[Beowulf] Opteron 235X: mobos & coolers

Tom Elken tom.elken at qlogic.com
Fri Apr 18 11:46:11 PDT 2008


Mikhail,

> We plan to add 4-cores Opteron 2354-based nodes.
> Is there some "contradictory evidences" for Tyan S2932G2NR
> http://www.tyan.com/product_board_spec.aspx?pid=453            ?

We've just started testing Opteron 2352 (2.1 Ghz) with Tyan S2912-E
http://www.tyan.com/product_board_detail.aspx?pid=547
and have had good success so far.

One difference between the 2912-E and what you specified below is that
the 2912-E has one PCIe x16 slot with x16 signal.  On 2932-E, both x16
slots have x8 signal.  With the x16 signal, you would have the option to
get DDR InfiniBand adapters that can get more peak bandwidth than with
the x8 signal.

With either the 2932 or 2912, you should specify that you want a
split-power plane mobo for maximum memory performance on 2300-series
(Barcelona) CPUs. Tyan indicates that with the -E suffix on the mobo
model #.
Here is the 2932-E if you like that better than 2912-E:
http://www.tyan.com/product_board_detail.aspx?pid=523
& http://www.tyan.com/product_board_spec.aspx?pid=523

-Tom

> -----Original Message-----
> From: beowulf-bounces at beowulf.org 
> [mailto:beowulf-bounces at beowulf.org] On Behalf Of Mikhail Kuzminsky
> Sent: Friday, April 18, 2008 10:54 AM
> To: beowulf at beowulf.org
> Subject: [Beowulf] Opteron 235X: mobos & coolers
> 
> We plan to add 4-cores Opteron 2354-based nodes.
> Is there some "contradictory evidences" for Tyan S2932G2NR
> http://www.tyan.com/product_board_spec.aspx?pid=453            ?
> 
> BTW, there is no AMD recomendations for 4-cores Opteron coolers.
> Is there some ideas which coolers are good for Opteron 2354 ?
> 
> Mikhail Kuzminsky
> Computer Assistance to Chemical Research Center
> Zelinsky Inst. of Organic Chemistry
> Moscow
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) 
> visit http://www.beowulf.org/mailman/listinfo/beowulf
> 




More information about the Beowulf mailing list