On 1/30/06, John Smith <[EMAIL PROTECTED]> wrote: > Hi, > > I find out some network card drivers (e.g. e1000 driver) use prefetch > instruction to reduce memory access latency and speed up data operation. > > My question is: > Suppose we want to pre-read a skb buffer into the cache, what is the > difference between using prefetch instruction and using a regular read > opertation? > 1. use prefetch instruction to stimulate a pre-fetch of the skb address, > e.g. prefetch(skb); > 2. use an assignment statement to stimulate a pre-fetch of the skb address, > e.g. skb1 = skb; > > I was told the data will be prefetched into a so-called prefetching queue > only by using prefetching instruction. Is this true?
prefetch won't stall the processor while data is fetched, i don't know of any "queue" as the data is fetched into cache. See the optimization manual at http://www.intel.com/design/pentium4/manuals/index_new.htm using prefetch is a difficult thing, as some recent threads on netdev should give you some more insight. Jesse - To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
