//flex table opened by JP

Click to See Complete Forum and Search --> : PCI MODEM TEST


Bleep
05-03-2000, 07:02 PM
PCI
MODEM TESTS
Having read over the years many discussions about PCI modems and some people haveing good luck with them and some not I decided to see for myself how PCI modems perform with different processors.
This is the method that I used---Same modem in 3 different machines--testing the modem connection speeds with a online connection speed test--The machines used are: All homebuilt - machine no 1 Ampton motherboard with AMD K6-2-350 64 megs pc100 ram . machine no 2 Tekram slot 1 with Celron 366 - 64 megs pc100 ram. Machine no 3 Soyo super 7 with AMD k6 2 500 - 64 megs pc 100 ram.
The ram in these machines are all Qbic memory. All machines have WD 10 gig hard drives with FAT 32
All of these machines fresh installs of Win 98 ver 1. one program installed, this was D-Net rc5 and this was shut down during these tests.
I installed the Modem in machine no 1 connected to the internet, ran the online speed check seven times and got connection speeds averaging 3400. I then installed the same Modem into machine no 2 and got average connection speed averaging 4400. Machine no 3 same result as machine no 1. These tests were repeated 4 times over a period of 4 days. These were done at around 3:00 AM CDT to try and minimize the number of hops that may effect the results. The resulting speed numbers were all rounded off.
Machine no 1 and no 3 (non Intel Chips) took a average of 94 seconds to verify user name and password. Machine no 2 averaged 14 seconds for user name and password. This I do not understand at all as I always thought this was a result of the server search function at the ISP.
It must be noted here that the machines were all close together and the phone line connection was switched beetween machines to eliminate any change due to line faults.
The results seem to me to be that even though PCI or non hardware modems use the CPU the speed of the CPU has no effect on the connection speed. It seems that the Celron always connected quicker and at a higher line speed than the non intel chips, I have no clue why this happens.
I also took note of the connection speed that is somehow congered up by the modem program and found this to be a very unrelable way to tell connection speed. At least in this instance.
I would appriciate any and all comments, if my methodology sucks and you know a different way I can test please tell me. If anyone knows why the Celron was more effective, I would like to know.
Bleep

Sharpy
05-03-2000, 07:14 PM
Can you post a link to the online test you used as i'd like to test my modem and post the results to compare?

Bleep
05-03-2000, 07:23 PM
Sure be glad to.
http://www.computingcentral.com/topics/bandwidth/speedtest.asp
Bleep

rrick31
05-03-2000, 07:35 PM
thanks bleep that was a very imformative post.sounds like you cover it all.i like to go to one page and time it via stop watch.i dont take much stock in connection rate but by the time.and as far a the intel connecting better i have heard that inel stuff runs better with windows because it was wriiten more intel than,amd,cyrix
thaks dude

RCN_Moose
05-03-2000, 07:36 PM
Hey Bleep-
That 3400 and 4400, is that K bytes/s or K bits/s?
I just tested my cable modem and got:
3709.1 Kbps or
454.5 K bytes/s

Just curious.
Moose http://sysopt.earthweb.com/forum/smile.gif

Bleep
05-03-2000, 09:22 PM
The 3400 is 34,000 bps. I used the normal statements relating to modem speed such as 56 K which I think is 56,000 bps. It is all relative as long as you use the same terminology on all tests.
Thanks for the comments.
Bleep

Sharpy
05-04-2000, 04:15 PM
Your line speed:
21.4 Kbps
2.6 K bytes/sec

56k PCI

EDIT - AMD K6-2 550


[This message has been edited by Sharpy (edited 05-04-2000).]