[Cbe-oss-dev] elaboration of Axon driver?

[Marcus G. Daniels]

Arnd,
> My current plan is to retire the current driver on the long run, and
> move to the standard ib-verbs interface with a new device driver
> implementation that also supports other systems than just the triblade.
[..]
> IFAICT, DAPL is basically dead for new development, but there is already
> experimental DaCS code on top of ib-verbs.
Ah.  It would make DaCS more appealing if it could be used on other 
kinds of IB-based clusters...
I wish it were easier to understand exactly what DaCS is doing.   For 
example, even with the raw Axon driver, when I swap from having the 
Opteron be the local and the Cell be the remote, to having the Cell be 
the local and the Opteron the remote, I get an error on the Cell side 
that says "Bad remote PLB5 address 0xa00000a500000018".   That seems to 
be some magic address of the Axon, but if I'm mis-using the interfaces 
it isn't obvious how! 

Cheers,

Marcus