[tcpm] Re: [e2e] Are you interested in TOEs and related issues (Resend)

Craig Partridge craig at bbn.com
Sun Mar 7 09:43:26 PST 2004


In message <510B9F0C-6DAB-11D8-AE57-000A95D7D4C0 at cray.com>, David Borman writes
:

>I wouldn't say in general that we found out that TOEs didn't work.  I 
>think a more qualified comment of what we found out was that in an 
>environment where the network is the bottle neck, TOEs did not provide 
>any cost effective benefits; and as CPUs got faster, the TOEs just 
>provided costs without benefits. :-)

I take a stronger view, but also agree we need to clarify.

>TOE has become a buzz-word that needs to be qualified.  Some folks 
>include Checksum Offload and Large Segmentation Offload when talking 
>about TOE.  When others talk about TOE, they refer to offloading the 
>whole TCP/IP stack.

I understood TOE as being offloading just about the entire stack.

I would agree that experience shows that certain functions can be offfloaded
very effectively, including checksums and some memory management.

I also agree with your thought question -- namely which functions to offload
in a high performance environment -- is the right one.

Craig




More information about the end2end-interest mailing list