PDA

View Full Version : pcmcia adapter shootout


georgew
09-04-2003, 11:25 AM
Ok, I finished my first round of pcmcia adapter card testing.

I'm testing the following cards:
Elan p111
Elan p011
Lucent 014884 (tall card)
Lucent 015219 (shorter "stepped" card)

The client machine is a dell inspiron laptop with an orinoco gold card.

The radio card used in all of the tests was a zcomax 200mw card.

In the first round of testing, all cards worked fine. This first test was just seeing if they all worked, and could drive a high-power card. I tested them in a Dell Optiplex gx100 500mhz machine.

I wanted to do speed tests with the slowest machine I could find. So I used a 300mhz machine based on an Asus motherboard. To make the test interesting, I packed the machine with a variety of cards, such that all 5 pci slots had cards in them. The purpose was to expose any bus-related problems.

The first round of testing was done with identical settings, using an FTP of a 4 meg file, and running the transfer multiple times to get not only the best readings, but to see how consistant the results were.

The card that seemed to do the best was the lucent stepped card. It had very consistant transfer speeds ranging from 582 to 587KBs, with a ethernet spedometer reading of 690KBs peak.

The lucent tall card would not allow the packed machine to boot. When I reduced the number of cards, the machine would boot, but not see the card. It worked in the Optiplex... I think, now I have to go back and retest that to be sure.

The Elan 011 card was the slow performer, giving 526KBs peak FTP speed, with an ethernet spedometer reading of 612KBs peak.

The Elan p111 had the fastest ftp transfer, and the slowest. Speeds were all over the place, mostly slightly slower than the Lucent stepped card, but ocassionally faster or much slower. The peak FTP speed was 592KBs, but the slowest FTP was 494KBs. The ethernet spedomether indicated a peak of 765KBs, but I'm guessing the high spedomether speed indicated a high-speed burst of error correction or some other negative event. I'm guessing the p111 was either having bus errors, or some other intermittant problem.

I will rerun this test with another set-up to see how easy the results are to duplicate.

These tests are all very dependant on the cards interacting with the motherboard, and any problems I detect could be related to the motherboard rather than the pcmcia adapter.

I will run another round of similar tests with a different test-rig to validate these test results.

I may well discover that different motherboards give different results.

lonnie
09-04-2003, 04:19 PM
I may well discover that different motherboards give different results.

:lol: :lol: Really?

lbiffle
09-08-2003, 04:09 PM
In the first round of testing, all cards worked fine.

Have you run any tests with more than one card in a system? I need 3 radios, and they could be PCMCIA or miniPCI.

-Les

bobbyc
09-08-2003, 08:51 PM
George let me know if you want me to confirm any findings; we have dell optiplex gx1, gx100s, and gx 110s. We have ricoh dual/single slots, and orinoco/avaya slots (TI chipset)
So far I've noticed that on my gx110 with the TI pci adapters, I have to pull the pcmcia card out and back in before any clients can associate to it.
Bob C

Bossman
09-09-2003, 10:10 PM
We are running a number of systems with 3 or 4 wireless adaptors.

All systems are PIV class systems with ASUS boards and Elan PCMCIA adaptors.

In the following setup, we pulled over 1 Mbit/s via FPT and a 100 MB test file.... all cards were Lucent (except client) and all routers Star-OS in an ASUS box or Soekris.

Asus PIV 4 Cards <-- 15.5 miles--> Soekris 2 cards <-- 16.5 miles --> ASUS PIV 4 slots (2xP423) 3 cards <--local--> P200 laptop with TT 2mb card.

- Arthur