Menu
Reply
  • 515
  • 27
  • 122
deviousiphone
Rising star
991 Views
Message 1161 of 4,479
Flag for a moderator

Re: Hub 3 / Compal CH7465-LG (TG2492LG) and CGNV4 Latency Cause

PhilMull, this is half the problem, that they do that, so everyday deliberately inflict this problem onto new customers, it could be argued that they can't help the Puma 6 fault, but it becomes mute when they hand it over to people deliberately!

  • 526
  • 25
  • 191
RidingTheFlow
Rising star
984 Views
Message 1162 of 4,479
Flag for a moderator

Re: Hub 3 / Compal CH7465-LG (TG2492LG) and CGNV4 Latency Cause

It looks like management is afraid to admit that SH3 has a problem and they think that hiding the issue is better somehow. Therefore they don't inform even own employees - which creates this mess.

 

Highlighted
  • 17
  • 0
  • 8
JJC1138
On our wavelength
957 Views
Message 1163 of 4,479
Flag for a moderator

Re: Hub 3 / Compal CH7465-LG (TG2492LG) and CGNV4 Latency Cause


@Gameruk wrote:
Can i leave my contract now because of the prise rise in november or do i have to wait for a email confirmation

I've been wondering that too. The precise wording of the contract clause is "The notice of termination must be given within 30 days of the increase in charges or changes to this agreement being notified to you" so if they follow that exactly then we have to wait. It's possible that they might let you cancel early, but I can imagine a scenario where the person you speak to says that's fine and puts through the cancellation, but then the system sees that they haven't sent you the notification yet and charges the early termination fee. So I'm not gunna risk it and will just keep eagerly checking the post every morning like I'm waiting for my flippin' Hogwarts letter.

  • 87
  • 0
  • 62
Xymox
Up to speed
832 Views
Message 1164 of 4,479
Flag for a moderator

Re: Hub 3 / Compal CH7465-LG (TG2492LG) and CGNV4 Latency Cause

We have had some pretty good discussion about the technical aspects of the issue over at DSLR. Mackey, who discovered the DoS and got root on the Arris SB6190 and has looked really carefully at whats going on has provided some some depth on whats causing the issue. This really looks untenable.

 

Intel has said officially they have started to release some patch to cable modem companies. We dont know who or what models. After the modem vendors work on it a bit it will go to ISPs and ISPs will work on it a bit and then it will come to us as users. We are weeks to months away.

No one in our discussion sees a way to fix the DoS and keep performance. The spike is just a symptom of the main issue. So it might not be possible to fix the spike without reducing the modem speed to like 60mbps.

Really the options for Intel seem to be:

1. Recall/replacement - staggering cost/lose market
2. Fix DoS fully, but at a loss of speed - lose law suit and market, still face recall because of not meeting speed tiers.
3. Rewrite everything using a new RTOS - huge development time, huge costs, stay in market & maybe have best chip.
4. Address DoS somehow keeping speeds but performance still abysmal because of OS. - Lose lawsuit, online reputation continues to ebb away at sales, slowly lose market.
5. New silicon + new RTOS. Best solution but may take years along with big investment.
6. MUCH faster silicon with bigger table, Same OS - Most issues resolved, still not as good as Broadcom but closer. Would need to sell cheaper VS Broadcom.

7. Just keep going, take lawsuit up the butt, stagger forward keeping sales going because of monopolies. **** US-CERT they have no legal powers. Keep paying people under the table to use Pumas. Ignore online BS as irrelevant.

  • 562
  • 9
  • 160
Guybrush85
Rising star
797 Views
Message 1165 of 4,479
Flag for a moderator

Re: Hub 3 / Compal CH7465-LG (TG2492LG) and CGNV4 Latency Cause

A lot of pages moved over the weekend, but unfortunately nothing in terms of any news from VM. 

As it's been said and done I think a lot in this thread, the idea of VM allowing other kit is a pipe dream. It's ok saying "they could", you could go all day with what businesses could do, but there is a reality here and that is that until a new piece of a hardware comes out, the hub 3 is what we've got. If you're fortunate enough to have a 2/ac, then good for you, but buying one off eBay or shouting at the staff won't make them suddenly start producing stock they simply don't have, or provisioning old equipment that shouldn't have been sold in the first place.

Xymox, you were almost correct in the it's Intel/Arris thing, but VM do have a hand in this. Whilst they didn't cause the issue and it's not within their remit to come up with the solution, they were aware of issues with this hub during the trial. At first when getting the hub, we were all provisioned for 8x DS channels and in terms of the latency issue, there appeared to be no difference between the older hubs and the hub 3. When the CMTS in areas were upgraded to either an Arris E6k or c-BR8 and hubs we started getting 16, 20 or 24 channels did an obvious issue happen and all of the BQM graphs we have start spiking all over. This was ignored, or at least thought to be a false positive and despite that (amongst other things), the hub was rolled out.

We talk about better testing, but even when that happens, it doesn't mean companies will act on it unfortunately. Hopefully the testing is pushed further up the chain and these kinds of issues aren't highlighted by UAT users like me and you.

The information from Datalink about Docsis 3.1 is intersting. Based on how certain variables reflect in the severity of the issue (such as DS channels from my example above), I was interesting to know how a puma 7 device acted 3.0 vs 3.1. Xymox, I don't think anyone's trying to say it's not an issue, it's just technically the differences are interested, so I think your passion for this maybe seems to be a bit missplaced/heated in that response.

Anyway, I will try and do an update later when I get some time and add in new/additional info.

--------------------------------------------------------
Look behind you, a three-headed monkey
  • 87
  • 0
  • 62
Xymox
Up to speed
741 Views
Message 1166 of 4,479
Flag for a moderator

Re: Hub 3 / Compal CH7465-LG (TG2492LG) and CGNV4 Latency Cause


@Guybrush85 wrote:

A lot of pages moved over the weekend, but unfortunately nothing in terms of any news from VM. 

As it's been said and done I think a lot in this thread, the idea of VM allowing other kit is a pipe dream. It's ok saying "they could", you could go all day with what businesses could do, but there is a reality here and that is that until a new piece of a hardware comes out, the hub 3 is what we've got. If you're fortunate enough to have a 2/ac, then good for you, but buying one off eBay or shouting at the staff won't make them suddenly start producing stock they simply don't have, or provisioning old equipment that shouldn't have been sold in the first place.

Xymox, you were almost correct in the it's Intel/Arris thing, but VM do have a hand in this. Whilst they didn't cause the issue and it's not within their remit to come up with the solution, they were aware of issues with this hub during the trial. At first when getting the hub, we were all provisioned for 8x DS channels and in terms of the latency issue, there appeared to be no difference between the older hubs and the hub 3. When the CMTS in areas were upgraded to either an Arris E6k or c-BR8 and hubs we started getting 16, 20 or 24 channels did an obvious issue happen and all of the BQM graphs we have start spiking all over. This was ignored, or at least thought to be a false positive and despite that (amongst other things), the hub was rolled out.

We talk about better testing, but even when that happens, it doesn't mean companies will act on it unfortunately. Hopefully the testing is pushed further up the chain and these kinds of issues aren't highlighted by UAT users like me and you.

The information from Datalink about Docsis 3.1 is intersting. Based on how certain variables reflect in the severity of the issue (such as DS channels from my example above), I was interesting to know how a puma 7 device acted 3.0 vs 3.1. Xymox, I don't think anyone's trying to say it's not an issue, it's just technically the differences are interested, so I think your passion for this maybe seems to be a bit missplaced/heated in that response.

Anyway, I will try and do an update later when I get some time and add in new/additional info.


I think you should read up on the technical issues discussed by Mackey recently so you have a better understanding of the issues in the links I provided a page back. The DoS issue wont be effected by the number of channels or docsis 3 VS 3.1 as its simple the PPS and a too small table. This issue is present in Puma 5, 6 and 7. Its pretty clear this cannot be fixed with a firmware update without serious side effects. This underlying issue of too small a table size effects a great deal of different network flows. On top of this the single core design running Linux rather then a real time OS creates issues that really cannot be overcome without more cores. The silicon seems to be fundamentally flawed in design.

Discussing number of DS channels and 3.0 VS 3.1 and how this effects things is almost irrelevant as the chip is clearly flawed and firmware does not appear to have much hope of addressing it.

If you feel otherwise I would encourage you to come have a discussion over at the thread on DSLR as we have a number of people there who have a good understanding technically of the issues involved along with Mackey who has root on a Puma 6 modem.

What concerns me is that everyone needs to stay focused and continue to apply pressure. Distractions like effects from the number of DS channels and 3.0 and 3.1 cant help resolve the issue. We know what the problem is now. Problem isolation is over. Its the chip and it appears to be baked into the silicon. VM needs to stop pushing these into the market as they appear to be flawed. Any work toward a Hub 4 based on the Puma 7 need to halt. Broadcom needs to be brought into the picture. This is going to be a difficult road. However you guys across the pond need to start pushing this.

So my views on this are that discussion about problem isolation are over. Its time to discuss how to fix the problem, how to move VM over to Broadcom.

  • 562
  • 9
  • 160
Guybrush85
Rising star
689 Views
Message 1167 of 4,479
Flag for a moderator

Re: Hub 3 / Compal CH7465-LG (TG2492LG) and CGNV4 Latency Cause

I think you're missing where I am going.

Ok, so perfect world for us on VM is we either get a Hub 4 that has a broadcom chipset and we get a firmware patch. We aren't getting 2ac's getting put back in stock and we aren't getting 2nd hand modems being provisioned.

So... if we postulate that the hub 4 is a puma 7 and the rumours are true and that it comes alongside 3.1, then as we have no access to said device, it is of interest to know the severity of it like for like with a puma 6 (3.0) and the when using Docsis 3.1. Why? Well:

  1. It could mean that VM take the stance of, ok, it's not great, but it's better and go with it anyway.
  2. If it's the same or worse, it might open the door to a broadcom device, which would be good (however doubtful)

So yes, it might not be 'helping the case', but we're talking about the hub 3 now, which was pushed out regardless of the issues at the time. 

--------------------------------------------------------
Look behind you, a three-headed monkey
0 Kudos
  • 166
  • 1
  • 16
fizzyade
Dialled in
654 Views
Message 1168 of 4,479
Flag for a moderator

Re: Hub 3 / Compal CH7465-LG (TG2492LG) and CGNV4 Latency Cause

God I wish we could use a cable modem of our own choosing, they could even carry on supplying their flawed SH3 as a freebie with the "it works fine for most people".

Of course, it's not going to happen, so we'll just have to wait until the SH4 and hope that it isn't based on a flawed chipset.

  • 213
  • 2
  • 74
boltedenergy
Superfast
612 Views
Message 1169 of 4,479
Flag for a moderator

Re: Hub 3 / Compal CH7465-LG (TG2492LG) and CGNV4 Latency Cause


@fizzyade wrote:

God I wish we could use a cable modem of our own choosing, they could even carry on supplying their flawed SH3 as a freebie with the "it works fine for most people".

Of course, it's not going to happen, so we'll just have to wait until the SH4 and hope that it isn't based on a flawed chipset.


So just thinking through this, the premise is that you normally buy any cable modem and it should work theoretically with most cable service providers?

Then the reason for VM not allowing this at all is more a business decision then a technical one. What for shane to come out and tell us how extremely difficult it would be for VM to do this. sod that though! If they really wanted to, they could make the required changes to their network for anyone greatly affected by the issue with a nice big disclaimer that they wont support you if you end up having any modem issues.

But that wont happen as we know as VM decided to lock their network down and only provide one type of modem, certainly not in the interest of helping their customers have more choice...

 

  • 526
  • 25
  • 191
RidingTheFlow
Rising star
574 Views
Message 1170 of 4,479
Flag for a moderator

Re: Hub 3 / Compal CH7465-LG (TG2492LG) and CGNV4 Latency Cause


@boltedenergy wrote:

Then the reason for VM not allowing this at all is more a business decision then a technical one. What for shane to come out and tell us how extremely difficult it would be for VM to do this. sod that though! If they really wanted to, they could make the required changes to their network for anyone greatly affected by the issue with a nice big disclaimer that they wont support you if you end up having any modem issues.

 


 Oh no, its not just for support reasons! Virgin Media is scared crazy of allowing third-party devices on the network - because several years ago they had massive troubles with hacked/cloned modems. Since then they tried everything to tighten security and now it lead to situation when they over-tightened it, and still scared (even when DOCSIS 3 is much more secure than DOSCIS 1 was), unable ever to re-provision own devices they have on hand Smiley Wink

Here is and example from these times back - https://www.theregister.co.uk/2009/03/19/virgin_media_cloned_modems/