[OSM-dev] Osmosis crashing due to lack of memory

marcus.wolschon at googlemail.com marcus.wolschon at googlemail.com
Thu Feb 26 12:57:15 GMT 2009


On Thu, 26 Feb 2009 12:15:44 +0000, Emilie Laffray
<emilie.laffray at gmail.com> wrote:
>> Using a gzipped planet input and gzipped output may be quicker if your
>> facing an IO bottleneck... the compression and decompression will use
>> more
>> CPU, but, significantly reduce the IO...
> 
> In this case, it is clearly obvious that I am CPU bound. One core of
> the CPU is clearly maxing out and the other is not used. I was just
> trying to think of a way of speeding up the entire process. The hard
> drive that I am using is a standard Sata drive and I have another one
> if needed connected on a different controller. I have a dual core
> processor. I am just trying to be as smart as possible in order to
> optimize the long times.

Are you sure to be using DMA for HDD-access?
It should be IO-bound for such simple tasks.

What do you mean with "maxed out"? "99% user" or
just just "0% idle"?
IO-processes (Windows: [x]show kernel times) use
cpu-time too.

(It's sad that Java-NIO does not get async file-io until
 Java7.)

Marcus




More information about the dev mailing list