[Beowulf] let's standardize liquid cooling

Lux, Jim (337C) james.p.lux at jpl.nasa.gov
Fri Sep 28 14:25:57 PDT 2012

An interesting idea..
BUT.. you might need some sort of "bench test" fixture.. thermally driven convection might work.

And I think you'd have mfrs worried about people filling a rack full of fanless nodes and then not turning on the fans.  Or the power failing to the fans, or, opening both doors.

I suppose some sort of overtemperature cutout would solve this: the mobo would just shut down if it got too hot.

Whatever happened to a Beowulf made of tower cases piled on brick and board book cases?

Or this incredibly adaptable scheme: http://climate.ornl.gov/~forrest/linux-magazine-1999/  note the easy reference information inscribed on each node in this novel heterogenous cluster.  It's all part of addressing: "Management of such a large, heterogeneous cluster poses many challenges."

Jim Lux

-----Original Message-----
From: beowulf-bounces at beowulf.org [mailto:beowulf-bounces at beowulf.org] On Behalf Of Bill Broadley
Sent: Friday, September 28, 2012 2:15 PM
To: beowulf at beowulf.org
Subject: Re: [Beowulf] let's standardize liquid cooling

Sounds expensive, complicated, and challenging.

How about a MUCH simpler proposal: eliminate fans from compute nodes.

Nodes should:
* assume good front to back airflow

Racks would:
* have large fans front AND back that run at relatively low
  rpm, and relatively quiet.
* If front or rear door open the other door would ramp up fans
  (and noise) to compensate.

My measurements of 1U nodes have shown a significant fraction of the power budget goes to fans!  Fans that fit in 1U (even 2U) nodes seem to generate more noise, heat, and vibration than airflow.

Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf

More information about the Beowulf mailing list