Roger Turk wrote:
>
> Clarence,
>
> I think that Michael Dawley's tests, if he does them as you suggested, would
> clearly show the speed difference between 1.5x and 1.6x. Both of his
> computers are running with an 8 MHz motherboard, while if we try to compare
> a 486/586 with a Pentium, albeit a 75 MHz Pentium, the motherboard for the
> Pentium is going to be faster and give skewed results.
>
> Clarence Verge wrote:
>
> >>Now, to complete the experiment, you should put 1.50b2 on the '486
> and test them again. To be fair, 1.50b2 should REPLACE 1.60b1 - not
> just be added to the drive. There are a lot of factors which affect
> that kind of test.
>
> I believe the '486 will still win if there really was a LARGE margin
> before. Since I found only about a 20% diffference in speed, there
> must be some other factor at work here. (IMHO)<<
Hi Roger;
It is my understanding that Michael ran his initial test with 1.50b2 on
a Pentium - probably 75Mhz and 1.60b1 on a 50Mhz '486. Since the Pentium
system lost the race I have to assume that it was due to something other
than the speed of 1.50b2.
Most likely it would be the use of different Arachne.cfg files. Selecting
a Pentium.acf doesn't OPTIMIZE Arachne for a Pentium, it slows Arachne down.
The Pentium.acf REQUIRES a pentium just to be usable at reasonable speed.
Other possible differences like hard disk performance and cacheing can really
affect the test results.
Please understand that I'm just trying to sort out the facts here. <G>
These are a few excerpts from my January posts comparing those different
versions on the same hardware and using the same Arachne.cfg:
Repost 33Mhz '486 data:
All browsers had to retrieve image source files from the hard disk.
Times are seconds with accuracy better than 5%.
Browser Opera 3.5 Arachne 1.50b2 Arachne 1.60b1 Netscape2.02
Load GIF
73,559 8 4.5 4.0 4.5
170,146 10 4.8 4.3 5.0
231,849 12 5.0 4.4 5.0
306,395 15 6.0 5.5 6.5
Load JPG
185,935 33 26 24.6 19
33,235 16 15 14.5 9
Group two: Timing measurements under 5 Sec were made electronically.
Browser Opera 3.5 Arachne 1.50b2 Arachne 1.60b1 Netscape2.02
Reload GIF
from CACHE
Cartwheel: 6.6 6.1 5.5 6.5
REFRESH DISPLAY
Cartwheel: 0.9 6.3 5.5 0.36
GOTO PAGE END
Cartwheel: 0.275 6.5 5.5 0.325
RELOAD JPG
from CACHE
Weather: 1.04 3.8 3.0 0.60
REFRESH DISPLAY
Weather: 0.92 3.8 3.5 0.36
GOTO PAGE END
Weather: 0.275 3.5 3.0 0.325
And the following is on a 20Mhz '386:
The modem is 33600 and was used at that rate at 0700 GMT. Times are seconds.
Arachne version: 1.50b2 1.60b1
Connect and LOAD
home.arachne.cz 51 Sec 48 Sec
RELOAD (CACHED) 22 Sec 17 Sec
REFRESH (F9) 9 Sec 7 Sec
Page END 6 Sec 6 Sec
Connect and LOAD
quote.yahoo.com 26 Sec 21 Sec
RELOAD (CACHED) 17 Sec 12 Sec
As you can see, the difference isn't great enough to allow a 50Mhz
computer to beat a 75Mhz one even without the Pentium being involved.
Something else is going on.
- Clarence Verge
--
- Help stamp out FATWARE. As a start visit: http://home.arachne.cz/
--
