Re: Performance degradation over time on VAX...

After some thought, I realized that my data copy overhead was not a good on=
e.=0A=
=0A=
When I ran "dd if=3D/dev/rsd0c of=3D/dev/null bs=3D1m count=3D200" on the v=
s4k60,=0A=
the times were consistent from 4.0.1 to 11rc5 (all around 76 seconds).=0A=
=0A=
So I asked this question on Google:=0A=
=0A=
  Why does the Unix dd command run slower with a small block size?=0A=
=0A=
Here is a portion of the response:=0A=
=0A=
  1. High System Call (Syscall) Overhead=0A=
=0A=
  Every time dd reads or writes a block of data, it must execute a=0A=
  read() and a write() system call.=0A=
=0A=
  Context Switching: Each system call forces the CPU to switch context=0A=
  from user space to kernel space and back.=0A=
=0A=
  The Cost: If you transfer a 10 GB file using the default 512-byte=0A=
  block size, dd must perform roughly 40 million system calls. If you=0A=
  use bs=3D4M, it requires only 2,500 system calls. Minimizing this=0A=
  context-switching overhead dramatically increases throughput.=0A=
=0A=
I see that "lmbench" includes several tests related to system call=0A=
overhead as well as context switching.  So that looks like a good=0A=
tool to try (plus it covers a broad range of topics).=0A=
=0A=
Thank you for the suggestion!=0A=