attachments/20240117/7228eaa5/attachment-0001.htm>
More information about the Beowulf
mailing list