1.67 GHZ G4 compares well to other CPUs in basic Pi test
This is a link to someones personal benchmark, but it appears to be honest, so I will post it here. According to this page (graph about the middle of the page), the Powerbook CPU is tremendous at crunching numbers in a pretty straightforward way. I noticed this when at my local Apple store the other day. The Powerbook seems a LOT faster than my P4 3 GHZ Alienware desktop at many tasks.
It is a program called SuperPi.
The G4 came out ahead of some pretty stiff competition.
I know the G4 is old news. It is nice for Powerbook owners (who must shell out some serious dough just to get the thing) to know that the G4 is still a serious competitor.
Link:
http://www.notebookreview.com/default.asp?newsID=2400
If this benchmark is any indication, the G4 does very well against the PM, too.
It is a program called SuperPi.
The G4 came out ahead of some pretty stiff competition.
I know the G4 is old news. It is nice for Powerbook owners (who must shell out some serious dough just to get the thing) to know that the G4 is still a serious competitor.
Link:
http://www.notebookreview.com/default.asp?newsID=2400
If this benchmark is any indication, the G4 does very well against the PM, too.
Comments
With two altivec units on a 1.67GHz chip, as is the case, you can theoretically get something in the range of 13+ GFlops if you design your mathematical program without your thumb in your ass. Incidentally, you can also get 26+ billion integer ops per second. Altivec rocks. The commoditization of software authoring killed it, though.
Originally posted by Splinemodel
With two altivec units on a 1.67GHz chip, as is the case, you can theoretically get something in the range of 13+ GFlops if you design your mathematical program without your thumb in your ass. Incidentally, you can also get 26+ billion integer ops per second. Altivec rocks. The commoditization of software authoring killed it, though.
First, I cannot find the Mac version of super pi. Second, how do you know that it relies so much on Altivec (except that this is the only logical explanation of the results)? Being a port, I would be surprised to find out that this is indeed true.
Originally posted by PB
First, I cannot find the Mac version of super pi.
OK, here it is. In the ftp server.
Version 2.0 of the super_pi for %s
Fortran source program was translated into C program with version 19981204 of
f2c, then generated C source program was optimized manually.
%s with compile option of "%s" was used
for the compilation.
-O3 -ffast-math -finline-limit=1000
It does not seems like Altivec, although I cannot exclude it since the optimisations are not detailed.
Originally posted by PB
I am not sure if the binary I found above is the same as the one in the Powerbook test. Anyway, the one I found has this:
Code:
Version 2.0 of the super_pi for %s
Fortran source program was translated into C program with version 19981204 of
f2c, then generated C source program was optimized manually.
%s with compile option of "%s" was used
for the compilation.
-O3 -ffast-math -finline-limit=1000
It does not seems like Altivec, although I cannot exclude it since the optimisations are not detailed.
It does not appear to be multiprocessor aware either. What the hell kind of benchmark is this.
yes, at this stage and going forward with mac os 10.4 iSteve has decreed "put all ye faith in accelerate.framework and be happy".