On Mon, 28 Feb 2000, Stephen Satchell wrote:
> Again, I beg to differ. Having worked on design of high-speed
> communications systems and in particular dealing with real-time issues, I
> can tell you that limiting interrupt count can be very, very
> important. For an early microcomputer system, we had to deal with just the
> sort of issue that you are talking about.
>
[more history deleted]
I _have_ measured the number of interupts per second that a PC could
withstand running Linux; This is very recent.
I was able to get something around 90K/sec when forwarding without utterly
slowing down the system; At 64byte packets thats close to 100Mbps; at
640Byte packets thats close to 1Gigabit. This is of course without turning
on Alexeys fast forwarding core. I think at some point the PCI bus becomes
your bottleneck.
Note also that this is not your normal packet load; This was generated by
a smarbits which has very tight inter-packet time. So i could probably
easily do more than 100Kpps;
Your traditional router does not use interupts; they poll the device
instead. This way they have more control.
If you use interupt batching it just means the packet may end up spending
more time on your node because it does not get processed ASAP. Thats my
contention. I dont see how you can argue against that ;->
cheers,
jamal
-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to [EMAIL PROTECTED]
