On 081222 19:02, Allan Schrum wrote:
Would it be worth trying to heavily load one of your computers to see if the
problem presents itself differently? It is obvious that your systems are fast!
Allan, at your suggestion, I repeated the trials on Computer 1 (details
in previous post) under two opposite load conditions:
Condition 1: Maximum Load
All Cygwin trials performed during continuous run of Stress Prime 2004's
Blend Test http://sp2004.fre3.com/ , a torture test which I use for
stability testing when overclocking. The Blend Test stresses both CPU
and RAM and keeps CPU usage at 100%.
Condition 2: Minimum Load
All Cygwin tests performed with minimum other apps running: just one DOS
window (to run Cygwin tests) and one Explorer window (to keep track of
bar's size).
Running a DOS window (cmd.exe) with Example 1 of my original post
tr \32 \0 < foo | tr \0 \32 > bar
I ran 21 trials under each load condition with the following results:
Condition 1 (Maximum Load)
foo = 5,138,895 bytes
bar = 5,138,895 bytes (no error) on 1 trial
bar = 5,136,384 bytes (1254 x 4096) on 6 trials
bar = 5,132,288 bytes (1253 x 4096) on 14 trials
Condition 2 (Minimum Load)
foo = 5,138,895 bytes
bar = 5,138,895 bytes (no error) on 1 trial
bar = 5,136,384 bytes (1254 x 4096) on 2 trials
bar = 5,132,288 bytes (1253 x 4096) on 18 trials
As you can see, there was little difference in Cygwin's behavior under
opposite load conditions. If anything, Cygwin did better under Condition
1 (Maximum Load), but I don't think the difference is statistically
significant.
Greetings,
Lawrence
--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/
