Q regarding channel ID, power and speeds

[CPCUK]

Hi,

This may be a really dumb question but I'm gonna go ahead and ask anyway.

When I was on 50Mbit and QAM64 I got 50Mbit on speedtest sites. Quite a while back now I noticed that I was moved over to QAM256. My power levels went quite high (from around 0dB to 13dB) but everything seemed fine. I added an attenuator to bring it back to around 0dB.

Anyway, I've noticed that my downstream channels are not consecutive, ie, 158,157,159,160 and although it might just be me clutching at straws I noticed that on speedtest sites I only obtained around 30Mbit to 45Mbit speeds. Logically I assume they are simply channel names and nothing more rather than required to be consecutive (as I said its a daft thought, lol).

Having upgraded to 100Mbit yesterday the engineer said that I should receive good download speeds.....we both laughed when I received 30Mbit!!!

To be honest I wasnt happy but I thought I'd se how things went. The engineer stated that my actual area shouldnt be too heavily populated with subscribers too!

Anyway, I ran 14 simultaneous downloads this morning which equated to around 12MB/s of downloads (great!) but why cant I see this speed on speedtest sites?

I stopped getting full speed I'm sure when I was swopped over to QAM256 six months or so ago.

I've also right now got 18dB of attenuation on the coax at the modem to give me a 0dB level? Should I require this amount really as it used to be around 6dB i the past.

Superhub power stuff shown below (looks ok to me)

Any help appreciated

--------------------------------------------------------------------------------

Startup Procedure
Procedure Status Comment
Acquire Downstream Channel 443000000 Hz Locked
Connectivity State OK Operational
Boot State OK Operational
Configuration File OK
Security Enabled BPI+


--------------------------------------------------------------------------------

Downstream Channels
Lock Status Modulation Channel ID Max Raw Bit Rate Frequency Power SNR Docsis/EuroDocsis locked
Locked QAM256 158 55616000 Kbits/sec 443000000 Hz 0.0 dBmV 41.5 dB Hybrid
Locked QAM256 157 55616000 Kbits/sec 435000000 Hz -0.3 dBmV 41.4 dB Hybrid
Locked QAM256 159 55616000 Kbits/sec 451000000 Hz 0.0 dBmV 41.5 dB Hybrid
Locked QAM256 160 55616000 Kbits/sec 459000000 Hz 0.0 dBmV 41.5 dB Hybrid
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV 0.0 dB Unknown
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV 0.0 dB Unknown
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV 0.0 dB Unknown
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV 0.0 dB Unknown


--------------------------------------------------------------------------------

Upstream Channels
Lock Status Modulation Channel ID Max Raw Bit Rate Frequency Power
Locked ATDMA 3 20480 Kbits/sec 45800000 Hz 42.0 dBmV
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV



--------------------------------------------------------------------------------

Primary Downstream Service Flow
Downstream(0)
SFID 10857
Max Traffic Rate 110000000 bps
Max Traffic Burst 10000 bytes
Mix Traffic Rate 0 bps


--------------------------------------------------------------------------------

Primary Upstream Service Flow
Upstream(0)
SFID 18540
Max Traffic Rate 10250000 bps
Max Traffic Burst 16320 bytes
Mix Traffic Rate 0 bps
Max Concatenated Burst 16320 bytes
Scheduling Type Best Effort

[Sephiroth]

Hi,

This may be a really dumb question but I'm gonna go ahead and ask anyway.

When I was on 50Mbit and QAM64 I got 50Mbit on speedtest sites. Quite a while back now I noticed that I was moved over to QAM256. My power levels went quite high (from around 0dB to 13dB) but everything seemed fine. I added an attenuator to bring it back to around 0dB.

Anyway, I've noticed that my downstream channels are not consecutive, ie, 158,157,159,160 and although it might just be me clutching at straws I noticed that on speedtest sites I only obtained around 30Mbit to 45Mbit speeds. Logically I assume they are simply channel names and nothing more rather than required to be consecutive (as I said its a daft thought, lol).
[SEPH]: They don't have to be consecutive, but they do have to be at a frequency separation no greater than 8 MHz one between the other as if they were a consecutive group

Having upgraded to 100Mbit yesterday the engineer said that I should receive good download speeds.....we both laughed when I received 30Mbit!!!

To be honest I wasnt happy but I thought I'd se how things went. The engineer stated that my actual area shouldnt be too heavily populated with subscribers too!

Anyway, I ran 14 simultaneous downloads this morning which equated to around 12MB/s of downloads (great!) but why cant I see this speed on speedtest sites?
[SEPH]: It's the downloads that matter. I get crap speedtest reports but better download speeds from sites with good bandwidth that are not busy. I don't know what's gone awty with speedtest.net, but I don't trust it any more.


I stopped getting full speed I'm sure when I was swopped over to QAM256 six months or so ago.
[SEPH]: When you went to QAM256, you went onto a different line card and most prolly with competing users whose activity is affecting you worse than when you were on the old line card. Contention, in other words.


I've also right now got 18dB of attenuation on the coax at the modem to give me a 0dB level? Should I require this amount really as it used to be around 6dB i the past.
[SEPH]: It depends what the engineer did at your street cabinet. Your level would rise if the engineer put you on a lower attenuation tap point, and it would be that high if you were near to the street cabinet. Or, they've replaced a weak amplifier in your street cabinet and suddenly, woomf.


Superhub power stuff shown below (looks ok to me)
[SEPH]: Yep - fine


Any help appreciated

--------------------------------------------------------------------------------

Startup Procedure
Procedure Status Comment
Acquire Downstream Channel 443000000 Hz Locked
Connectivity State OK Operational
Boot State OK Operational
Configuration File OK
Security Enabled BPI+


--------------------------------------------------------------------------------

Downstream Channels
Lock Status Modulation Channel ID Max Raw Bit Rate Frequency Power SNR Docsis/EuroDocsis locked
Locked QAM256 158 55616000 Kbits/sec 443000000 Hz 0.0 dBmV 41.5 dB Hybrid
Locked QAM256 157 55616000 Kbits/sec 435000000 Hz -0.3 dBmV 41.4 dB Hybrid
Locked QAM256 159 55616000 Kbits/sec 451000000 Hz 0.0 dBmV 41.5 dB Hybrid
Locked QAM256 160 55616000 Kbits/sec 459000000 Hz 0.0 dBmV 41.5 dB Hybrid
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV 0.0 dB Unknown
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV 0.0 dB Unknown
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV 0.0 dB Unknown
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV 0.0 dB Unknown


--------------------------------------------------------------------------------

Upstream Channels
Lock Status Modulation Channel ID Max Raw Bit Rate Frequency Power
Locked ATDMA 3 20480 Kbits/sec 45800000 Hz 42.0 dBmV
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV
Unlocked Unknown 0 0 Ksym/sec 0 Hz 0.0 dBmV



--------------------------------------------------------------------------------

Primary Downstream Service Flow
Downstream(0)
SFID 10857
Max Traffic Rate 110000000 bps
Max Traffic Burst 10000 bytes
Mix Traffic Rate 0 bps


--------------------------------------------------------------------------------

Primary Upstream Service Flow
Upstream(0)
SFID 18540
Max Traffic Rate 10250000 bps
Max Traffic Burst 16320 bytes
Mix Traffic Rate 0 bps
Max Concatenated Burst 16320 bytes
Scheduling Type Best Effort

[Mick Fisher]

Hopefully you are disabling AV (specifically the web protection component) before using a speed tester.

[CPCUK]

I dont have any AV on my system.

The power levels I constantly check to be honest and adjust accordingly. Like a said I used to have around a 6dB attenuator in place but when QAM256 got installed it shot up to 13dB and I had to use more attenuation. At no time did VM ever come around yet whatever they did it certainly changed my power levels.

The last time I was looking inside the roadside cabinet was when I had 20Mbit and had all sorts of problems. Different power settings were adjusted in there back then and if they have changed since I was on 20Mbit I have no idea. That was a while ago mind you.

It just amazes me that none of the speedtest sites ever show my full speed anymore even though it does appear I'm getting good 'actual' speeds when downloading.

I'm a little reluctant to call techs in as I feel it'll just be a case of 'ignore the speedtest sites as they arent accurate in most cases'. Would just like to know why I cant see the full speed on them.

Thx

[Mick Fisher]

Off the topof my head..............

Do you have an up to date version of java installed?

Pop up blockers may have an effect.

An AV you used to have and haven't uninstalled fully.

If your download is working properly as verified by a site known to be capable of giving a throughput rate consistant with whatever tier you are on, usenet for instance, then the answer must lay on your machine or attached router perhaps.

[CPCUK]

Thanks for your info.

This machine (Win 7) has no AV or pop up blockers other than the Windows own stuff from WIN Update.

It is very odd that I used to be able to get speedtest sites to show my speeds properly and not now. I also agree its strange that I can get good throughput from dl's too which does lead me to believe it could be something ultimately on this machine.

I'm gonna hook up a Win XP machine shortly and do the same speedtests and see how that might go. Will post back if things look up :)

Cheers

Java normally gets updated whenever I see the message 'Java update available' on screen.

At this moment in time I am using the Superhub as a modem and from there directly to my pc. With or without my gaming router attached so far has made zero difference.

[Mick Fisher]

This machine (Win 7) has no AV or pop up blockers other than the Windows own stuff from WIN Update.

Might be worth temporarily disabling those during speedtesting just in case. Also, try different browsers and make sure to clear your cache, agin just in case. :)

It is very odd that I used to be able to get speedtest sites to show my speeds properly and not now. I also agree its strange that I can get good throughput from dl's too which does lead me to believe it could be something ultimately on this machine.

I cannot play trailers from IMDB on this XP SP3 desktop but have no problems on a laptop or iPad. I guess it is because of some adblocking soft I have installed. However I have tried disabling everything that I can find that may be involved with any success. I've given up and now watch trailers on Youtube. That's PC's for you.

I'm gonna hook up a Win XP machine shortly and do the same speedtests and see how that might go. Will post back if things look up :)

Cheers

Java normally gets updated whenever I see the message 'Java update available' on screen.

At this moment in time I am using the Superhub as a modem and from there directly to my pc. With or without my gaming router attached so far has made zero difference.
I hope you crack the issue and look forward to hearing the explanation. :)

[CPCUK]

I tried my XP machine which I use as a server only and scored 65Mbit on speedtest.net !!!! Tried again and back to 30Mbit, lol, and unable to get past it.

I'm not bothered by it as I know my actual speeds seem ok.

On my Win 7 pc I updated Java and checked my Windows spam blocker thingy was disabled. It actually was already off anyway.

I now think its something installed on my PC's which stop the speedtest sites from operating correctly.

When Windows 8 comes out I'll install it and then retry and see what happens.

Thanks.

[Socro]

Don't know if this is true or not, but some speedtest servers do not allow you to reach 100mb speeds.

[CPCUK]

Yes, that is true but the one I used to use and get reliable results on was speedtest.net and that one does cater for 100Mbit.

I'm trying to find out why it doesnt show up thats all as I do get the full 12MB/s or so using multiple downloads.

---------- Post added at 21:54 ---------- Previous post was at 21:44 ----------

Hmmm, not too sure whats just happened but I used the TCP Optimizer utility and set it to OPTIMAL settings. Rebooted, went to the Newbury test site and achieved 74.67Mbit down and 7.24MBit up (normally Upload speed is 9.2Mbit on London tests).

So, it looks like you at least have to just keep trying different servers until you hit the right one that works for you, lol, makes little sense really but hey, at least I can see 75Mbit at least!

[Chrysalis]

there is extra factors on speedtest.net

it goes via flash which adds significant cpu requirements.
also VMs peering is dodgy and could be affecting some speedtests, certianly namesco is capable of far higher than 100mbit as I have tested it with servers hitting 700mbit/sec.
rWIN may also affect it although with it been 4 threads I wouldnt expect it to have a significant effect on the 10mbit service. On old operating systems tho it can easily affect it on any other tier.

As a rule of thumb if things feel laggy/broken or my net has been dodgy anyway I will do a speedtest on speedtest.net.
If the test is slow I will then see if it gets backed up by manual downloading (usually of multi gigE servers I run). 95% of the time its accurate, occasionally its not and thats usually down to VM peering or core network issues.

[CPCUK]

Since last night I have tried new settings, cleared cache, checked this and that. Put settings back to default abd also the 'optimized' ones for TCP etc and I get the usual 30mbit. Some servers give me 1Mbit (lol).

Back to how it was but actual dl's are still fine.

[Chrysalis]

well in your case for whatever reason speedtest.net isnt showing you the capability of your connection speed although its highlighting a problem somewhere.

[qasdfdsaq]

Yeah, Namesco is one of the better UK ones.

Incidentally, most of the ones in the UK seem limited to ~100-200mb while most of the ones on the continent seem to be 1gbps.

[RB2004]

Netcetera also seems good think that's a newer addition but I get good results

[CPCUK]

An Update!

Well I've been messing about with Windows 7 TCP settings, Winsock and all sorts. Trying to get the best settings etc but in the end I just reset everything to Windows default.

As I have mentioned I know I get full speed Downloads as I have opened 14 different download files and counted the total MB/s achieved :)

I decided that I was going to try all the UK speedtest.net sites one after the other and have now done so.

The results are truly astounding. Whether all sites are capable of testing 100Mbit connections I dont know but my download speeds rantged from 2Mbit up to 103.3Mbit.

I have found that the Lancaster server is the best to test with (in my case), followed by Kingston Upon Hull. I'm located down the south of the UK but these servers are impressive!.

The Lancaster speedtest gives me consitantly over 100Mbit (104.44Mbit actual) download with 8.59Mbit upload (I get over 9.2Mbit upload on some of the other locations).

So, if your speedtest results arent that great and your SuperHub power levels etc look fine then at least try the other server test locations.


http://www.speedtest.net/result/1713877859.png

[Mick Fisher]

Namesco, London usually works for me.

Glad you have sorted things out. :)

[qasdfdsaq]

I just speedtested all the speedtest.net servers in the UK, Ireland, France, Belgium, Netherlands and Luxembourg. While Namesco still gives me a decent download speed on non-VM connections, Lancaster is actually coming out best overall.

[CPCUK]

Basically, I havent actually changed anything on my pc to get a good result.

I used to use London test server all the time when on 20mbit and that showed full speed. Then on 50mbit it was fine (showed 50Mbit) until I'm sure speedtest.net changed its site design (may be just me though) and I could only achieve 30Mbit results. It also depended on ping. If ping was over 25ms I got a good score and if under that it would suck! Strangely.

Lancaster I think was the 2nd furthest speedtest.net UK server from me yet gives great results (nice routing maybe?).

This just goes to show the potential for many customers to complain about their VM broadband speeds when it could be perfectly fine. Of course some connections do have issues. What is needed is a 100% reliable speedtest site / server and not some random speed one which can only confuse the poor guy testing his speed.

The only really reliable test is to use multiple downloads to max the advertised connection speeds and see what you actually obtain.

[Sephiroth]

CPCUK has confirmed what I've said in a few places about speedtest.net reliability.

Of late, I've found the London hosts to be unreliable when compared in the same systematic way as CPCUK has described.

I must admit I hadn't alighted on Lancaster, but Hull was always (and still is for me) pants. Until now, my benchmark has been Paris, they've added Massy (also OK) and Lancaster, because of the low ping time, I now trust.

So it can't be Flash. Flash isn't CPU intensive judging by the CPU graphs on my dual-core.

So I agree with CPCUK's conclusion that actual downloads are the acid test. But you have to find a site that can dispense at the full speed.

[Chrysalis]

When I am able to get full speed which is 30mbit on my quad core i5 it can use about 20% cpu power.

Assuming its 3.3x that requirement to do the 100mbit test then it needs quite a bit of juice. Given that I expect most people in the uk now browse on portable type devices as well like laptops and ipads.

[qasdfdsaq]

My best ever results currently are from Lancaster (as per my sig), and all the Paris ones - in fact all the France ones came out better than most of the UK ones.

Paris gave me 115-138 down and 75-127 up, which in itself is quite variable but I only did two tests. Massy was actually better - 210 down, 226 up. There seem to be two "classes" of speedtest servers - dunno if it's related to them using different IP settings (e.g. large windows), but they either give me ~20-40mbps upload results or 150+ with nothing in-between.

In fact, on some platforms (Win7x64) running the same test to the same server in IE could give me 200mbps upload but in Firefox only 40mbps, yet the same downstream result. Odd!

In case anyone's interested I decided to speedtest all* of speedtest.net's servers.



*By "all" I don't actually mean all. I'm adopting VM's fairly loose interpretation of "all", "everyone" and "unlimited". I actually just did the UK, Ireland, France, Belgium, Netherlands, Denmark and Luxembourg.

The results are below. They should really be taken as an sorted list, rather than absolute quality ratings.

I only did one test on most of them, under one condition at one time. Conditions may not reflect any particular situation on the VM network It gives us an idea of the relative speed capabilities of each server. The "Yes/No" columns are based solely on my opinion of their suitability for testing each tier.


Group 1 - Superfast uploads - Great if you have a symmetric line, fibre, etc.
Apart from a few with consistency issues, they're all good for 100mb+

Server Country Download Upload Ping 20/30mb 50/60mb 100/120mb

Lancaster UK 520.7 278.9 5 Yes Yes Yes
Luxembourg LU 304.8 226.9 18 Yes Yes
Massy FR 297.2 225.4 21 Yes Yes Yes
Manchester UK 246.2 158.8 11 Yes Yes Yes
Contern LU 235.4 211.0 19 Yes Yes Yes
Naaldwijk NL 220.6 161.0 25 Yes Yes Yes
Malmo DK 207.8 149.1 34 Yes Yes Yes
Merzig LU 196.8 157.2 31 Yes Yes Yes
Steinsel LU 196.7 228.2 19 Yes Yes Yes
Mechelen BE 192.9 182.6 27 Yes Yes Yes
Alkmaar NL 184.9 160.5 26 Yes Yes Yes
Clichy FR 183.4 190.1 19 Yes Yes Yes
Eindhoven NL 167.7 170.8 28 Yes Yes Yes
Bonneville FR 163.5 160.2 30 Yes Yes Yes
Aubervilliers FR 158.9 211.6 21 Yes Yes Yes
Paris FR 115.3 127.5 22 Yes Yes
Lyon FR 106.3 175.5 30 Yes Yes
Dronten NL 89.5 194.5 19 Yes
Copenhagen DK 66.5 154.9 36 Yes

Group 2 - Standard uploads. Fine for cable.
I'd be careful if you have 10mbps+ UL Infinity or other FTTx

Server Country Download Upload Ping 20/30mb 50/60mb 100/120mb

Cappelle Ij NL 364.5 36.4 21 Yes Yes Yes
Maidstone UK 277.8 57.6 11 Yes Yes Yes
Kingston UH UK 262.0 36.9 20 Yes Yes Yes
Rotterdam Qweb NL 248.5 51.5 20 Yes Yes Yes
Capelle NL 243.2 51.5 31 Yes Yes Yes
Roubaix FR 205.7 46.2 15 Yes Yes
Randers DK 188.3 54.3 39 Yes Yes Yes
Bournemouth UK 167.4 26.2 13 Yes Yes Yes
Veendam NL 167.2 16.6 22 Yes Yes
London UK 138.0 36.7 11 Yes Yes Yes
Dublin Eircom IE 137.7 16.6 30 Yes Yes
Ashford UK 128.3 42.3 12 Yes Yes
Sittingbourne UK 123.0 51.2 12 Yes Yes
Dublin Vodafone IE 98.5 16.5 26 Yes Yes
Amsterdam NL 96.5 30.9 23 Yes Yes
Liverpool UK 94.9 27.4 25 Yes Yes
Gloucester UK 94.9 27.1 25 Yes Yes
Dublin Digiweb IE 94.8 26.2 25 Yes Yes
Newbury UK 94.4 20.0 19 Yes Yes
Norwich UK 87.1 26.7 21 Yes Yes
Limerick IE 83.5 22.1 26 Yes Yes
Galway IE 81.4 30.2 23 Yes
Leeds UK 80.7 14.5 21 Yes
Skipton UK 64.9 20.2 25 Yes
Birmingham UK 64.4 30.0 21 Yes
Coventry UK 63.9 72.7 11 Yes
Rotterdam Luna NL 63.0 32.7 25 Yes
Guernsey UK 58.7 18.2 25 Yes
Brussels BE 57.3 61.8 22 Yes
Utrecht NL 55.2 25.2 36 Yes
Aberdeen UK 54.6 11.8 33 Yes

Group 3 - Epic fail servers. Crap at just about everything.
Probably best avoided for anything until they'e fixed.

Server Country Download Upload Ping 20/30mb 50/60mb 100/120mb

Canterbury UK 46.3 12.9 14 FAIL FAIL FAIL
Dundalk IE 36.5 8.5 25 FAIL FAIL FAIL
Arnhem NL 33.0 57.7 25 FAIL FAIL FAIL
Derry NI 28.8 23.6 25 FAIL FAIL FAIL

Some day I might get round to doing automated tests (e.g. 10+ per server) and working out some standard deviations to estimate stability. But not till I get my own connection stable.


---------- Post added at 21:25 ---------- Previous post was at 20:19 ----------

When I am able to get full speed which is 30mbit on my quad core i5 it can use about 20% cpu power.


Curious. I hit about 40% on my dual-core i7 (laptop) but only at 300mbps+. Given that an older 3Ghz Core 2 Duo has no issues exceeding 500mbps+ on Speedtest.net, I wouldn't say CPU is an issue.

[Sephiroth]

When I am able to get full speed which is 30mbit on my quad core i5 it can use about 20% cpu power.

Assuming its 3.3x that requirement to do the 100mbit test then it needs quite a bit of juice. Given that I expect most people in the uk now browse on portable type devices as well like laptops and ipads.
I don't see why 100 meg should use 3x CPU of 30 meg.

Anyway, I was addressing the point made that Flash uses a lot of CPU power. I didn't think so on current machines.

[Chrysalis]

seph thoughts on these figures.

30mbit ftp 0.9 %cpu usage
http 1.4% cpu usage
flash 19.6% cpu usage O_o

pingtest.net cpu usage 4.6% O-o for pings :LOL:

incidently on my sisters laptop when I tested speedtest here a while back it couldnt max out the speed as the cpu pegged to 100% during the test. On my laptop it hits 60% or so.

[Sephiroth]

Simples. 19.6% (or indeed 40%) is not a lot of CPU power. In any case, peops doing speed tests are hardly likely to be doing anything else on their PC for fear of affecting the test.