Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[kitz]

OSX flash has an oddity that single thread connection max's out at 12 Mbps.

Thank you Stuart, I think you may have highlighted and brought to our attenuation, an important fact that Ive not seen mentioned before, but should perhaps be more widely known. 

since OSX is *nix based it is perfectly feasible that Linux has a similar issue albeit not quite as restrictive.

Now that you mention it I recall having a conversation last December with someone on Linux who was having the same issue.  He thought it may have been MTU, but despite all his efforts couldnt resolve it, so it may be wider spread than we think.   Perhaps its not been as noticeable before due to the fact that not many lines had such high speeds to let it be so obvious.

[broadstairs]

Yes I certainly think I had not noticed it on my ADSL2+ 8mbps connection and it's probably likely no one on ADSL lines would see it but get to FTTC and it becomes an issue. It would be interesting to see a single threaded speed test which is not flash driven to compare but I dont believe there are any, at least none I've found.

Stuart

[kitz]

This may be interesting.  No flash or Java

http://speedof.me/

SpeedOf.Me also downloads large contiguous sample files. This is what you usually do when downloading a webpage or media. This is while other speed test services download small chunks in parallel and do heavy adjustments to calculate your approximate speed.

By contiguous it seems to imply separate files once after another (and not parrallel/multi-threaded], which is what I observed when running it.   Seems to work fine for my downstream, but my upstream results are off :/

[broadstairs]

Works fine for me, even upload but then I only have 2mbps up 

I'm thinking of trying an upgrade to 80/20 with TT as you can select it to try for 30 days and if then you find no benefit you can switch back.

Stuart

[burakkucat]

This may be interesting.  No flash or Java

http://speedof.me/

I have been using that throughput speed tester ever since Eric mentioned it a year, or so, ago.

[kitz]

I have been using that throughput speed tester ever since Eric mentioned it a year, or so, ago.

Oops I must have either missed Erics post or forgotten    I found it via google.   Ive now added it to the list of Speedtesters on the main site, so hopefully I wont forget again. 

[burakkucat]

No harm done, I'm sure. 

[Chrysalis]

This may be interesting.  No flash or Java

http://speedof.me/

SpeedOf.Me also downloads large contiguous sample files. This is what you usually do when downloading a webpage or media. This is while other speed test services download small chunks in parallel and do heavy adjustments to calculate your approximate speed.

By contiguous it seems to imply separate files once after another (and not parrallel/multi-threaded], which is what I observed when running it.   Seems to work fine for my downstream, but my upstream results are off :/

it claims to be accurate, yet according to that site I Can download faster than my sync speed, and the upload speed maxes out at a very low level suggesting they have the buffer set way too low for fast upload connections.

--edit--

interesting the sendbuffer setting in firefox works for this speedtest.

it was on 65k (the default is 128k) and that gave me approx 2mbit.
at 256k it gave me 12mbit
at 512k it maxes my upstream and I get a result of 18.19mbit.

downstream still is over my sync speed tho

default would probably give about 5-6mbit.

[broadstairs]

Now I have got my account sorted with TT (a couple of minor glitches) I have just ordered an update to Fibre Large. Nice thing about TT is that I can try this for 30 days and if it turns out not to be worthwhile I can turn it off again and revert back to Medium. So in a couple of (working) days it should switch me over.

Stuart

[roseway]

I hope it goes well Stuart.

[broadstairs]

Well it switched just after midnight and now my sync is 72120/20000 kbps but having run a few speed tests my actual speed is around 44000 kbps down and around 16000 kbps up. My snrm is now 6.4db down and my CRC and ES error rates have gone up from practically zero to an average of 15529 crc/day and 549 ES/day. So unless I can find out what is causing this and find a way of increasing my speeds this experiment will not last beyond the 30 days I have signed up for.

I'll let it settle for today and then tweak the snrm up to see if I can ding a value which maximises speed and minimises errors and only if it ends up much better than I was getting will I stay on this upgrade.

Any advice gratefully received 

Stuart

[broadstairs]

Does anyone know if the DSLStats snrm tweak firstly works on the ZyXEL and second if it does then will it work on an FTTC connection. I just tried it but my snrm did not re-sync any differently after trying a +3db increase using DSLStats. Still sitting at 6.4 db with 24 hours average at 22000+ CRCs!

Stuart

[roseway]

I don't think that you can tweak the SNRM on FTTC by the normal method. The DSLstats SNRM tweak should work with any compatible modem/router, but not on FTTC.

[broadstairs]

I don't think that you can tweak the SNRM on FTTC by the normal method. The DSLstats SNRM tweak should work with any compatible modem/router, but not on FTTC.

Eric thanks. Do you or anyone else know any abnormal  method for doing this?

Stuart

[Dray]

Some Broadcom based modems allow you to cap the sync speed I believe, which will cause the SNR Margin to rise