Hello michi, Sorry for late reply. comments are inline: On Sat, Apr 18, 2009 at 12:23 AM, Michael Blizek <michi1@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> wrote: > Hi! > > On 14:01 Fri 17 Apr , Devesh Sharma wrote: >> Hi michi, >> Its is using a TCP/IP protocol, program is Intel's MPI benchmark IMB, >> and CPU is surely not a bottelneck because its Quad core Quad socket >> machine with 64GB of physical mem. its a proprietary device for >> Systems Area Networks used for cluster computing. But the main problem >> that I have observed is, my device do not get any packets to post >> during this data size thats why I am trying to look beyond my device >> related functions. > > It seems like this benchmark is quite latency sensitive. > > Is the benchmark TCP based? If yes, you can try (as root): > echo "1" > /proc/sys/net/ipv4/tcp_low_latency this did not helped in my case...:( > > Is there any packet loss on the link? What does "ping -f" and packet loss is very less......after data transfer in TBits I observer on 5 tx packet drops > "ping -f -s 4096" say? How much does the latency increase if you send > bigger packets? Is there a "magic value" which causes a "sudden increase" > (e.g. 4051 is much slower than 4050)? Yes there is a magic number for packet size of 4040 bytes in ping -s 4040 gives proper results but of ping -s 4041 every alternate packet get delayed upto 1000 ms. > > Does this problem persist if the kernel is compiled without smp support (just > for diagnosis)? This I have not tried. > > -Michi > -- > programing a layer 3+4 network protocol for mesh networks > see http://michaelblizek.twilightparadox.com > > -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
