Igor Schein on Thu, 4 Mar 1999 10:07:41 -0500

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

MPQS weirdness


I've encountered something which puzzles me.

In this example, I deal with 2 systems, Solaris@Ultra-60 and
Linux@P233. Benches for them are 5225 and 10500 respectively,
i.e. exactly a factor of 2. 

Now, I factorint(,6) the same 75-digit number on both machines at \g5.
Time between 5% checkpoints was 15m on Ultra-60 and 2h15m on PC!!
In other words, Ultra-60 is more than 9(!) times faster on that
particular task.  

Both machines were empty at the time of the run.  PC has 32M RAM, but
GP only uses less than 10M, so the memory is not an issue.  

>From what I understand ix86's integer arithmetic is in general faster
than Sparc's.  

I would like to hear some comments on this issue.  Are the benches