Re: MPICH-GM in kernel space?

Ashish Bijlani wrote:
> Hi People,
> 
> Does MPICH-GM implementation exist as a kernel module? 

No. Out of curiosity could you detail for what purposes you would need such a 
kernel module.

> If no, then is
> it a good idea to implement it? 

Probably not. There would be major problems with such a port, including the 
fact MPICH-GM does not provide the ability to report all kind of errors to the 
MPI user (and aborting in the kernel is not a good idea), the configuration of 
MPI_COMM_WORLD might not easy to adapt to kernel use.

Moreoever if you want to use the MPI API in the kernel, I would suggest you 
base your work on MX rather than GM (see the announcement on the 
www.myri.com). It would probably be easier to implement minimal MPI wrappers 
around the MX API to provide the MPI subset you need rather than trying to 
port MPICH-MX. The basic MX communications primitives are very close the MPI 
ones. Probably the best solution would be to directly use the MX API.

> Also, please guide me to the document
> explaning how MPICH uses GM API.
> 

There isn't really any public one.


Loic