Menu
Reply
  • 58
  • 0
  • 9
yabba
On our wavelength
819 Views
Message 2951 of 3,074
Flag for a moderator

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


@wrote:

It isn't a single lost packet, it's eight-twelve entire frames that we're talking about.  I'd agree that a few lost packets are neither here nor there, but when you're have a PC trying top catch up on ten complete screen refreshes, that's rather a lot of data in a fast moving game.  There's still only the two options I've indicated - the PC has to make up and blend two different images, or there's going to be discontinuity on the screen.


 

No there is not any "lost frames". Your screen might run at 120hz, game servers do not.

The client renders a continuous smooth, interpolated image. If there's any jumping or warping this only happens in the event of extreme packet loss
and this has no bearing at all on texture mapping or screen tearing.

Look, people are playing games all the time with pings over 100. b4nny has a recent video on his twitch of
him playing on European servers from the West coast. His ping was around 160. His screen wasn't tearing and
glitching. He still played significantly better than everyone else.

Of course, that's not to say VM, intel et al don't have to fix this issue, but if your games or web browsing is a laggy mess or
you're losing a ton of packets, are warping around in games etc, or if you suck and get killed a lot then that's really nothing at all to do with this issue.

This issue is barely noticeable and it's typically 5 or 6 packets out of 200 on the pump 6 test (far more packets are delayed by the test than by the issue)

I gave you a link, you have much to read and learn about lag compensation in games - this thread is not really the place to discuss it.

0 Kudos
Reply
  • 3
  • 0
  • 0
halewoodj
Joining in
731 Views
Message 2952 of 3,074
Flag for a moderator

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

Another Hub 3 user here, who has now had the new firmware pushed down:

https://www.thinkbroadband.com/broadband/monitoring/quality/share/d49117c0bf5d6be38b4289e40e1ba7e1c408a698-13-02-2018 

Same version as others mention - it's pretty clear on the above graph where the update took place!

Sadly I still get a good amount of red on the puma-test, but I'm thinking that may now be another issue with our line - need to work out if the signal levels and such are all in tolerance.

 

0 Kudos
Reply
Highlighted
  • 325
  • 39
  • 135
Andruser
Fibre optic
707 Views
Message 2953 of 3,074
Flag for a moderator

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

@ yabba

" but if your games or web browsing is a laggy mess or you're losing a ton of packets, are warping around in games etc, or if you suck and get killed a lot then that's really nothing at all to do with this issue."

Go back and read what I actually wrote before pontificating.  If you don't think this is the place to discuss it, then don't respond, but it seems to me you're still not grasping the practical impact of a 150-250ms lag spike on a client/server game, but you're now trying to close down the discussion and have the last word.

"I gave you a link, you have much to read and learn about lag compensation in games - this thread is not really the place to discuss it."

The only reason it has become a discussion point is because you and a couple of others are denying there's any effect of the Puma 6 lag spiking.  If that's the case, why have they spent a year trying to work out a fix?  Why have so many people been complaining bitterly, and some even taken their business elsewhere?

0 Kudos
Reply
  • 34
  • 1
  • 2
philjohn
Tuning in
661 Views
Message 2954 of 3,074
Flag for a moderator

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

Wasn't your complaint that it was causing screen tearing? I don't think anyone has said that latency spikes DON'T cause an issue for twitch shooters and their ilk (or games with woeful net code - which is often overlooked, I've written net code for a game, it's one of the hardest parts to get right).

0 Kudos
Reply
  • 87
  • 2
  • 82
Datalink
Up to speed
621 Views
Message 2955 of 3,074
Flag for a moderator

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

With reference to the above discussion, the correct answer is:

A.  some of the above;

B.  none of the above;

C.  all of the above;

D.  C, or the longest answer

 

Personal opinion, the answer is C.  all of the above.  

The one point that everyone seems to be forgetting is that with the Puma 6 modem, if you're running UDP through the modem, which a good many games do, you have a combination of latency and packet loss.  Running GRC's DNS survey, I've seen losses up to 9% on a Puma 6 modem.  So, there are a couple of issues here.  Firstly, are you all talking about the same online game.  I suspect that the answer is no.  That will change the discussion as potentially, the amount of UDP traffic could have a very wide range, and as a result, so will the observable effects, game to game.  Secondly, there is no question that there is UDP packet loss with the Puma 6 modem.  The question then becomes, what effect does that have on the game graphics.  Does the game deal with the combination of latency and missing packets gracefully, or, as has been indicated, is there an observable discontinuity frame to frame.  I suspect that games are designed with a certain amount of packet loss and latency in mind, but, if you encounter an extended number of UDP packet losses combined with high latency, as might be possible with a Puma 6 modem, how does the graphics react?  That will probably change, game to game and will depend on how much effort the game designers put into handling missing or highly delayed packets.  If I'm thinking about this correctly, the game servers itself won't transmit game graphic data.  Please correct me if I'm wrong on that one.  But, what would happen is that the positional data, on your game display would not be correct, both for yourself and for the other players in the game.  As your UDP packets are either lost or highly delayed in transit thru the Puma 6 modem, the game plays on, with everyone moving, including yourself.  After a series of missing and/or highly delayed packets, the next received update should correct everyone's position in the game world.  I would expect to see position jumps and possible screen discontinuities as the positional differences may exceed the games capability for a graceful change from one frame to the next.  So, you might not see a huge change frame to frame, but I would expect to see some discrepancy in the frame itself, similar to a tear in the image.  Then the question of graphics sync modes comes into play.  How is that affected when the player jumps from one position to another and his or her view of the game world suddenly changes?  I'm not here, I'm actually 10, 20 feet away, or further.  Your view of the game world around you should change, its just a question of how much it changes.  Does freesync or G-sync for example hold the image until the image buffer is refreshed and how does either sync system know that the whole image is correct when there is a position jump that has occurred?  That's yet another question in the mix.  So, at end of the day there are a few factors that go into this discussion which could lead to a range of effects, from minimal to absolutely noticeable.  Arguing about it makes no sense unless you're all on the same game, recording the game graphics output with a high speed video recorder to settle any arguments about the contents of the frame image.  The question of the day is whether or not the update will have the positive effect that you're looking for?  

On a related note, if you're looking to test UDP, follow the instructions in my post, which is located in the DSLReports forum:

http://www.dslreports.com/forum/r31737637-

 

That will run a one second interval DNS query to your chosen DNS address, which in this case should be the VM DNS address.  The result is two way UDP traffic.  That instruction plan is only for a one second interval, so, kill the BQM and let this run for a 24 hour period to see if in fact there is any change in the UDP throughput.  As this only runs a one second interval query, you'll need the longer term test to see what the results look like.  I'm working on posting a faster scripted instruction set for ICMP, TCP/IP and UDP.  (Time to hurry up and post it Smiley Sad)That will really show if there are any changes to any of the above protocols with the new update.  Fwiw, if this is the same update that was pushed out to the Rogers modems for ICMP purposes, Rogers completed that update in Sept 2016.  

0 Kudos
Reply
  • 196
  • 2
  • 39
wotusaw
Superfast
575 Views
Message 2956 of 3,074
Flag for a moderator

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

Suddenly my cable test was aweful. Think I found the answer..cable double test.jpg

0 Kudos
Reply
  • 9
  • 0
  • 1
plums1234
Tuning in
535 Views
Message 2957 of 3,074
Flag for a moderator

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

I don't know how to post images on here so here is the imgur link for my screenshot DSLR PUMA 6 test as suggested earlier how to post images

PUMA 6 Test: https://imgur.com/a/cTP9i

Here is the imgur link to screenshot of my BQM, its obvious when the update took place: https://imgur.com/a/eHAGv

Hope it works this time

0 Kudos
Reply
  • 34
  • 1
  • 2
philjohn
Tuning in
487 Views
Message 2958 of 3,074
Flag for a moderator

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

The thing is, rendering pipeline for a video game is fixed - everything is setup and then rendered - there's no stopping the rendering of a given frame halfway through - it waits until it's done, and then when the packets have come in stuff will jump around, so the tearing is an unrelated issue.

At 60fps each frame takes around 16ms to render.

0 Kudos
Reply
  • 87
  • 2
  • 82
Datalink
Up to speed
456 Views
Message 2959 of 3,074
Flag for a moderator

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

What happens when the packet(s) are delayed by 200 to 300 ms?  In theory you have a string of packets arriving one after the other, with some packets potentially lost as they timed out in the modem itself. I suppose as you indicated "stuff will jump around".  I can't see anyway around that.  

0 Kudos
Reply
  • 58
  • 0
  • 9
yabba
On our wavelength
356 Views
Message 2960 of 3,074
Flag for a moderator

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


@wrote:

@ yabba

" but if your games or web browsing is a laggy mess or you're losing a ton of packets, are warping around in games etc, or if you suck and get killed a lot then that's really nothing at all to do with this issue."

Go back and read what I actually wrote before pontificating.  If you don't think this is the place to discuss it, then don't respond, but it seems to me you're still not grasping the practical impact of a 150-250ms lag spike on a client/server game, but you're now trying to close down the discussion and have the last word.

"I gave you a link, you have much to read and learn about lag compensation in games - this thread is not really the place to discuss it."

The only reason it has become a discussion point is because you and a couple of others are denying there's any effect of the Puma 6 lag spiking.  If that's the case, why have they spent a year trying to work out a fix?  Why have so many people been complaining bitterly, and some even taken their business elsewhere?


 

It's not a question of having any last word. You clearly don't understand lag compensation at all and I gave the word here to the developers of a game engine. Carry on in your ignorance or read their pages - it's up to you. This is a matter of fact though, not opinion. You can't debate it.

As for complaints. People have complained in here bitterly about gaming issues since the community was created and they'll continue to complain long after this puma 6 issue has gone away. Complaints are what you should expect to see here. They are not evidence of anything. There will be plenty of issues causing lag and packet loss in games for people other than intel's flakey chipsets.

IME there's barely any noticable difference at all in well written game engines, for example. Not the least because multiplayer gaming wouldn't work at all if the odd packet got dropped or arrived late. And I've seen no evidence at all on my connection that these packets happen every 2 seconds. It's more like every 30 seconds here. Pings to game servers are low and I've played the same as ever.

Indeed, I only know about this issue because I read about it - and then when I upgraded to V6 and they sent a hub 3 I was expecting a high ping in games and lots of issues but nope, nothing. I can run a few of the tests you are all fetishing and get graphs with yellow lines on them, or 5 or 6 red stars at the bottom but, aside from that, there's no visible evidence at all that this has affected copious amounts of web browsing, streaming netflix and youtube and playing Team fortress 2.

It's moot, there's an updated firmware now. Enjoy it.