Welcome to the MacNN Forums.

canadave · Feb 20, 2009, 06:14 PM

Hi,

I have a Jan. 2008 Mac Pro, stock. I'm using a WRT54GL router running Tomato v1.23, and my ISP features 15Mbps download speeds.

My Mac Pro sits about 20-25 feet away from my router. I've recently had occasion to test download speeds of large files using wireless, and immediately tried downloading the same files while connected with a 25-foot Ethernet CAT6 cable (in fact I tried two different cables). Consistently, I get about 350 KBps download speeds over wireless...but about 210 KBps download speeds using the Ethernet.

Now, I know Ethernet performance degrades with distance. But I have to admit I was surprised at what seems to be an extreme degradation of download speeds over Ethernet. Has anyone experienced anything like this? Is there some setting I've got to adjust on the Mac Pro to get Ethernet speeds at, or better than, the wireless? Or is it just a fact that 25-foot Ethernet is always going to be slower than wireless?

Thanks,
Dave

Cold Warrior · Feb 20, 2009, 07:19 PM

If your ISP offers 15 Mb then 350 KB is pretty sad. It's more likely that your download source is capping out at those speeds. Have you tried measuring speeds while copying a large file from one local network computer to another -- specifically one wired computer to the Mc Pro over wifi, then the same computer to the Mac Pro over wired?

tooki · Feb 20, 2009, 09:12 PM

And also, if Ethernet is so much worse than wireless (assuming your tests are statistically significant), you probably have a bad cable or something. But regardless, I agree with Cold Warrior, you need to test locally, without using the Internet.

ghporter · Feb 20, 2009, 10:46 PM

With an advertised speed of 15Mbps downstream, and you just getting just a tiny trickle through, there is a big problem, and no ISP can survive with that sort of performance. However, you need to know what is going on within your network so that if there is something wrong with your ISP connection, you can tell 'em how you're sure.

However, it's possible (maybe probable) that there is something wrong with your network setup. Ethernet performance over 25 feet should be imperceptibly different from over 0 feet, especially with Cat6 (even mediocre Cat6).

Can you post how you're connected, from the wall to the router? I'm thinking there's something odd about your modem's connection to the router, or about the connection TO the modem from the outside world. You're using a cable ISP, right? Knowing which one would be helpful.

Oh, and since this really isn't about your Mac Pro, but rather about your network, I'm going to move this to the Networking forum.

canadave · Feb 20, 2009, 11:57 PM

Thanks everybody (sorry about posting in the wrong place, I'd figured it was probably something on the Mac Pro causing the issue).

I'll definitely try local testing...should've thought of that to begin with. I'm sure it's not the cable, since I tried two different Ethernet cables and the same result occurred with both.

As an aside, Cold Warrior: Is 350 KBps really a slow download speed? For what it's worth I was downloading the Second Life install .dmg file from their site. I was under the impression that my download speed won't be close to the ISP's throughput (15Mbps in my case), but will rather be limited by the site I'm downloading from, in which case I thought 350 was about right. Am I mistaken?

Glenn--thanks for confirming what I thought about Ethernet--25 feet should be imperceptible. My setup is as follows:

EastLink.ca ISP (Canada Maritimes) --> Motorola SB 5101 cable modem --> WAN port of WRT54GL router running Tomato 1.23 --> Ethernet LAN cable to Mac Pro, and wireless to Macbook Pro. It's a straightforward setup, nothing out of the ordinary.

Cold Warrior · Feb 20, 2009, 11:59 PM

Yes, that's what I was getting at -- your download is limited by what the site can give you, not what your ISP can. 350 KB from a site for web download is good. But I was thinking that this site's speed could easily be erratic -- 350 KB at one time but 210 KB fifteen minutes later.

canadave · Feb 21, 2009, 12:17 AM

Ah, okay, I thought you were trying to say something else.

Yep, I understand what you're saying, but I tried it a bunch of times immediately one after the other--I'd download via Ethernet, let it run for about 15-20 seconds, note the speed, then cancel the download, immediately switch to the wifi and disconnect the Ethernet, download again, note the speed, rinse, repeat. It was extremely consistent, even over the 5 or 6 times I tried it at various times of day, over different days, etc. I feel pretty confident in saying that the Ethernet is for some reason slower than the wifi. I'm just not sure why....

canadave · Feb 21, 2009, 01:25 AM

OK...weirder and weirder!

I just tried a file copy from the Mac Pro to my laptop upstairs, via wifi first and then Ethernet. The Ethernet turned out to be about 30% faster! So I tried the "Internet download test" again, but I got the same results as before--the wifi download is TWICE as fast as the exact same download via Ethernet!

In fact, Internet download speeds from Second Life on my laptop via wifi, TWO FLOORS above the Mac Pro and the router, were consistently at 450 KBps, while Ethernet download speeds at the Mac Pro a few seconds later were only at 270 KBps, as they've always been via Ethernet!

So Ethernet rocks wifi on the LAN, but it's the reverse situation regarding WAN speed. Huh?

tooki · Feb 21, 2009, 07:55 AM

Originally Posted by ghporter

Ethernet performance over 25 feet should be imperceptibly different from over 0 feet, especially with Cat6 (even mediocre Cat6).

Just posting to dispel the myth, again, that Cat 6 is needed for... anything. All Ethernet standards in use currently, including gigabit, are specified for Cat 5 (not even Cat 5e!!). (Well, 10Base-T is actually specified for Cat 3.) Consequently, more expensive cables may only help in some corner cases where you are struggling against interference.

That said, a 25 foot Ethernet cable is nothing. No length of cable that is within the limits of the standard (100m, or about 330ft, for 100BaseT) should cause any detectable difference in speed.

tooki · Feb 21, 2009, 07:58 AM

Originally Posted by canadave

So Ethernet rocks wifi on the LAN, but it's the reverse situation regarding WAN speed. Huh?

Is it possible there's a Quality of Service setting on your router firmware that's causing this?

Also, are there sources of interference near the Mac Pro that could be affecting it? Things like microwave ovens, some 2.4GHz cordless phones, lots of Bluetooth devices, etc.

ghporter · Feb 21, 2009, 11:27 AM

Originally Posted by tooki

Just posting to dispel the myth, again, that Cat 6 is needed for... anything. All Ethernet standards in use currently, including gigabit, are specified for Cat 5 (not even Cat 5e!!). (Well, 10Base-T is actually specified for Cat 3.) Consequently, more expensive cables may only help in some corner cases where you are struggling against interference.

That said, a 25 foot Ethernet cable is nothing. No length of cable that is within the limits of the standard (100m, or about 330ft, for 100BaseT) should cause any detectable difference in speed.

With ANY good quality cable, 25 feet is, as you say, nothing at all, but the way Cat6 is made should make it a little more accepting of the abuse we give all Ethernet cables. I've managed to get excellent performance from "recycled" bulk Cat5 in some rather interesting environments, with runs in the hundreds of feet. New cable and especially quality new cable should be fan-freaking-tastic at 25 feet.

tooki · Feb 22, 2009, 01:45 PM

Higher-category cable can reject more interference than a lower category. That's it. And because the higher the category, the longer the actual wire pairs inside are, then a higher category could theoretically reduce the maximum cable length. (Not that that's normally an issue.)

So for real-world applications, I sincerely doubt that going beyond Cat. 5 will provide any detectible improvement. And again, my reason for posting is to make it crystal clear that higher-than-category-5 cable is not required by ANY ethernet standard, please don't give people that impression!

ghporter · Feb 22, 2009, 04:04 PM

I had thought that the higher twist rate and overall larger diameter of the cable would make Cat6 more physically robust. Harder to bend too sharply, harder to pinch, etc., that sort of thing. I hadn't really thought about the way the higher twist rate of the pairs would interact with physical vs. electrical length... I do see the point of DE-emphasizing the mistaken assumption that "higher standard cable is better and/or more desirable," and heartily concur!

tooki · Feb 22, 2009, 07:44 PM

I don't think the category has anything to do with physical robustness -- if there are differences, they're trivial. There are differences in types of cable (e.g. plenum rated cable sucks for making finished cables), but that's independent of category.

dimmer · Feb 22, 2009, 09:07 PM

Actually, the specifications for 10GbE require Cat 6 or Cat 6a cabling, so if you are installing cable as an infrastructure piece (inside walls, over ceilings, etc.), you should definitely be using 6a. If it's just a patch cable, no real need.

canadave · Feb 24, 2009, 08:23 PM

Well, I have to say I'm at the end of my rope here.

I've definitely confirmed: the Ethernet on the Mac Pro downloads from the WAN *twice* as slowly as my Macbook Pro's wifi does. I've tested it a million times, there's no mistake.

I'm still not any closer to understanding why, and I'm just about ready to reformat the darned thing.

My last question--could it be some sort of weird half/full duplexing thing on the Mac Pro's Ethernet port (although i went into Network preferences and the Etherhet speed is set to "Automatic")? Or perhaps something to do with Boot Camp and Windows XP that I have installed as well? I'm just tossing ideas out there...really have no clue.

Doofy · Feb 26, 2009, 04:10 PM

Originally Posted by canadave

Yep, I understand what you're saying, but I tried it a bunch of times immediately one after the other--I'd download via Ethernet, let it run for about 15-20 seconds, note the speed, then cancel the download, immediately switch to the wifi and disconnect the Ethernet, download again, note the speed, rinse, repeat.

Note that there's another possibility here.

If the following conditions occur:

1) You always download the same file.
2) You always try it in the same order - i.e. "ethernet first".
3) Your browser/downloader has some kind of "resume" mechanism.

Then the extra speed you see on the second run (the wifi) could be influenced by the browser/downloader picking up and resuming the download - which will show higher transfer rates (because at that point, a lot of it's coming off your HD).

RMXO · Feb 26, 2009, 05:19 PM

do you have the latest firmware for your router?

Doc HM · Feb 28, 2009, 10:32 AM

Do you have another Mac (or a laptop owning friend). Try connecting your mac to the other directly, ethernet port to ethernet port (no router) and copy a nice big file from one Mac to another.

If your copy speed is still slow you have a physical problem with your ethernet port on the Mac. It's not likely but it could happen.

ghporter · 11:43 AM

Originally Posted by tooki

I don't think the category has anything to do with physical robustness -- if there are differences, they're trivial. There are differences in types of cable (e.g. plenum rated cable sucks for making finished cables), but that's independent of category.

All of that is moot though, just ramblings I added without realizing I'd be going off topic until too late.