Those ADSL/cable bandwidth speed test sites... are they accurate?

[Deja Q]

I've been looking at internet bandwidth speed tests to note my throughput, as opposed to the connection Qwest patched me in at. I've been to many such sites, on both PCs on my LAN. Often I get ~1200kbpd down, but recently noted that when I join Unreal Tournament 2004 online games, people start complaining of lag and disconnecting - one person referring to me directly because there was no lag before. Even Doom3, which in general is slow for multiplayer games, lags horrendously for ME to the point of unplayability so I have a problem.

So I've been to those sites again; running MANY tests. My download speed varies from 400 to 1200 now. Usually 1000+ but it does take a dramatic downward dip that cannot be normal because these gaming lag problems haven't happened before.

The modem's WAN status screen says an unuusally high number for "Far End RS FEC (I/F)". In a 3 hour period, I've gotten 4200/0 on that indicator. I've read online users' who have the Near End indicator ramp up to extreme numbers, but haven't found a thing on "Far End" at all.

I replaced the NIC in both systems but that did not help.

(one system is Linux, the other Windows with enough anti-spyware/virus programs and the problem happens on the linux box even when the Win box is shut down, and I don't go to naughty sites and I don't open strange e-mail attachments.)

I called up Qwest, who said that only the WAN Status screen gives a true connection rate (and proves they haven't yet upgraded me yet to their new 5mb connection that was to be done Thu or Fri...). They basically couldn't give me any help. They couldn't even answer what "Far End RS FEC (I/F)" means. All I (think I) know is that FEC is a type of error checking. But is ~4000/0 in a ~3 hour period too many? The other error values are much lower (though, in the same time period, th Near End RS FEC value was 721/0.)

Could my modem have gone wonky? It's an Actiontec 1524. I live 1.1 miles away from the central hub and my connection was rock solid before, so I am convinced this is a new problem that is not a virus on any of my PCs.

What else can I do to determine the problem as Qwest is giving me the cold shoulder? (and as I prefer DSL to cable, plus their horror stories, I likely won't be switching to comcast.)

Thx!

[TrueAmerican]

1. Checkout

http://www.broadbandreports.com

Use the tools section and forums. They will help you. You many have some latency issues.

Thanks

[pnorman]

3. Thanks for that site!

Here are my results:
2005-06-19 07:42:45 EST: 1273 / 317
Your download speed : 1304181 bps, or 1273 kbps.
A 159.2 KB/sec transfer rate.
Your upload speed : 324628 bps, or 317 kbps.

I picked Speakeasy.org for no better reason that it happens to be my own ISP. And I'm assuming that the fact that they're just across the street from my ap't, wouldn't have any bearing on the above speed ( ;) ;) ;) )

pnorman

[Deja Q]

5. Thanks!

They had a couple of more comprehensive tools than cnet or the other two places I'd tried... there are no latency issues on my end but I'll keep checking my sped through them exlusively and log the details.

Thx again!

[CabalPowered]

2. Maybe do a traceroute?

If you know the IPs of the servers that is...

Maybe LAN performance?

[Omphaloskepsis]

4. Have you tried capping your upload?

Maybe the ACK packets are getting stuck. I use a Macintosh with cable. Grain of salt and all.

[Peachhead22]

6. Sorry to get off your topic a bit, but...

...what's the typical download rate of cable HSI? I've got SBC DSL and my buddy has Adelphia Cable broadband. I'm getting around 275 Kbps he's getting around 500 Kbps (both download speeds, both measured through Firefox's bandwidth tester for comparison purposes). His connection is faster than mine, but I was expecting his cable to be several times faster, not less than twice as fast.

[Orsino]

7. No, they are not and cannot be accurate.

Your own ISP's speed test, perhaps, would be a good measure of your throughput, but third-party servers can only try to open a session with you, and measure the session bandwidth. Heavy loads on such third-party servers may interfere.

To use a variation on the analogy I give my customers: when a cop stops you for speeding, he cannot tell how wide your driveway is.

In other words, the bandwidth you purchase is only between you and your ISP, and--we presume--through that ISP. If you're into online gaming, you are probably traversing several peering points and other ISPs to reach a destination server.

I still get the occasional complaint from a customer who has recently upgraded from DSL to my T-1 service, or from a single T-1 to multiple, or even to T-3. He'll wonder why he isn't getting faster throughput to his favorite web sites, and I'll have to break the news to him: he's only upgraded his aggregate bandwidth, which has little bearing on session speeds, the latter of which typically max out at a few hundred kbps (or maybe a hundred kBps). Widening your driveway, I tell him, doesn't get you to work any faster.

[LynzM]

9. That's a great analogy

Thanks for sharing it :)

[masshole]

8. Good site for speed tests and diagnostics

www.pcpitstop.com