Well, the thing that hurts the 83820 is that its interrupt
mitigation capabilities are rather limited. This is where napi
helps: by turning off the rx interrupt for the duration of packet
processing, cpu cycles aren't wasted on excess rx irqs.
As to the lack of bandwidth, it stems from far too much interrupt
overhead and the currently braindead attempt at irq mitigation.
Since the last time I worked on it, a number of potential techniques
have come up that should bring it into the 100MB realm (assuming it
doesn't get trampled on by ksoftirqd).
-ben
-- "A man with a bass just walked in, and he's putting it down on the floor." - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/