Print

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

[banger]

Since the cold weather has come my DS attainable has gone up from 80 to 82 and SNRm from 3.3 to 3.6db DS. Also my US SNRm has gone from 6 to 6.7db. I used to get better speeds in cold weather on ADSL and ADSL2+ is this the same sort of symptom?

[22over7]

I've very little trust in whatever algorithm it is that computes one's "attainable" speed. I think it is a confabulation between the modem and .. something else, like the exchange.
But I'd be delighted if someone could persuade me it means something.

My DS attainable is a shade below 100Mbps. (I've lost a whopping 3Mbps in the last 9 months!) It syncs at 80Mbps. I get in reality data throughput around 72Mpbs .
My US attainable is about 22Mpbs, It syncs at 20Mpbs. I get in reality about 18Mbps. (I'm on an 80/20 Plusnet FTTC contract.) All SNRM's waggling around 7dB.

My explanation, to myself, is that my DS is massively interleaved (INP 16, INP 48, 20-something ms latency)), and that about 3/4 of my DS throughput is the transmission of error-correcting/detecting bits. (My line is terrifically stable.) My US, about 1/4 of the DS, doesn't bother so much to correct things, so is the main source of the errors my line experiences. I only got these numbers after capping my line for several months.

Anyway, what does "attainable" mean? Is it remotely interesting, or even meaningful?

[j0hn]

Probably a small crosstalker has ceased their service or had a pair swap.

The attainable calculated by Broadcom chipsets is pretty consistent.

[ejs]

The "max attainable net data rate" is an estimation of the line speed you would get if it re-connected now, based on the current conditions.

My explanation, to myself, is that my DS is massively interleaved (INP 16, INP 48, 20-something ms latency)), and that about 3/4 of my DS throughput is the transmission of error-correcting/detecting bits.

Did you mean D=16 INP=48? That's using retransmission (G.INP), there is no massive interleaving. It's a tiny little amount of interleaving, arranged so that the interleaving covers each retransmission unit (a DTU), which aren't very big.

It's also important to note that the max attainable and actual net data rate figures are calculated already taking into account FEC/interleaving overheads, you don't subtract them again.

[22over7]

Thank you both ejs and john. Your remarks are valuable, and I'll try to understand them.

@j0hn:

My explanation, to myself, is that my DS is massively interleaved (INP 16, INP 48, 20-something ms latency)), and that about 3/4 of my DS throughput is the transmission of error-correcting/detecting bits.

Did you mean D=16 INP=48? That's using retransmission (G.INP), there is no massive interleaving. It's a tiny little amount of interleaving, arranged so that the interleaving covers each retransmission unit (a DTU), which aren't very big.

Gosh! Sorry for probably  (1) screwing up quoting, (2) general misunderstanding of G.INP. I had thought that the effect of retransmission would be more drastic on latency. Where can I read more about G.INP?

@ejs:
I('m sorry but I)  don't know exactly what "line speed" means, as opposed to "net data rate", or if I have even seen them anywhere, on DslStats, or MDSLW. Please forgive me. There is a lot of difficult-to-remember-terminology in this area.

[Moderator edit in an attempt to fix the quotes.]

[tubaman]

You can read all about G.INP at http://www.kitz.co.uk/adsl/retransmission.htm
 

[ejs]

My apologies. I tend to use "line speed" or "line rate" because I don't really like the term "sync speed". I believe the proper technical term is "net data rate". It's like net pay, the tax (FEC overhead) has already been deducted.