[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: Macs and 64-bit Memory Addressing
I need to correct myself a bit. I suddenly realized that Stata only have one
data type (double) that is 64-bit. This would mean that overall Stata
performance doesn't rely very much on 64-bit calculations. However,
calculations with doubles should be a lot faster with 64-bit code anyway.
Also, I think it would be a good idea to consider adding 64-bit integers to
Stata to take full advantage of the G5 (and others). With 64-bit integers,
it would be possible to use integers in many applications instead of floats.
For example, it would be possible to store currencies in an integer (as cent
or 1/100 cent) instead of a float (dollars and cent). I guess that
calculations with 64-bit integers would be more efficient than with floats.
Den 03-11-04 16.20, skrev "Michael Ingre" <Michael.Ingre@ipm.ki.se>:
> email@example.com <firstname.lastname@example.org>:
>> Anyway, let's remember that Apple kept the G5 under wraps for a while and
>> Metrowerks (who have been at the compiler game a lot longer than Apple and
>> saved Apple's bacon when the first PowerPC was released) will add G5 support
>> to their compiler eventually. And when they do, I expect the same sort of
>> minimal speed gains I'm seeing now with gcc 3.3 because the sort of
>> optimizations they perform are not of great benefit to Stata.
> I guess that you have made a good choice of sticking with Metrowerks. And
> probably, G5 support will be added to code warrior long before I have the
> means to get my own G5 from apple.
> However, what you're suggesting is a bit puzzling to me. I always thought
> that there would be a huge boost in performance (several times faster rather
> that a few percent) when calculating 64 bit data types in native 64-bit code
> compared to 32-bit code. And this would (presumably) greatly increase speed
> in programs like Stata that rely heavily on calculations with 64-bit data
> Apple suggest that calculations with 64-bit integer and floats cold be done
> in one clock cycle with the G5. But, they don't tell how many cycles a G4
> would need for the same operation.
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
* For searches and help try: