Geekzone: technology news, blogs, forums
Guest
Welcome Guest.
You haven't logged in yet. If you don't have an account you can register now.

View this topic in a long page with up to 500 replies per page Watch this topic Create new topic
1 | ... | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | 32 | 33 | 34 | 35 | 36 | 37 | ... | 235
1793 posts

Uber Geek
+1 received by user: 212


  Reply # 692008 27-Sep-2012 08:31 Send private message quote this post

Nice one! Ouch, yeah, those CRC errors o.O



1216 posts

Uber Geek
+1 received by user: 117

Trusted

  Reply # 692561 28-Sep-2012 00:39 Send private message quote this post

Well ain't that a b***h!



That didn't last long! :_(

After upping my SNRM to 16 I was still getting 1 CRCPM, so I hacked my config to up it to 17 which took my CRCPM down to just over 0.4 (at least for the short time that I was still on DLM-1), but unfortunately about 15min after I made that change, I was spontaneously resynced and now I'm on DLM-2 again... <sigh>.  Interestingly, as you can see, my CRCs are still pretty high even after I've been put back on DLM-2, so I'm not entirely sure what the story there is!

So, I'm not sure if DLM doesn't like an SNR of 17 or whether it had just had enough of all the CRC errors, but either way, I'm back to square one!

Not quite sure what I'm going to do now.  I could:
1) Leave it on 17db and hope that it's enough to get me back on DLM-1 (but even if I do, I doubt I'm going to stay there long)
2) Up the SNRM even more to try to reduce my CRCs to the point where DLM-1 is stable long term (but I'm not really willing to go below 30mbit downstream, which would happen if I went any higher with my SNRM)
3) Give in and tweak the crap out of my SNRM to get the maximum throughput possible, latency & CRCs be damned!
4) Get my line checked as I think based on the fact that I'm only 300m from the cabinet, in a new subdivision & have a fairly reasonable line attenuation figure I should be getting better error rates.  As a comparison, sidefx has 12db line attenuation, but his connection is faster and less error prone than mine!  I have always suspected that my line isn't optimum, I had lower than expected ADSL connect speeds as well.

I think I'm going to go for option 4 first and see if I can get some improvement and otherwise fall back to option 3 :)

1793 posts

Uber Geek
+1 received by user: 212


  Reply # 692671 28-Sep-2012 08:49 Send private message quote this post

Bummer man :( Have you tried changing the RFI setting and seeing if that makes a difference? I've got mine set one notch to the left, but did it at the same time as changing the SNRM by 1, so not sure which made the most difference.

I still sorta suspect that the way DLM "measures" user-initiated resyncs vs error resyncs is not as smart as you think and that may explain why it changed you not long after a resync. It would also explain why DLM bumped me up to the 16ms profile really quickly initially as that was around the time my son kept turning off my modem (and the fritzbox had the setting enabled to auto-resync each night)

96 posts

Master Geek


  Reply # 692945 28-Sep-2012 14:53 Send private message quote this post

Today is day 7 for me. I expected a resync today but I passed the 48 hour period with no resync. I was kind of expecting DLM1 maybe. I thought I'd do a reset of the router just to see.


No DLM change but I have synced 0.5Mbps faster and error rate has been really low for the last 5 hours since the reset. It was 0.11 before the reset.





378 posts

Ultimate Geek
+1 received by user: 48


  Reply # 692969 28-Sep-2012 15:38 Send private message quote this post

I was put back on DLM-1 last night:



The error rate is high, though actually quite a lot better than last time I was on DLM-1 when I was seeing as many as 400 CRC errors per hour. What is interesting is that my sync rate is quite a bit lower than it was last time (48000+), and actually it's lower than it was on DLM-2. The line attenuation is up 1db too, which is a bit odd. I think I will leave it for 8 days to see if DLM makes any changes and then resync manually.

I haven't changed the line settings at all. It'd been on DLM-2 since August 28th. Prior to that it was on DLM-1 since July 24th. Coincidence that these changes seem to be happening about a month apart?

96 posts

Master Geek


  Reply # 692977 28-Sep-2012 15:50 Send private message quote this post

stevehodge: I was put back on DLM-1 last night:



The error rate is high, though actually quite a lot better than last time I was on DLM-1 when I was seeing as many as 400 CRC errors per hour. What is interesting is that my sync rate is quite a bit lower than it was last time (48000+), and actually it's lower than it was on DLM-2. The line attenuation is up 1db too, which is a bit odd. I think I will leave it for 8 days to see if DLM makes any changes and then resync manually.

I haven't changed the line settings at all. It'd been on DLM-2 since August 28th. Prior to that it was on DLM-1 since July 24th. Coincidence that these changes seem to be happening about a month apart?


Wow your line stats are very similar to mine. When you were on DLM2 was kind of CRCPM rate were you seeing?

1793 posts

Uber Geek
+1 received by user: 212


  Reply # 693125 28-Sep-2012 19:51 Send private message quote this post

Hmm... I decided to bump my SNRM back down 1 to 13... I'll wait for it to settle down before taking a screenshot but error rates aren't looking too good. Current at 2 CRCPM and 42FECPM. Looks like SNRM at 14 might be the sweet spot on my line.



1216 posts

Uber Geek
+1 received by user: 117

Trusted

  Reply # 693135 28-Sep-2012 20:11 Send private message quote this post

@stevehodge - from what I have seen historically, you will need to keep your CRCPM at least under 0.4 and possibly as low as 0.2 (unconfirmed) in order to stay on DLM-1. You might want to look at upping your SNRM a bit to get your errors under control.

.@sidefx - Yeah, it's hard to say, but I'm going on what the official Chorus docs say on the subject.

1793 posts

Uber Geek
+1 received by user: 212


  Reply # 693139 28-Sep-2012 20:21 Send private message quote this post

SamF: 
.@sidefx - Yeah, it's hard to say, but I'm going on what the official Chorus docs say on the subject.


But did the chorus docs actually spell out how they detect a user initiated resync? or do they just say something along the lines of "user initiated resyncs don't count towards MTBE"? It might be that they then define "user-initiated" as:

A resync where the modem is down for more than 180 seconds
or
The first 2 resyncs per week

or something along those lines, which looks like how BT defines user resyncs (in simplistic terms) in their patent.





1216 posts

Uber Geek
+1 received by user: 117

Trusted

  Reply # 693141 28-Sep-2012 20:23 Send private message quote this post

No, they didn't spell out how this was detected unfortunately so you may well be right.

378 posts

Ultimate Geek
+1 received by user: 48


  Reply # 693158 28-Sep-2012 21:04 Send private message quote this post

helis:
Wow your line stats are very similar to mine. When you were on DLM2 was kind of CRCPM rate were you seeing?

Pretty low as you can see from the graph - the errors jumped at 4am when DLM resync'd. The highest in the graph before that was 6 in the hour between 11pm and 12am which equates to 0.1 per minute.


SamF: @stevehodge - from what I have seen historically, you will need to keep your CRCPM at least under 0.4 and possibly as low as 0.2 (unconfirmed) in order to stay on DLM-1. You might want to look at upping your SNRM a bit to get your errors under control.

To be honest I'm not sure CRC per minute is how DLM is looking at this. One problem I have with CRCPM is that I think the error rate is going to be higher with higher usage (after all you can only have an error if there is data being transmitted). I believe it is more likely that DLM is looking at the number of successfully transmitted bits per CRC error or something similar. That means CRCPM is only half the story - you also need to look at how much data was transmitted.

For me I'm still seeing too many strange things going on with DLM to try to tune for it. For example, why did DLM decide to put me back on DLM-1 now? We believe it makes decisions at least weekly and nothing has changed for me this week versus the last 2 or 3 weeks. Another example is DLM resyncs without changes - why? Or your switch and then switch back a day later - I was DLM-1 for a month with worse error rates than you so why did it downgrade you after only 1 day and yet leave me for a month?

I'm going to leave it alone for at least a week and see what happens.



1216 posts

Uber Geek
+1 received by user: 117

Trusted

  Reply # 693183 28-Sep-2012 21:50 Send private message quote this post

stevehodge: To be honest I'm not sure CRC per minute is how DLM is looking at this. One problem I have with CRCPM is that I think the error rate is going to be higher with higher usage (after all you can only have an error if there is data being transmitted). I believe it is more likely that DLM is looking at the number of successfully transmitted bits per CRC error or something similar. That means CRCPM is only half the story - you also need to look at how much data was transmitted.


I'm only going on the Chorus docs which state that DLM makes its determination based on mean time between errors.  This has a direct mathematical correlation to CRCPM.

stevehodge:For me I'm still seeing too many strange things going on with DLM to try to tune for it. For example, why did DLM decide to put me back on DLM-1 now? We believe it makes decisions at least weekly and nothing has changed for me this week versus the last 2 or 3 weeks. Another example is DLM resyncs without changes - why? Or your switch and then switch back a day later - I was DLM-1 for a month with worse error rates than you so why did it downgrade you after only 1 day and yet leave me for a month?.


Yes, there are some odd / unknown aspects to DLM.  I believe that there are some DLM settings which are not visible on the client side.  As for weekly changes, DLM is able to make changes weekly, but it doesn't necessarily mean it will make a change every week.  It is also able to make changes much more quickly if it deems the line to be too unstable.  I also think, based on a number of people's experiences, that a few weeks ago, DLM wasn't making changes for anyone which could explain your month on DLM-1 without any changes.

96 posts

Master Geek


  Reply # 693196 28-Sep-2012 22:27 Send private message quote this post


To be honest I'm not sure CRC per minute is how DLM is looking at this. One problem I have with CRCPM is that I think the error rate is going to be higher with higher usage (after all you can only have an error if there is data being transmitted). I believe it is more likely that DLM is looking at the number of successfully transmitted bits per CRC error or something similar. That means CRCPM is only half the story - you also need to look at how much data was transmitted.



I thought this too but I have downloaded 10gig in one hit at speeds of 2-3MB/s and had no CRC errors during the 40mins or what ever it was and now I've come back to see I've had 4 CRC errors in the last 15mins. I doubt I've had 50meg of data transfer over the link in the last 15mins.

1793 posts

Uber Geek
+1 received by user: 212


  Reply # 693199 28-Sep-2012 22:46 Send private message quote this post

Well, error rates looked way too high, so I decided to switch back to SNRM of 14!  Unfortunately my error rate is still looking quite high...  If it ain't broke don't fix it.



helis: 
I thought this too but I have downloaded 10gig in one hit at speeds of 2-3MB/s and had no CRC errors during the 40mins or what ever it was and now I've come back to see I've had 4 CRC errors in the last 15mins. I doubt I've had 50meg of data transfer over the link in the last 15mins.


Yeah I sort of thought this too, but have been watching CRC errors on and off in relation to downloads and I'd say anecdotally they're unrelated, so not really sure how they measure them.






1216 posts

Uber Geek
+1 received by user: 117

Trusted

  Reply # 693200 28-Sep-2012 22:50 Send private message quote this post

I've found that sometimes changing SNRM doesn't seem to make any difference to error rates until you restart the router.

1 | ... | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | 32 | 33 | 34 | 35 | 36 | 37 | ... | 235
View this topic in a long page with up to 500 replies per page Watch this topic Create new topic

Twitter »
Follow us to receive Twitter updates when new discussions are posted in our forums:



Follow us to receive Twitter updates when news items and blogs are posted in our frontpage:



Follow us to receive Twitter updates when tech item prices are listed in our price comparison site:




News »

Trending now »
Hot discussions in our forums right now:

Moment of Truth?
Created by BarTender, last reply by Hammerer on 19-Sep-2014 13:52 (385 replies)
Pages... 24 25 26


Mr. Key to extradite Kim Dotcom?
Created by TimA, last reply by SaltyNZ on 18-Sep-2014 09:20 (126 replies)
Pages... 7 8 9


Spark DNS Issues - Amazing - Broadband Service Alert
Created by PeteS, last reply by Demeter on 15-Sep-2014 14:13 (307 replies)
Pages... 19 20 21


IOS8 - Network Load
Created by FireEngine, last reply by FireEngine on 19-Sep-2014 06:20 (37 replies)
Pages... 2 3


2014 Holden SS (V8) or Ford XR6-T (in-line 6 turbo)
Created by joker97, last reply by ilovemusic on 16-Sep-2014 14:34 (71 replies)
Pages... 3 4 5


Maybe some politicians should go back to school?
Created by jarledb, last reply by DarthKermit on 18-Sep-2014 18:27 (31 replies)
Pages... 2 3


FS: Server time
Created by deathslocus, last reply by CamH on 17-Sep-2014 21:31 (30 replies)
Pages... 2


Warning: Rage Ahead - Campbell Live and childhood poverty
Created by kawaii, last reply by gzt on 15-Sep-2014 22:57 (231 replies)
Pages... 14 15 16



Geekzone Live »
Try automatic live updates from Geekzone directly in your browser, without refreshing the page, with Geekzone Live now.

Are you subscribed to our RSS feed? You can download the latest headlines and summaries from our stories directly to your computer or smartphone by using a feed reader.

Alternatively, you can receive a daily email with Geekzone updates.