need advice on improving wireless throughput

I have a T42. Is it normal for wireless to have less than half the throughput of a hard-wire connection? My router is Linksys WRT54GL. Here are some examples with links:

FILE SOURCE      802.11g    Ethernet (KB/s)
speedtest          240         560 (42%)
filefront          255         575 (44%)
space-multimedia*   76         270 (28%) 

*site apparently enforces 275KBs hard limit on download speed

My internet connection is 7Mb DSL, although in practice, I rarely get higher than 5. So the result from the speed-constrained site (28%) is the best indicator of the actual wireless throughput ratio. That's a huge hit. It's even worse (22%) when comparing PC-to-PC file transfer (2200 KB/s via wireless vs. 9900 KB/s via ethernet).

So far, I've upgraded the firmware for the wireless card and router, changed mode from 'mixed b/g' to 'g only', tried different channels, even disabled security encryption (temporarily). None of these individually or collectively made a hair's difference, perhaps 5% at best. In all cases, signal strength was excellent (5 bars). I also tried a different router to rule out any issue with my Linksys. And finally, I replaced the wireless card with an Intel Pro 2200bg, which is supposed to be faster (and cooler) than original Philips 91P7301. Performance is much the same as before.

Do other folks get similar results? Perhaps nothing is broken, but rather that wireless imposes an unavoidable and huge (>70%) hit on throughput.

Thanks for your reply. I understand what you're saying, but I need to get more specific. I mean, 802.11g has 54Mbs capacity (6750 KBs). And since I've demonstrated sustained speeds of up to 2200 KBs over my local lan, wireless is obviously not a bottleneck per se. Also, by running tests in batches and repeating on different days, I've eliminated the effects of any intermittent interference.

This is not just an academic exercise. I'm considering purchasing a blu-ray DVD player with netflix support, and I'm limited to wireless. One possibility is to upgrade my router to 802.11n, but it may not make a difference if the bottleneck is elsewhere.

The reason I'm using Ethernet as a benchmark is because it has almost no overhead and thus reveals upstream constraints. For example, my DSL is constrained to about 5 Mbs, and the space-multimedia file server is constrained to about 270 KBs. I'd like to understand why wireless can only sustain a transfer of 70 or 80 KBs, yet it can sustain more than 3x that amount when downloading from a non-constrained server (although still well short of my DSL pipe).

Since my OP, a friend tested his wireless/ethernet ratio using space-multimedia server linked above and reports a sustained transfer rate of 240 KBs via ethernet and 202 via wireless (802.11g), which is better than 80%. This seems to disprove the notion that low wireless throughput is expected and unavoidable. The main difference in his setup is that his wireless card is PCI-e, whereas mine is mini-PCI. Could the interface type explain the huge difference in throughput?

Another possibility is a local sustained interference source. Is there a diagnostic tool that shows percentage of packets that have to be retransmitted?

Is the link you tested 802.11g or n?

OK, so I'm envious! Do you know if your wireless card is PCI-e or mini-PCI? I found MSI-1013 specs but it only says "optional 802.11g wireless".

I'm curious to see your results with MSi via wireless (since it has mini-PCI).
Edited by ginahoy - 4/6/11 at 9:57pm

Still, 80% of 6Mb is a heck of a lot faster than what I'm experiencing. I probably just need a new router. THANKS for checking that for me!

I'm considering an N class router. However, I'm not sure if T42 built in antenna will support N. Do you know anything about that? I saw some mini-PCI N-class cards on ebay, with the same Main/Aux antenna connections as my current card, but most PCI-e cards (which I can't use) have at least 3 antenna connections.