>>>>> "David" == David Mosberger <davidm@napali.hpl.hp.com> writes: >>>>> On Fri, 27 Feb 2004 18:44:38 +1100, Peter Chubb <peter@chubb.wattle.id.au> said: >>>>> "Scott" == Scott Feldman <Feldman> writes: >>>> Is this related to NAPI? >>>> I'll assume "the gigabit" card is e1000: Peter> Nope, DP83820. I'm benchmarking a driver that runs in Peter> userspace against the in-kernel driver; both benefit from the Peter> interrupt speedup. David> So why is irq enabling/disabling on the critical path for this David> driver? Because IRQs are disabled for every hardware interrupt, and enabled after each interrupt returns. So the latency between hardware interrupt and the driver seeing the interrupt *was* (prepatch) around 5 microseconds; minimum latency between interrupts was around 10 microseconds. This is time on a busy system to run out of receive buffers (you can get a new packet every 640ns with minimum-sized packets, and it takes time to process each packet) With the patch minimum time between interrupts should drop to around 2us, but I need to measure it again. -- Dr Peter Chubb http://www.gelato.unsw.edu.au peterc AT gelato.unsw.edu.au The technical we do immediately, the political takes *forever* - To unsubscribe from this list: send the line "unsubscribe linux-ia64" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.htmlReceived on Fri Feb 27 16:42:29 2004
This archive was generated by hypermail 2.1.8 : 2005-08-02 09:20:23 EST