Subject: Re: Python Pystones
From: Troy Benjegerdes (hozer@drgw.net)
Date: Tue Jan 11 2000 - 18:15:39 MST
On Tue, 11 Jan 2000, Luke Tymowski wrote:
> Hello,
>
> What pystone benchmark results are people getting with G3s and G4s?
>
> The benchmarks I've seen suggest that x86 so far is the best platform for
> Python and Zope development. (I've got x86 and MIPS boxes, and I've seen
> results for Sun.) But I'll soon be getting another Linux box, and I'd like
> to get a G3 or G4. But if it doesn't run Python as well as x86 then there
> isn't much point.
>
> I've not yet come across any G3 or G4 Pystone results.
>
> I couldn't find anything on either the Yellowdog or Python sites.
Python is really great from fast prototypeing, but it really *isn't*
designed with fast execution in mind.. From everything I've read on it,
the idea is to develop a prototype in Python, and if it's slow, write
python modules in C or some other 'fast' language for the parts that are
slow.
As for benchmarks, here goes: (the PPC numbers are from Black Lab Linux,
which is using gcc-2.95.1)
G4/400:
Pystone(1.1) time for 10000 passes = 1.52
This machine benchmarks at 6578.95 pystones/second
G3/400:
Pystone(1.1) time for 10000 passes = 1.49
This machine benchmarks at 6711.41 pystones/second
G3/450:
Pystone(1.1) time for 10000 passes = 1.36
This machine benchmarks at 7352.94 pystones/second
Pentium III/500
Pystone(1.1) time for 10000 passes = 1.72
This machine benchmarks at 5813.95 pystones/second
I find it really strange that the G4 is slower than the G3, but I'm happy
to report they are all way faster than the Pentium ;). The PIII was
running a default install of Red Hat 6.1.
--------------------------------------------------------------------------
| Troy Benjegerdes | troy@blacklablinux.com | hozer@drgw.net |
| Unix is user friendly... You just have to be friendly to it first. |
| This message composed with 100% free software. http://www.gnu.org |
--------------------------------------------------------------------------
This archive was generated by hypermail 2a24 : Tue Feb 01 2000 - 17:50:57 MST