Re: internet latency
Posted: 06 Jun 2019, 15:55
but why does it not show up as loss then or ping fluctuations to 400 ms bc i can clearly feel the difference in my game when i get better hit reg but my ping doesnt change
Who you gonna call? The Blur Busters! For Everything Better Than 60Hz™
https://forums.blurbusters.com/
it can be within the packet data and client software. or the first google search of "udp ping" offers a tool. tools are built into other software as wellnick4567 wrote:because of the way udp packets work i dont think it would show anything wrong i dont really think theres a way to test this
You still missed the point. It has nothing to do with what is being measured and reported by the default system tools/the game itself in the terms of ping and packet loss. Obviously, something is happening to the packets, delay, loss or both, it is just not being reflected properly by the standard measuring tools. My speculation is, packets are being delayed and delivered late and out of order, which creates all the problems that people are experiencing. In terms of packet loss, there are no missing sounds, just a feeling of firing an empty bullets at the enemies. Typical packet loss feels and manifests itself differently, and is also being properly measured and reported by the default system tools/the game itself when it actually occurs. And this is precisely why this whole thing creates the problem i described, people that are affected by this look at everything else but at their own internet connection as a potential culprit.open wrote:so yeah. everything you listed is a form of delay or loss. congratulations on agreeing with me
Many lurkers and readers do read these threads. Sometimes my replies are intended for readers of the thread rather than the original posters.Chief Blur Buster wrote:(This reply not addressed to anyone in particular)
Chief Blur Buster wrote:Disclaimer:
Many lurkers and readers do read these threads. Sometimes my replies are intended for readers of the thread rather than the original posters.Chief Blur Buster wrote:(This reply not addressed to anyone in particular)
Oftentimes, I've seen many outside parties cite/quote Blur Busters threads at other forums or reddit. Sometimes there comes a point where I just simply have to chime in (Because I saw misinformation on reddit from someone citing a Blur Busters thread). So sometimes while my replies sometimes add noise inside a thread, they can reduce noise outside the thread.
Not everything has measuring tools easily developed/available.
Just like for displays, there were often no measuring tools available for specific kinds of artifacts. Even for esports we need application developers to create new measuring apps that are available for download. Oftentimes I see some very complex apps for display testing that costs tens of thousands of dollars with a specialized piece of equipment such as "high speed colorimeter camera" or a "motorized MPRT measurement camera" --
But laypeople can't just download a tool (or use an existing diagnostic already installed in OS or website) to figure out that a display problem is being caused by something that these measurement devices can measure. Same for networking too as well.
There are lots of networking problems that definitely exists that don't have easy measuring tools available for them. One example is Cisco has a lot of high end expensive ISP network analysis stuff that actually measures some really interesting data -- since the backbone people need these expensive tools to properly measure their big iron. A lot of that data doesn't ever get seen by you nor me, unfortunately. No user-downloadable EXE can tell you. Unfortunately I do not know if this is the case here or not, but I definitely keep an open mind (Blur Busters policy). The fact is, sometimes it's not easy to measure without a big spend...
Likewise, NVIDIA and some other manufacturers have created apps that can measure stuff that the bloggers haven't yet. (I do bring out display testing inventions that makes it possible for testing at much lower budgets than was possible ten years ago). But not everything is cheaply/easily measurable.
So many things come out of left field for displays. And for networks too.
Even a weird situation of slightly varying the packet size or temporal spacing between the packets, unexpectedly triggers unmeasured latency. All that innovation in traffic shaping, great-firewall-style-analysis and security scanning (random latencies for suspicious packets) used in many countries, auto-throttling algorithms (like LTE carriers use for congested nodes), and automatic packet reprioritizations (unintentional or intentional; ala distortions to network neutrality), creates all kinds of new undiscovered latency issues in recent months and years that our tools don't always cover.
It's becoming worse and worse, even if average latencies are falling -- as the latency noisefloor falls lower, the smaller latency noise is appearing that we formerly didn't care to measure. 16ms VSYNC ON console refresh cycle makes 1ms ping jitter invisible, but 4ms 240Hz VSYNC OFF makes 1ms-worsenings of ping jitter sufficiently visible enough with certain games that don't have algorithms good enough to filter that out. Certain things are continually creating unexpected secret ping jitter in games that don't show in your measuring tool. Etc. Etc. Etc. Specific game creating network pattern, or the combinations that occur (e.g. gaming+streaming occuring simultaneously) that, even with a perfect gaming router, is triggering something deep in the network missed by a common measuring tool. That ISPs have done has created some unexpected side effects that rarely show up. Measuring tool A catches something that measuring tool B doesn't. Likewise, the gaming stuff can interact with complex network algorithms that keeps getting continually slapped hand-over-fist onto the Internet. Throw in the IPV6 soup and and it's becoming horrendously complex for measuring-tool-makers.
The existence of 240Hz revealing latency flaws and motion flaws more visibly than 60Hz, even where the measuring tooldesigners haven't considered -- even 1ms moves a tearline down 25% screenheight at 240Hz -- and 1ms frametime delay creates 1 pixelwidth of stutter per 1000 pixels/sec motion. For the most sensitive individuals, even 1ms network delays aren't always fully successfully filtered out in the soup of algorithms that games have, and some are much more sensitive to tiny problems than others. The game is much to blame, but as the noisefloor falls everywhere (lower lag, higher Hz, faster GPU, etc), the noisefloor of other network latency problems starts to appear that few or nobody is currently bothering to measure. Many are display pros or network pros but few are simultaneously display-pros-AND-network-pros.
There are many unmeasured interactions. In the very simplistic example -- an average may be steady but a median may change -- or vice versa -- and create a whole different lagfeel that is unmeasured. And even an average/median may stay mostly unchanged but the volatility pattern changed (different latency inconsistency feel) and not all network lag tools may make this show up like a christmas tree until it's properly scatterplotted in a brand new chart visualization a different tool did not have. Yet one-or-the-other may be creating more problem for lag artifacts for a specific individual missed in measurements. And complex interactions with a video game's lag compensation algorithm can create better/worse outcomes for a specific volatility-pattern graph than a different game.
And on the measuring front, users may hate doing certain things much like they hate English class or might hate Math class at school. It's a bit frustrating when someone doesn't want to troubleshoot/measure in the same way we do, or that our point is being missed.
The point is that the network-lag is horrendously complex underneath the surface, far moreso than every poster here thinks. It's no wonder some of us don't even bother touching specific Pandora's Boxes! The software running in carrier routers is orders of magnitude more more complex than it used to be 20 years ago, creating new complex latency patterns that are harder and harder to measure sometimes. Just as we think something is a solved problem, extra complications layer in, and we're back to square one. Now, even what I say, is probably not what is happening, but the point is: It's not always fun measuring or trying to understand/interpret measuring processes, and it's complex enough to easily be misunderstood, even if it is something far simpler than the Pandora's Box underneath the network-lag surface.
Nontheless, we keep an open mind on hard-to-measure stuff.
It is not my intention to derail discussion, so apologies -- my generally-on-topic reply is also addressed to lurkers/readers of this thread too who haven't posted anything. This is often my Blur Busters Forum approach when I see potentially contentious topics.
(Carry on, feel free to pretend I didn't reply to this thread.)
Cheers.