I remember (a long time ago) digging around for a
computer to test one of the large commercial Smalltalks - VA IIRC. It did
not go well, and in the process of checking on how the machine performed
otherwise, I ran something out of the then-current M$ Office and quickly
discovered that the particular machine was not up to my needs. The
thing just crawled, and the interesting part was watching it draw, several
times(!!), one of the common dialogs. It was quite a display of
inefficiency, and was repeated with every user interaction on that dialot.
The point being that faster machines were able to overpower the bad code, which
is pretty much what has kept MS in business.
Byte codes and message sends are important, but so is efficiency and feedback to the user.
I think it has no sense to test with several images. Look adrian
comment
"This basically measures how many bytecodes and messages the VM
can
execute per second."
So, I guess it is a benchmark for VM +
hardware. I don't think the image can influence much here. Am I
correct?
best,
Mariano
On Thu, Jul 16, 2009 at 1:38 PM, Esteban A. Maringolo
<emaringolo@gmail.com>
wrote:
Here
are my numbers.
Computer: Thinkpad T61,Intel Core 2 Duo T7300 @ 2GHz, 2GB
RAM)
OS: Windows XP SP3
Seaside 2.8.3 One Click
Image:
'157248157 bytecodes/sec; 4708854 sends/sec'
Seaside
2.8.4 One Click Image:
'158907510 bytecodes/sec; 4592283
sends/sec'
pharo0.1-10373web09.07.2:
'156479217 bytecodes/sec;
4413998 sends/sec'
Regards,
Esteban A.
Maringolo