float V double

Try it out. Just make two simple loops with some calculations (one for floats, one for doubles) and print the time they took.

1) all operands are converted to 80 bit extended format
2) doubles are 2x as big => more memory pressure
3) speedwise, the only difference is FPU precision: double fdivs and fsqrts are ~25% (estimate) slower.

civguy: Zimbabwe? Did''t it use to be ''somewhere in northern europe''?

Sorry for hijacking the thread but how would you go about calculating the time it took? specifically, what function would you use? (yes i''m new to this ).

You can use getClockTicks, or clock(), or QuerryPerformanceTimer... or anything else you''d like to use. The longer you let it run (more iterations) the more accurate your results will be .

As I don''t know C++ I can only give you the psuedocode:

-Record the current system time in milliseconds
-Run a loop with the calculation a known number of times
-Again, record the number of milliseconds
-Subtract the start time from the end time and divide by the number of loops
-This will give you the time taken to perform one calculation

---

#Old Steve, he said to Xerox "Boys, turn your heads and cough"
And when no-one was looking he ripped their interfaces off#
Three Dead Trolls in a Baggie on the truth behind MacOS 1

SketchSoft | SketchNews

---

quote:

---

Original post by Jan Wassenberg
civguy: Zimbabwe? Did''t it use to be ''somewhere in northern europe''?

---

Sshh.. They might be watching