I strongly suspect that the variations in timings are due to _ (infinity). _ occurs with higher probability in %?.>:200 200$200 than in %?.>:200 200$200000.
The timing variations are also hardware dependent. In Google, enter "Pentium 4 infinity multiply speed" and select the first search result (from Cyngus). ----- Original Message ----- From: Fraser Jackson <[EMAIL PROTECTED]> Date: Saturday, September 2, 2006 3:42 am Subject: [Jprogramming] Timing of some matrix products > The following give times of some matrix products using two forms. > > The first example uses integers. > > The remainder give reproducible results with a range of floating > point > numbers. All matrices are of the same shape. What is the reason > for the 30 > to 1 ratio of times depending on the values of the floating point > numbers > for the matrix form, and 7 to 1 ratio for the table form? > > Was I wrong to be surprised by these differences? > > loadd 'd:\j\j601w\temp\1.ijs' > ts =: 6!:2 , 7!:2 > > > a =: ?. >:200 200 $200 > f =: (+/ @: * )"1 1 > ts 'a +/ . * |: a' > 0.0302508 1.05024e6 > ts 'f /~a' > 0.0853203 531008 > > a =:% ?. >:200 200 $200 > ts 'a +/ . * |: a' > 0.625971 1.05024e6 > ts 'f /~a' > 0.721225 1.06029e6 > > a =:% ?. >:200 200 $2000 > ts 'a +/ . * |: a' > 0.0839973 1.05024e6 > ts 'f /~a' > 0.165096 1.06029e6 > > a =:% ?. >:200 200 $20000 > ts 'a +/ . * |: a' > 0.0277373 1.05024e6 > ts 'f /~a' > 0.106258 1.06029e6 > > a =:% ?. >:200 200 $200000 > ts 'a +/ . * |: a' > 0.0211127 1.05024e6 > ts 'f /~a' > 0.0991123 1.06029e6 > > > 9!:14'' > j601/beta/2006-08-23/07:45 ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
