Could you share more detail on how you went about gathering the data used in the graphs? I took a packet capture on the next hop after the PS4. This packet capture was then analyzed with a hacky perl script which just picked out the parts of the TCP header I wanted, and aggregated them by-connection at a 10s granularity. I doubt the code going to be useful for anyone, but I do like it as an illustration of just how simple this kind of ad hoc analysis can be even without explicit tool support.
That's a good guess! Some sort of new failure of exactly that code is what I was hoping to find. But unfortunately it's not the case. All of the connections were using TCP timestamps, but the autoscaling was not in effect at all despite that. They're clearly manually setting the receive buffer size with setsockopt , and that disables the autoscaling. I didn't check whether non-PSN connections had autoscaling enabled or not.
Thanks for this very interesting blog post, BTW.. Have been annoyed by this particular set of issues for a very long time. Strangely, I never thought that the software could be such an egregiously bad actor in this case. I think a more egregious issue is why the PS4 is so much slower over WiFi than over Ethernet - would it really just be congestion at the AP?
But I think that ought to be demonstrably false when I can download at high speeds on a laptop that said, I'm most likely not using any of the same sources as the PS4. I honestly think that the set of devices released by Sony around the same time as the PS4 suffer similar problems - Currently, our TV cannot "find internet" over our WiFi, and that's when it manages to find and connect to our WiFi in the first place! If they wanted to limit the rate of downloads it would make a lot more sense to set the allowed rate to e.
Any thoughts on why Wifi speeds are just so slow? Is it a similar issue where the software caps speeds of downloads if the system sees it's on a Wifi connection? You're on the right track, except that it needs to go a step further. The allowed rate should not be static, like 1Mbps. It should actually depend on the network conditions.
The core goal of making sure background downloads don't interfere with games or streaming is reasonable. It's just that it's implemented in the wrong way, and triggered even when not necessary. If I understand correctly, I would not be able to notice these issues at all with my 5 Mbps adsl.
Welcome to Germany. Seems like a "feature" that only surfaced with the advent of high Speed internet. Interesting to have some figures on this matter although I thought everyone was aware of the fact that apps being open at all limited the download speed. That said I have apps set to be suspended in rest mode but have measured how fast the same download took Both with ps4 on and in game and in rest mode with game suspended and even accounting for changes in Internet bandwidth that occurred in that time period the download when in rest mode was almost twice as fast.
Something not explained by changes in Internet bandwidth. Out of interest I don't think you said we're you using normal Ps4 or ps4 pro? As the pro in theory has faster clocks peed on cpu which might effect how games using cpu effect any download limiting in those cases it also has the newer WiFi chipset I believe. Thanks for that, community. We need people like you in the company, rather than just business folk! It appears that the ARM coprocessor is not actually used for downloads, even though it was the original plan.
Sony stated this officially soon after the PS4 launch, and nothing I've seen suggests that things have changed since then. This was on an non-slim PS4. Yes, there could be a difference between "game running" and "game suspended in rest mode"; specifically it does look like actively playing a game can cause the download to genuinely starve for CPU.
You can find the section where this is discussed by searching for "Horizon". Excellent article thanks. Explains why pointing at my synology nas proxy sorted out the performance for me. I just wonder if receive window affects ps now applications as well. Sony would shoot themself in the foot if it does since ps4 allow to play ps now game while normal game is idling in the background.
Can this be easily verified I wonder? It'll be easy to verify by taking a packet capture. PSN is due to collapse. Its not built for the future. Cant even change your name or region on it. Only a matter of time, and they thought that hack a few years ago was bad. Interesting article! My friends and I do not only have the slow download problem, but something even worse: The "preparing download" takes over minutes times, often even longer than the download itself especially when loading updates for installed games.
Did this issue ever occur to any of you guys? I really wonder if there is a fix for this. Interesting Article. However, there must be other reasons for slow speeds. Otherwise, pausing and resuming a download should not help, which it often does. Sometimes dramatically so. Impressive analysis, thank you for sharing.
I noticed on your graphs you were testing in the late morning, did you see a variability throughout the day? My reason for asking is that in the US, netflix would often be the worst from to , I'm curious if you matched that pattern.
That might indicate an overall network load correlation. The PS4 has a special chip that can download files while the rest of the PS4 sleeps, right? Probably some integration with that that's causing it.. This was covered in an earlier comment. Downloads being handled by a coprocoessor was the original plan, but that's not what Sony implemented.
Even in rest mode, it's done by the main CPU. My belief is that they've done this intentionally to prevent performance issues from downloading quickly, probably with a big issue being disk access from downloading so fast, as well as affecting latency for online games and CPU usage. While using a local proxy is a good workaround the PS4 at least the original model can be almost unusable while downloading especially multiple downloads , causing both the games and OS to stutter and the controller to become unresponsive, taking 3 or 4 seconds to react to a button press.
Sure, it could be that this is what they were going for. The static limit on the receive window means that your setup still ends up unusable due to the downloads being too fast, while other people have their downloads unnecessarily throttled to a speed far below the effective disk speed.
Without it, downloads would take hours, with it, minutes. Is it possible the latest system software 5. Can't say for sure without actually redoing the testing. Download speeds can vary for all kinds of reasons; the only way to be certain are systematic tests.
But my feeling is that they haven't fixed it. At least I had an awkwardly slow download last week, after having forgotten to close a game. But it could also have been just a coincidence. It all depends on your Internet connection. I live in the middle of nowhere and get 1Mbps. I can just about stream SD video as long as nothing else is using the connection, and so I applaud anything Sony are doing to throttle downloads when anything else is using the Internet connection.
The problem here is that the Youtube or Netflix app is on is an awful way to detect whether somebody else is using the connection. First, it has false positives due to including the cases where the app is actually active or not. If Youtube is suspended in the background, there's absolutely no way. Second, it has a ton of false negatives due to only looking at what's happening on that PS4 system. That's an absurd assumption: in the modern world a household will have half a dozen devices using the same internet connection.
The current solution does not help at all for the case where I'm trying to watch a Youtube clip on my computer while the PS4 downloads something, or someone is trying to read a webpage on their phone, etc. Like I wrote in the post, there are proper ways of doing this by looking at the actual network traffic patterns.
The "what apps are kind of running" approach chosen for the PS4 is just totally broken. I don't know if it's just me but my Playstation 4 has had absolutely terrible internet speeds. Nothing I've done has worked, and this has been a problem since the latest firmware update. Had zero problems, downloads started immediately and were consistent until they were finished. On the Playstation 4 the speeds fluctuate all over the place, sometimes it changes dramatically when I pause and resume.
Then it slows down more and more until it becomes unbearable. Then there's the questionable sign in problem along with PSN overall being medicore in the past few months. So how does an old, 12 year old console from the past generation give me better speeds and better consistency than a modern console that is supposed to be good? It's very frustrating.
I'm this close to buying a new Slim PS4 model just to see if the download speeds and performance are any better. Could please Test the slim version? I would like too see what has actually changed or improved on the Slim model. I am having very slow downloads speed on my PS4 that has wired connection to the router and nothing running in the background. Why have we gone from buying decent full games that download fast on ps3 to broken half a games and really slow downloading.
I am going off gaming completely just because if I want to buy and download a game I have to expect to play it the next day after I originally wanted to play it and then I just go off wanting to play the game. Was trying to download expansion packs for modern warfare was going to take 16 hours for 16GB install. Closed all my apps running I forgot to close including Modern Warfare and Netflix As an antispam measure, you need to write a super-secret password below. Today's password is "xyzzy" without the quotes.
Why PS4 downloads are so slow. Posted on in Networking , Games. That means Sony has a big job to keep the PlayStation Store status healthy at all times. Let's say you're trying to download a patch for Call of Duty: Modern Warfare and your PS4 says there's not enough room, even though there should be.
The reason you can't download and install the. Still, nothing at all. Just the little yellow light on my ps4 controller to show that it's charging. I tried using an xbox one controller as well to see if that was the problem, but my computer didn't detect that either. See here for an article how to do this for some popular routers. This is another very common reason why PS4 downloads can go slower than expected — there are simply a lot of people using the internet at the same time in the home and more generally , and this can increase network congestion and increase the time it takes for downloads to complete.
This problem is worse on Wi-Fi, because it is more prone to congestion. A router can only handle traffic demands on Wi-Fi sequentially, in a queue based system, and not simultaneously. More devices connecting equals more traffic demands, equals longer downloads, as the router can only do so much at once and has to keep everyone happy.
However, there is a setting called Quality of Service or QoS, which you can try and use on your router to manage traffic and prioritize your PS4 on the home network. This is a setting which can tell your router to handle all traffic demands from your PS4 first, before anything else.
It is mainly used to reduce ping on busy networks; nevertheless you can try it if you like to see if it also improves download speeds as well at busy peak times. Nevertheless, it is good practice to use QoS to reduce ping at least on busy networks, and may offer some help with download speeds. More generally, it is also true that downloads can run slower more generally at peak internet use time evenings and weekends , when lots of people are indoors and using the internet.
One thing that can commonly catch people out is not realizing the difference between download speeds and file sizes. This is something that caught me out when I was expecting a patch download to be completed in seconds when in fact it took many minutes.
Put simply, download speeds from your ISP are measured in megabits per seconds, whereas file sizes are measured in megabytes or gigabytes, a much larger unit of data. A bit is eight times smaller than byte, and therefore a megabit is eight times smaller than a megabyte, and a thousand times smaller still than a gigabyte. There are megabits in a Gigabyte so this gives some idea of why large downloads can take so long.
At a download speed of 50 megabits not megabytes per second we can expect to download megabits or just over a third of a gigabyte per minute. If you have a large game of 20 gigabytes or more to download you can now see why it will take so long. A file that large will take around 1 hour to download at this speed. Online gamer and general home networking enthusiast.
I like to create articles to help people solve common home networking problems. There's lot of talk of IP addresses nowadays, and it's even relatively easy to find the IP address of your device using online tools or through network settings. But are IP addresses actually unique Skip to content. Tweet Pin Share. Faster Settings, Best Speeds!
0コメント