[Zope] Re: Intel vs AMD

[Myroslav Opyr]

Hi,

as Guido told:
> pystone uses time.clock(), which (on Unix/Linux) measures CPU time,
> not wall clock time.

Thus speed is pure processing speed and small decrease in pystones is 
extra expenses for process switching.

I'm updating list on http://plone.org/Members/interra/hardware
There is a bit of ambiguity in AMD processors clocking... I was trying 
to put not real frequency but its Intel equivalent, correct me if I am 
wrong.

m.
-- 
Myroslav Opyr
zope.net.ua * Ukrainian Zope Hosting
e-mail: myroslav@zope.net.ua

Myroslav Opyr wrote:
> I've found you as translator of pystones.py test into Python (from C). 
> Can you give explanation of the below fact of running several parralel 
> pystones.py processes and not affecting results. Is pure processing time 
> calculated?
> 
> Thank you,
> 
> m.
> -- 
> Myroslav Opyr
> zope.net.ua * Ukrainian Zope Hosting
> e-mail: myroslav@zope.net.ua
> 
> Marius Gedminas wrote:
> 
>> On Tue, Dec 10, 2002 at 09:44:18PM +0200, Myroslav Opyr wrote:
>> [ This was about a Dual-CPU Xeon, IIRC ]
>>
>>>> and running two pystones at once:
>>>>
>>>> [bkc@strader ~/Zope]$ Zope-2.6.0-linux2-x86/bin/python 
>>>> Zope-2.6.0-linux2-x86/lib/python2.1/test/pystone.py & ; 
>>>> Zope-2.6.0-linux2-x86/bin/python 
>>>> Zope-2.6.0-linux2-x86/lib/python2.1/test/pystone.py
>>>> [1] 500
>>>> Pystone(1.1) time for 100000 passes = 5.01
>>>> This machine benchmarks at 19960.1 pystones/second
>>>> Pystone(1.1) time for 100000 passes = 5.02
>>>> This machine benchmarks at 19920.3 pystones/second
>>>> [bkc@strader ~/Zope]$
>>>
>>> Effectively doubling perfomance...
>>
>> This is a *single* CPU Athlon (XP 1800+, 1.5 GHz)
>>
>> $ python /usr/lib/python2.2/test/pystone.py
>> Pystone(1.1) time for 10000 passes = 0.53
>> This machine benchmarks at 18867.9 pystones/second
>>
>> $ python /usr/lib/python2.2/test/pystone.py &\
>>   python /usr/lib/python2.2/test/pystone.py &\
>>   python /usr/lib/python2.2/test/pystone.py &
>>   python /usr/lib/python2.2/test/pystone.py
>> [1] 23242
>> [2] 23243
>> [3] 23244
>> Pystone(1.1) time for 10000 passes = 0.52
>> This machine benchmarks at 19230.8 pystones/second
>> Pystone(1.1) time for 10000 passes = 0.53
>> Pystone(1.1) time for 10000 passes = 0.53
>> This machine benchmarks at 18867.9 pystones/second
>> This machine benchmarks at 18867.9 pystones/second
>> Pystone(1.1) time for 10000 passes = 0.54
>> This machine benchmarks at 18518.5 pystones/second
>>
>> They all seem to run sequentially, and the numbers do not change even if
>> I try to run 32 of them in parallel.
>>
>> Marius Gedminas