Home
Artists
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register
Partychan Telegram ThePornDude
Home Artists Posts Import Register

Maybe It's Time To Review Starlink (Patreon)

Published:
2022-09-24 09:35:58
Imported:
2022-09

Content

Hi everyone, Rich here. Just finished my breakfast and I've been looking back over a massively stressful week. While RTX 4090 hardware arrived with time to spare, the crucial DLSS 3-enabled software only arrived a week ago, so it was crunch-time ever since. This project also posed big challenges to us - like how to capture 4K at 120fps, without which the content would not have been possible at all! All of this and more explains why early access was a little thin on the ground this week, so apologies for that. 

However, I did also finally get an upgrade to FTTP internet this week (900 down, 110 up) and at some point soon, the Starlink dish on the roof of my house will be taken down. Let's just say I won't be missing the 100W baseload it added to power consumption in my home! What this does mean though is that in my office right now, I have both Starlink and gigabit internet available, side by side - and they could be compared.

I was thinking of two PS5s hooked up to Atomos 4K recorders, both consoles playing the same multiplayer games. And then both PS5s could be observing the actions of a third PS5 elsewhere. When that third player shoots, for example, we could measure the latency difference between the two PS5s at my place. The two Atomos recorders put up timecodes as they record, so a high speed camera shot could be used to synchronise those feeds. I only wish there would be a way to synchronise the third feed located elsewhere (if the two local recordings start an MP game absolutely at the same time, we could use that as a sync point for the third, I suppose!)

I guess my question is this: what kind of tests do you think I should do? My only limitation on requests is that I think they should be PS5-based as it's the only machine where I have two of them - and I do think identical client hardware should be used. Are there any other lag tests beyond what I've described above that I should be considering?

Files

Comments

ShwaMiller

https://www.google.com/amp/s/arstechnica.com/tech-policy/2022/09/ookla-starlinks-median-us-download-speed-fell-nearly-30mbps-in-q2-2022/%3famp=1 It's getting interesting.

Anonymous

Hey Rich, I live in a fully off-grid home in the hills of the Australian rainforest, where Starlink is my only option for high speed internet. It's decent enough for my needs (uncapped and fast enough downloads of massive single player games mostly), I would much rather have cheaper and faster wired Internet. Why do _you_ have Starlink? ?

Anonymous

I like the test you outlined but I think an more comprehensive streaming services review/comparison would be excellent content. Understanding in the best of conditions what visual quality and response time you will get with various controllers wired/wireless/WiFi (Google stadia) From there going down in internet speeds and latency. Also the massive cost if we would do game streaming over mobile internet due to the massive volume of data even if you were in 5G with a moderate level of latency.

Anonymous

While latency tests would be interesting, I wonder how well they translate to other locations.. seems like things like starlink almost have to be evaluated by each user at their particular location

Anonymous

Using PS5's is a good idea for a more controlled environment. You may as well test how persistent are the download speeds are then. Either use a PC and download from a server, or download the same game to each PS5 and see if Starlink slows down at any point. You will have to download to the same SSD (model) to keep it further controlled though, and reduce bottlenecks. Nevertheless, the more data, the better. Others have already suggested testing cloud gaming in those environments too. Perhaps GeForce Now with Reflex enabled, to add onto further to Tom's prior coverage? Unless that is considered unnecessary. Hopefully others will contribute better suggestions

Anonymous

Maybe get the timestamp of the controller press on a PS5 then take the timestamp on the 3rd PS5 (observer in other location) when it sees the action take place and then use those timestamps as an estimation of the latency?

Anonymous

Share Play - Is there anything in this feature worth looking at, network-wise?

Anonymous

🤔 Scratching my head as to why your bandwidth is async instead of sync when it's FTTP. Because the bandwidth is almost Docsis 3.0 cable like. Is the backend based on Fibre or no? Because that would decide what kind of latency should you see. In terms of testing between Starlink and Your new connection, it's tricky because in your test case it may not be p2p (depending on game). You might actually be testing latency from one ps5 to the hosted game server back to the other ps5. There could be difference of hops between the two how they'll reach the hosted server which will decide which connection is better. Ideal testing would be with a game that allows P2P multiplayer. It will still depend on the hops but at least they would be trying to communicate with each other directly rather than the hosted server. Another test could be to use GeForceNOW or xCloud and doing multiplayer. The hosting should occur either locally or from a single source for both players. And then you will be testing streaming on Starlink VS FTTP.

Anonymous
>>92257753

We have been over this, it is just how internet is sold in the UK, higher upload speeds are locked to business connections at much greater prices under the theory that 99.9% of residential users just don't need the extra upload speeds and lets face it, that theory is correct.

Anonymous

I can't say I've ever played an online game and thought, wow I wish I had gigabit upload speeds to make this better, because as long as the ping is reasonable there is very little benefit to the additional bandwidth for gaming.

Anonymous
>>92257753

Hmm not necessarily in the world of iCloud, Google Photos, Amazon Photos and 365. People are making 4K videos on their phones which needs to get backed up in the cloud. People are working from home copying data to mapped drives/One Drive etc. Regardless it's not about whether they need it or not. It's about artificially restricting it. So you're telling me this is actually pure Fibre back to PHub but being restricted? In that case the latency should still be far better than Starlink then. My main concern was whether was just Fibre to the last mile with Coax backend.

Anonymous
>>92258741

Oh yes there's no advantage to additional bandwidth for MP gaming. (those use cases are different as I said above) But there's a significant difference in latency and jitter if it's pure Fibre backend FTTP VS Coax backed FTTP. It's similar to having good framerate but having bad frame pacing over the network. Hence my original question/confusion.

Anonymous

In theory, Starlink could use the full 1c signal speed of light over long distances, instead of the 0.7c of fiber or 0.5c of electric wire. So lag could be less for connection to a server on a different continent. Local servers would not have that benefit. A game that lets you select matchmaking region could be used to test if that signal speed difference holds up in practice for gaming. Fortnite has that option, but I'm not sure if that's the case on all platforms: https://www.epicgames.com/help/en-US/fortnite-c5719335176219/battle-royale-c5719350646299/how-do-i-change-my-matchmaking-region-in-fortnite-a5720300391579

digitalfoundry
>>92251543

Because up until this month, infrastructure had not been updated since 2010 meaning that the highest download speed I had was 32Mbps. Which is far from disastrous but I routinely had circa 200Mbps with Starlink. So do I wait an entire working day to download a game or do I wait a couple of hours? I was also hoping for a boost to upload speeds of 8Mbps but for the most part, it was actually slower.

digitalfoundry
>>92251877

Right, but there are certain fundamental challenges Starlink has to contend with that will be common across the entire network. Your base latency is 20-30ms higher, for example.

Anonymous
>>92251877

Sure that's possible, i guess that would also be an interesting thing to test, how consistent it is across locations :) i just mean it would be a shame to make a super detailed test, only to find the result be lost in the noise location variance... still though it would be interesting to see how it fares, maybe its just enough to make a note of saying everybody will probably see slightly diffrent results, atleast to avoid some negative comments

digitalfoundry
>>92258741

The difference vs Starlink is stark. In fact, I can tell the difference between Starlink and standard fibre internet in online games.

digitalfoundry
>>92260017

That is an interesting theory! I'm not sure it'll be born out in real world testing based on my general experiences mind you.

Anonymous
>>92258741

Do you think it's the additional bandwidth or significantly the better ping?

digitalfoundry
>>92255311

Yeah, but you'd need to be able to synchronise video capture of the remote PS5 with the local ones! I believe that MP games do start at exactly the same time, but the only way to check this would be by checking that the two local ones also start at the same time. Synchronising the feeds is key.

digitalfoundry
>>92250657

This is an interesting point because I've had Starlink for about 15 months now and the service - while still fast - has indeed degraded. I have felt it most in upload times.

Anonymous
>>92260017

But it really depends on how they are routing right? If they are only using Starlink satellite to route to the nearest CO or PHUB and it's basically acting as a replacement for last mile then it won't help. Generally for long distance theory to work the assumption would be that the satellite is actually directly pushing the packet down to another continent's ground station. It would be amazing if they do that!

Anonymous
>>92258741

For MP games, it definitely would be the consistency and reliability of packet transfer over Fibre. Wireless packet drops and jitter is brutal even if the ping is similar. Although I assume ping times would be different as well.

ShwaMiller
>>92250657

Yeah, it seems fairly widespread. Looking forward to the coverage!

Anonymous
>>92260017

It's possibly they only make that feature available to private partners (like high-frequency trading), or it's not active yet. In either case, gaming would often favor a ground connection. But if it worked, it could really cut down the worst case ping.

Anonymous

The PS5 test you mentioned would be interesting as a real world example, however any detailed review of Starlink would need months long data of the connection e.g. https://www.thinkbroadband.com/broadband/monitoring/quality or https://support.aa.net.uk/CQM_Graphs_-_Gamers for example, what are peak times etc, as it could be mostly useable for gaming in office hours but in the evening be completely different.

Anonymous
>>92251543

Yeowch! Uploads of videos must have been painful even with Starlink. I didn't imagine there were still places in the UK with that poor an infrastructure. Enjoy the efficiencies to your work flow 😊 On topic, how about comparing consistency of game download speeds? I don't have a PlayStation to compare against, but I notice my Xbox download speeds are relatively low and highly inconsistent, and rarely reach the heights of Steam downloads (which do a much better job at maxing out Starlink's bandwidth and are also much more consistent). Does FTTP provide a much more consistent connection?

Anonymous
>>92250657

I was one of the first ones in my area to get Starlink (a day before major flooding knocked out traditional network infrastructure!), downloads routinely hit 200-300 Mbps. 6 months later, 100-200 Mbps is more common. Not unusable by any means, definitely a measurable degradation. Still miles better than anything else available to me! Uploads and latency have been pretty consistent at 10-15Mbps / 50ms

Anonymous
>>92260017

I'm pretty sure Starlink still connects you to your local groundstation, which then uses traditional wired infrastructure to connect you to the rest of the Internet. I don't think the inter-satelitte laser transmitters are in widespread operation yet, if at all. That might indeed be a gamechanger!

ShwaMiller
>>92250657

Interesting. They must have very loose definitions of what their service provides in their contracts?

Anonymous

You could also use a network testing tool like PingPlotter to compare latency, jitter, and packet loss. Latency is important, but consistent latency (low jitter) is just as important for online gaming. I suspect StarLink is going to have higher jitter just due to the nature of it being satellite-based. It also depends on the destination of the packet. Your FTTx provider could have better inter-connectivity to data centers and Internet Exchanges which keeps latency and jitter low and can help avoid paths on the Internet backbone that are susceptible to congestion. I don't know enough about StarLink's backbone but once that packet hits the satellite, it has to get to the ground somewhere and that will rely on fiber and electronics to transport the traffic to its ultimate destination. And if StarLink has less optimal path connectivity and diversity on its network, than that only makes things worse. So yes, Fiber is a more reliable medium but there are other factors. Would be a fun analysis to see, especially to traceroute to the same destination on both networks to compare!

Anonymous
>>92257753

In the states, some ISP's use the regulatory bodies to determine their speed packages, and since the government is always about 10 years behind technology, you end up with situations like this where the FCC still thinks in terms of DSL internet, which is an asymmetric medium. But the ISP's use the guidance because of the funding and regulatory implications. It has gotten much better in the last couple of years to where you mostly see symmetrical speeds, at least for the higher end packages. I wonder if its a similar thing in Europe.