On 2 November 2002 10:09, Andi Kleen wrote:
> > That depends on size. If you do huge memcpy (say 1 mb) it still
> > wins by wide margin. Not that we do such huge operations often,
> > but code can check size and pick different routines for small
> > and big blocks
>
> The kernel nevers does such huge memcpys. It rarely does handle any
> buffer bigger than a page (4K)
Okay I take that.
How did you determined that movntXX stores are net loss?
I thought about that and didn't came to a working solution.
It's easy to time memcpy() but harder to measure susequent
cache misses when copied data gets accessed. We can read it
back after memcpy and measure memcpy()+read, but is entire
copy gets used immediately after memcpy() in real world usage?
We're in benchmarking hell :(
-- vda - 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/
This archive was generated by hypermail 2b29 : Thu Nov 07 2002 - 22:00:24 EST