Caitlin> This is not a question of extra work, just assigning the
Caitlin> responsibility to the correct layer -- that is to the
Caitlin> device specific code. This is a proper recognition that
Caitlin> the logic is in fact device dependent.
BTW in general I agree with this. But if we could get a measurable
performance win by moving the locking up a layer, and not penalize any
real hardware, then as I said I would be inclined not to worry about
hypothetical future hardware. As Michael said, we can always change
it back if we want to.
- R.
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
