Comcast XFINITY → [Speed] Why is download so low if...

If I have an upload of 3.6 Mbps, shouldn't I have 25 +/- Mbps for a download? I only get about 13 Mbps, and that is the case for a couple months now. I could swear when they first hooked me up it was closer to 25.

Another question:

My latency before tonight has always been around 60+ ms, but tonight is is about 21 ms. Why would latency suddenly improve like that?

I hope my latency stays low like this.

I don't know which plan, for sure, I am on, but it is a bundle package without contract for about 140 dollars. I know I am paying more than if I would have gone with a contract, but too many uncertainties to do that right now.

Recently there was a serious problem with certain channels not working, and I thought it was my box, but evidently it was some sort of system problem, either the neighborhood (where the signals aggregate) or at the office, who knows, but I was getting artifacts, some channels wouldn't work at all, some showed as being unauthorized so I couldn't watch them even though my agreement included those channels (such as MSNBC).

So 4 Mbps upload, which download speed is paired with that? Could they be giving me 4 instead of 2 by accident?

Well without going into a lot of checks, if you have a router, please disable WMM under the QoS tab and try again.

What are you measuring latency to, or is that the value shown by the speed test? And what speed tests are you using?

As mentioned, some routers have QoS settings that can throttle speeds. An older router or modem could also be an issue, along with how you are connected. Are you wired to a router or direct to the modem, or using wireless?

Ok, disabled WMM under QoS. I am using speedtest.net. Speed jumped to 16+ down...

Comcast's speed tiers vary widely (and change frequently) based on your location. Their (deliberately) annoying practice of including PowerBoost in their advertising for their speed tiers just makes it more difficult for a customer to know exactly what they are paying for what.

Try using ShaperProbe to test your speed, and connect a PC directly to your modem (reboot both the modem and the PC after making that connection) to eliminate your router as a possible source of speed limiting.

Here are the ShaperProbe results that I just got:

DiffProbe release. January 2012. Build 1008.
Shaper Detection Module.
 
Connected to server 4.71.210.239.
 
Estimating capacity:
Upstream: 17941 Kbps.
Downstream: 95086 Kbps.
 
The measurement will take upto 2.5 minutes. Please wait.
 
Checking for traffic shapers:
 
Upstream: Burst size: 3895-4484 KB;
Shaping rate: 2128 Kbps.
 
Downstream: Burst size: 18257-19729 KB;
Shaping rate: 12786 Kbps.
 
For more information, visit: http://www.cc.gatech.edu/~partha/diffprobe
 
 

The "Shaping rate" values are your real sustainable speeds. I am on a 12/2 speed tier, and the ShaperProbe test shows that (it also shows the potential 95 mbps PowerBoost induced results that on-line speedtests usually don't take into account).

Here is my current result from your favorite speedtest.net site:

Shaper probe is crashing on my system. I don't know why.

I seem to recall seeing reports that earlier releases of ShaperProbe had problems on Windows 7, but I thought that had been fixed with the newer versions of that program. Are you running the January 2012. Build 1008 release, or an older release?

The one in that link, not the previous version, but the one above.

I had some problems with the older version sometimes hanging and not completing the test even on Windows XP (one of the reasons that I did not used to recommend using that test), but the Jan 2012 version has worked with no problems for me. Hopefully if someone else reading this thread is still seeing problems with the newer version they might be able to suggest a workaround for you (if there is one).

Vudu Speed Test seems to correlate pretty well for me with ShaperProbe (and what I can see in my realtime bandwidth on my Tomato router) for download speed. The download test is a little longer then others.

»speedtest.vudu.com/cdn1/

The short upload test gives an exaggerated (power boosted?) number.

Just for Ss and Gs, rather than try to initiate/run the program from Firefox's downloads window, I started it from the directory, and it runs.

DiffProbe release. January 2012. Build 1008.
Shaper Detection Module.

Connected to server 213.244.128.168.

Estimating capacity:
Upstream: 3678 Kbps.
Downstream: 14333 Kbps.

The measurement will take upto 2.5 minutes. Please wait.

Checking for traffic shapers:

Upstream: Burst size: 5094-5152 KB;
Shaping rate: 2053 Kbps.

Downstream: No shaper detected.
Median received rate: 14037 Kbps.

Your maximum and sustainable downstream speed of ~14mbps with no detected PowerBoost is kind of different from any speed tier I have seen offered by Comcast (but as I originally said, Comcast's speed tiers do vary widely by location).

Are you in an area that has not been upgraded to DOCSIS 3, or do you have a non DOCSIS 3 modem? If so, you may actually have 16/2 service, but your node is so busy that you are not able to reliably reach 16mbps downstream (or have any PowerBoost available).

And FYI, the previous speedtest.net result I posted was apparently due to the DoS attack on my email server. Once that attack dissipated, that test went back to its usual PowerBoost inflated value:

It's an Arris and I think it is DOCSIS 3. It is pretty heavily populated where I am.

You can tell if you have multiple bonded downstream channels by looking at your modem's status pages (usually at »192.168.100.1 but some modems, and most gateways can differ).

If you only have a single downstream channel, that (and a heavily loaded node) could explain your oddball downstream speed and the lack of PowerBoost.

There's 4 channels showing for downstream.

I'm getting as low as 10 Mbps right now.

Sounds as if you are on a really overloaded node, or you have some problems with your connection or modem (I am assuming that you have eliminated your router as the bottleneck by testing with a PC connected directly to the modem).

No, I haven't hooked directly to the modem lately. But IIRC, I wasn't seeing anything better than 15 when I WAS. I am using an ethernet over powerline kit, and it works very well. I used to live where there was 60 down and 8 (?) up with a different cableco, different residence.

I don't have AC in the room where the modem and router are in, and it is hot, so it will have to wait. But I don't think the problem is internal to my residence. For one thing, as I said, latency is drastically improved.

Are you 110% sure that it works very well? If I had known that earlier in this thread, it would have been at the top of my "usual suspects" list.

As I mentioned, I could be wrong, but I recall measuring about 15 down with the direct connection. Also, I have wireless and that doesn't change things.

I WILL check this just to be certain...

Wireless is usually number one on my "usual suspects" list (and it is why my initial suggestion to you was to test with a PC directly attached to the modem), but that is only because I don't run into Ethernet over powerline as often (and you did not initially mention it).

Ok, I DO get 24 over a direction connection to router in living room where other computer is. So I tried the bedroom (system we are talking about, where the AC is) over wireless instead of over PoE, and got the same speed, about 14. At this point I am beginning to wonder if it has something to do with my computer here.

Wireless SHOULD be different to some extent, but I get identical.

Sounds like you are on the Performance tier. Here in CT I get 24 - 25 down/3.5 - 3.7 up (including PB). I would suspect that you should be getting right around the same performance (on Sharper I get 20 down).

What do you mean "over PoE?". That would be for dsl not Comcast, right? At any rate, if you are getting about 14, whether direct ethernet or wireless, then perhaps it has to do with a setting within that computer.

BTW, I lose nothing speed wise on my wireless connection through my router, anywhere within reasonable range.

What is your wireless encryption. Make sure it is WPA2-AES.

There's WPA2 "enterprise", "enterprise mixed", "personal", and "personal mixed". Which should I choose?

Pete, I meant EoP, easy error to make, I have an Ethernet over Powerline kit, Zyxel. At the old place it worked fantastic, and now I remember, it was working well here too, until a month or two ago, IT SEEMS, I do get 25/4, but with the EoP devices it drops to 6-14/3.6. The wiring here sucks, so maybe they are getting worn out (I hope not!) I could run a long ethernet cable (I just ran a 20 footer between rooms to verify it isn't the computer), but it would be very messy and I don't have such a cable right now.

I think that the OP meant EoP (Ethernet over Powerline), at least that is what was being discussed earlier in this thread. And FYI, PoE means Power over Ethernet (often used to supply power to Wireless Access Points in remote locations that are not near a power outlet, and it is also frequently used to power IP phones), and it is in no way related to PPPoE (Point to Point Protocol over Ethernet) which is probably what you were referring to as being related to DSL.

EDIT: Oops, I guess the OP answered your PoE question while my tired arthritic old fingers were typing.

you should be getting what you pay for on each billing cycle.If you pay say 50 bucks a month for 15 down and 3 up then that is the number you should be concerned about don't worry about if you got 80 down and 50 up yesterday because you have no complaint with your provider with that nonsense,only the numbers on your bill. Their is fine print that goes along with this as well which means those boost signals are not guaranteed and the list goes on and on.I have been getting 36 down and 7 up for the longest with sustained transfer at 32 Mbps or 3.2MBs and that is with a 1 gig file but I pay for 15Mbps and that is what i expect and that other number is complementary as far as i am concerned because my service here is literally blazing.My friend has dsl and he can download a game demo on Xbox live and literally has to start it go to work and then play when he gets home if its over a gig but I am playing the same demo in 10 minutes.

Incidently, indeed, the devices have a red LED, indicating slow speed. I recall when I first moved in, they were green. I know they changed something here because in June when it was in the 90s, a heater was burning me up and they had to have a pro come out and change something. Maybe they changed something electrical (radiator heat here, but who knows...)

There is no thermostat in here, so I depend on how it is set in another apartment.

The wiring here is bad, as I say. They have conduit running along the wall near the floor feeding outlets (definitely not code), but I searched high and low for a place that I could afford, so this is it...

WPA-AES is Personal. Once set check your speed with WMM enabled and disabled.

For some strange reason, when I've tested current N gear with WMM turned off, I literally could not obtain a connection wirelessly.

If you disable WMM, you will only get a G connection at best. The N specs require QOS, which is what WMM is.