Re: NEC benchmarks/matrix transposition.

I made the matrix transposition change and compared the two codes on my 75meg Super Sparc and got about a 10% speed improvement. This is why I suggested people ought to come clean with mods they have made to the code when announcing they have done tests.

Here is something I think we could use as a benchmark.

CM
CM
CM
CM
CE
GW33,20,.0,-.025,.28,.0,-.025,-.89,.004,
GM1,2,0.,0.,0.,0.,.025,0.,33.,
GM1,5,0.,0.,0.,-.025,0.,0.,35.,
GM5,1,0.,0.,0.,.0,-.05,0.,36.,
GW49,25,.14,0.,-.19,.14,0.,.19,0.014,
GM1,1,0.,0.,0.,0.,0.,-.61,49.,
GM0,0,0.,0.,0.,.17,0.,0.,0.,
GW100,12,0.,-.75,.39,0.,-.75,-1.,.004
GM1,59,0.,0.,0.,0.,.025,0.,100.,
GE0,0,0.,
EX0,49,13,00,1.,0.,
EX0,50,13,00,1.,0.,
FR0,1,0,0,500.,0.,0.,
RP0,1,359,1010,90.,0.,0.,1.,0.,0.,
EN

I have trimmed down the number of segments slightly to try and get it to be around 1000 segments. It was previously about 1800 segments.

Regards J.
Received on Thu Apr 11 1996 - 06:30:00 EDT