Re: Scheduled Transfer Protocol on Linux

[Jeff Garzik]

Larry McVoy wrote:
> Err, if you had actually done this, you'd find that your statements
> are unsupportable in practice.  Please show me an application that has
> anything, even with an order of magnitude, like the number of locks
> taken/released per second in IRIX or Solaris.

Has anyone done any research on what I called 'NUPA' -- NUMA for
processors, basically ??

This is probably a totally crazy idea, but I was wondering how possible
it would be for Linux to seamlessly support, say, a PowerPC Mac or Sun
with an x86 accelerator card.  Or, to get even crazier, offload some
processing onto a GeForce or one of those video cards with dual Rage128
chips.

If 'NUPA' support existed, it seems like a prerequisite would be
eliminating many of the scaling problems typically associated with
OS<->hardware SMP interaction.  ie. creating spinlocks useable by only a
small portion of the available processors in the system.  and that's
only the tip of the iceberg.

-- 
Jeff Garzik         | "Vegetarian" is the Indian word
Building 1024       | for 'lousy hunter.'
MandrakeSoft, Inc.  |