[Pkg-openssl-devel] Bug#613490: Troubleshooting: Verification/Help Requested

[Scott Schaefer]

Hm .... now this is very (??) .. um ... "interesting"  (note this is 
stable 0,9.8-4squeeze1)

Assuming no silly user error, the code execution time has quite high 
variability.  High-level math and/or statistics is not my area of 
expertise.  However, given these results, do you agree it is plausible 
your results are simply due to algorithm variability ?  Can you test 
multiple runs on a machine to which you have access ?


Executing four consecutive runs for each bit-size 256-4096 ...

----------------------------------
#! /bin/sh

RUNS=4
for BITS in 256 512 1024 2048 4096
do
   for R in `seq 1 $RUNS`
   do
      time openssl dhparam -outform DER -5 -out parameters.dh $BITS > 
/tmp/dh${BITS}.$R 2>&1
   done
done
----------------------------------

Results  (Yes -- this is relatively old box):

*** 256 ***

1) 0.03user 0.00system 0:00.08elapsed 46%CPU
2) 0.71user 0.00system 0:00.78elapsed 91%CPU
3) 0.54user 0.00system 0:00.57elapsed 93%CPU
4) 0.44user 0.00system 0:00.49elapsed 91%CPU


*** 512 ***

1) 5.45user 0.00system 0:05.50elapsed 99%CPU
2) 4.53user 0.01system 0:04.60elapsed 98%CPU
3) 8.89user 0.01system 0:08.94elapsed 99%CPU
4) 4.92user 0.00system 0:04.98elapsed 99%CPU


*** 1024 ***

1) 17.50user 0.02system 0:17.59elapsed 99%CPU
2)  5.81user 0.00system 0:05.88elapsed 98%CPU
3) 40.87user 0.02system 0:40.95elapsed 99%CPU
4) 47.88user 0.02system 0:47.97elapsed 99%CPU


*** 2048 ***

1)  21.57user 0.00system 0:21.64elapsed 99%CPU
2) 284.93user 0.02system 4:44.98elapsed 99%CPU
3) 206.71user 0.02system 3:26.74elapsed 99%CPU
4)  20.87user 0.00system 0:20.92elapsed 99%CPU


*** 4096 ***

1) 18903.08user 0.40system 5:15:02elapsed 100%CPU
2) Still running after 5h 10m