Proton VPN created a feature that accelerates your VPN by 400%

[Whispre]

19 hours ago, Kisai said:

Again, this is literately common knowledge for PC gamers.

 

Funny, I've been a PC gamer since Windows 3.11, most of my friends and colleagues are also PC gamers... and "nagel" is not a word that has ever entered any conversation. I'd even be willing to bet that none of them have ever even heard of it.

 

I've been a Sysadmin AND gamer for over 25 years, and I'd have to side with leadeater on this one 100%, both on the nagle argument and the relationship between latency/throughput/bandwidth and buffering.

 

[nich99]

That sounds like paid ram downloading

[Kisai]

7 hours ago, Whispre said:

 

Funny, I've been a PC gamer since Windows 3.11, most of my friends and colleagues are also PC gamers... and "nagel" is not a word that has ever entered any conversation. I'd even be willing to bet that none of them have ever even heard of it.

 

 

 

 

This is literately a thing. "Leatrix fix" is nothing more than setting those three options you see TCP Optimizer. Regardless of the game, you will eventually find someone talking about reducing lag by disabling the nagle algorithm, even if the advise turns off the nagle algorithm and delayed ACK.

 

You can find references in google going back to Google's inception, but it really blew up with the launch of WoW.

 

Quote

I've been a Sysadmin AND gamer for over 25 years, and I'd have to side with leadeater on this one 100%, both on the nagle argument and the relationship between latency/throughput/bandwidth and buffering.

 

Then you haven't cared enough to test it. I have. These are from 2010.

 

 

Without the context from the game, the healing (among other skills) is the most obvious skill in the game that shows you when the nagle algorithm is turned off. You can't cast healing below 200ms with nagle in default settings. The blue-haired character is a second pc with nagle turned off, spamming the healing skill as fast as it can be clicked. So if you did not disable Nagle in this game, you would be left at a severe disadvantage in all PvP activities, and also in PvE events in competition with other players.

 

World of Warcraft (which I have not played) is similarly affected.

http://www.xtremesystems.org/forums/showthread.php?237964-Gaming-Tweak!!!!-Disable-Nagle-s-algorithm , and you'll see these "lag fixes" typically called "leatrix fix".

 

You'll see it again and again in optimization guides.

https://www.howtoguides.org/MMO-improve-latency-in-WoW-and-LoL-in-Windows

 

All I can say is that if you don't think this is a thing, then you aren't a part of the games community enough to be aware of it. 

 

Type "Nagle Algorithm" and an online game, and then try the same query with "leatrix fix" and you'll usually see the same thing over and over.

[Whispre]

16 hours ago, Kisai said:

All I can say is that if you don't think this is a thing, then you aren't a part of the games community enough to be aware of it. 

I never said it "wasn't a thing" just saying that it is not as common as you seem to believe. Also, not sure a comparison from 2010 is a good one given today's gig fiber environment.

 

Anyway, just don't equate you and your friends and some online people knowing about this with it being "common knowledge". If people have to google it to find out about it, it's not that common.

[Lurick]

34 minutes ago, Whispre said:

I never said it "wasn't a thing" just saying that it is not as common as you seem to believe. Also, not sure a comparison from 2010 is a good one given today's gig fiber environment.

 

Anyway, just don't equate you and your friends and some online people knowing about this with it being "common knowledge". If people have to google it to find out about it, it's not that common.

Not only that but all the results from their post show just about everything from 2010 to 2015 in almost every result indicating things have changed (shocking I know) since that time.

[leadeater]

7 hours ago, Lurick said:

Not only that but all the results from their post show just about everything from 2010 to 2015 in almost every result indicating things have changed (shocking I know) since that time.

I'd also be willing to bet 100% of the fix is TCP Delayed ACK off and zero to do with Nagle as a whole (and I still doubt it). Because nobody ever actually just does one setting at a time to evaluate what each one does and if all of them are required.

 

Neither have I ever heard of using VPNs to lower pings in games being popular, that literally defies all logic. You'd have to be on some seriously bad ISP/infrastructure configuration for that to be even remotely possible.

 

Also the second result in the posted Google search lol. Do a search yourself for Leatrix TCP Delayed ACK, first result. Very clear correction to what is going on there too, not TCP_NoDelay only TcpAckFrequency, well what do you know, I do know something 

 

This whole flat 200ms thing due to Nagle or TCP Delayed ACK just isn't true

Quote

TcpAckFrequency
Key: Tcpip\Parameters\Interfaces\interfaceGUID

Value Type: REG_DWORD—number

Valid Range: 0–255

Default: 2

Description: Specifies the number of ACKs that will be outstanding before the delayed ACK timer is ignored. Microsoft does not recommend changing this value from the default without careful study of the environment.

https://docs.microsoft.com/en-us/previous-versions/windows/it-pro/windows-server-2003/cc739819(v=ws.10)?redirectedfrom=MSDN

 

200ms is only used if no other condition is met and the established session has not used session flags to configure the session appropriately. Literally all you need is 2 delayed ACKs by default and the bundled packet will get sent immediately.

 

As a once MMO fiddler to see what is possible how do I know the posted video isn't just using a cheat to get around cast time restrictions, something that is very easily done even with anti cheat tools.

[Kisai]

5 hours ago, leadeater said:

 

 

As a once MMO fiddler to see what is possible how do I know the posted video isn't just using a cheat to get around cast time restrictions, something that is very easily done even with anti cheat tools.

Because BOTH of those machines are mine. Geez. Y'all so desperate to deny what people do out of desperation to stay competitive in MMO games, first it's "deny", then it's "oh they must be hacking"

 

I'm absolutely used to seeing people pull this nonsense to keep their competitive advantage. You'll note that all the sites that say not to tamper with windows settings say to use a VPN. The topic of this thread.

 

Using a VPN never helps games except to get around a geoblock. Kiss your TCP hardware offload goodbye. Changing the windows settings "as THE solution" is not a great solution, but some game developers (in particular, mobile games) still never implement the settings in the software itself, most of them being Korean and Japanese games that aren't used to the >10ms latency everyone in North America gets.

[RejZoR]

Buffer size on LAN adapter also matters. Have it too big and data will sit in buffer causing latency, have it too small and you'll have packet drops if they can't fit into buffer. If you get it just right it makes huge difference.

[leadeater]

7 hours ago, Kisai said:

Because BOTH of those machines are mine. Geez. Y'all so desperate to deny what people do out of desperation to stay competitive in MMO games, first it's "deny", then it's "oh they must be hacking"

Well no I'm saying it's really unlikely because changing TCP ACK settings isn't going to get around cast timers in an MMO. Unless what you are casting has no casting cooldown then yea it might work but the difference will absolutely not be as pronounced as shown, what you see and what is actually happening quite often are different things. You spamming cast is 100% going to cause more than 2 delayed ACKs in the buffer in well under 200ms.

 

Only thing I'm desperate about is ensuring misinformation isn't being perpetuated because it literally doesn't help anyone, in fact it makes it worse.

 

I did not say it was hacking I specifically said for all I know that's what it is. Unless you have hundreds of data points for each game for each setting on and off at different times with actual ways to check what is really happening all it is is you saying trust me it works when there is no technical backing for how it would actually do anything.

 

I don't know, maybe I expected you to read and understand the information being presented to better inform yourself. Some of the settings changes might help but best cases are when the latency is exceedingly high already where immediately sending ACKs might actually make a difference rather than every second packet.

 

But this whole "Applications are designed around 200ms", can we please stop that. It's categorically not true, it's not correct for how the TCP stack works in any OS or in the standards itself.

 

7 hours ago, Kisai said:

but some game developers (in particular, mobile games) still never implement the settings in the software itself, most of them being Korean and Japanese games that aren't used to the >10ms latency everyone in North America gets.

Even in Korea and Japan below 10ms is exceedingly good, probably a good few that can get that but I'd doubt less than 8ms. I've personally never had problems competing on ADLS2+ on servers all the way over in Australia on many many games and for MMO's I'd be lucky if there even is an Australia server.

[Kisai]

3 minutes ago, leadeater said:

Well no I'm saying it's really unlikely because changing TCP ACK settings isn't going to get around cast timers in an MMO. Unless what you are casting has no casting cooldown then yea it might work but the difference will absolutely not be as pronounced as shown, what you see and what is actually happening quite often are different things. You spamming cast is 100% going to cause more than 2 delayed ACKs in the buffer in well under 200ms.

Just so you know, during the entire time I played that game, they never fixed it. You can spam the 5 casts all immediately. On the old forums for that game I even had RTT graphs showing exactly what was going on. 

 

Nagle on, the game would get buffered. Nagle off the game buffered nothing, and anything that wasn't explicitly server controlled could be fired off as fast as you could press it. The script kiddies on that game went so far as to hack the packets, and it would send players flying all the way across the map. This was not a very well secured game at the best of times, and I quit playing it during what was known as the "black robe" incident where the hackers figured out how to destroy targeted characters from the client end.

 

 

3 minutes ago, leadeater said:

Only thing I'm desperate about is ensuring misinformation isn't being perpetuated because it literally doesn't help anyone, in fact it makes it worse.

And I'm not saying you're wrong, I'm saying this is what is happening, and the proof is in the pudding.

 

3 minutes ago, leadeater said:

I did not say it was hacking I specifically said for all I know that's what it is. Unless you have hundreds of data points for each game for each setting on and off at different times with actual ways to check what is really happening all it is is you saying trust me it works when there is no technical backing for how it would actually do anything.

I could do better at the time. I had THREE PC's available to run the game on simultaneously (aka multi-boxing). I only used those to explicitly proof points to people like you who don't believe things unless they see it. The other two PC's were not powerful enough to capture video, so they were literately run at 640x480x15fps. I'm sure I could look through my archive of stuff I did record and find examples of people spamming healing, crafting, and going out of bounds.

 

I've always regarded the nagle/delayed ACK tampering as a symptom of naive game development.

 

3 minutes ago, leadeater said:

I don't know, maybe I expected you to read and understand the information being presented to better inform yourself. Some of the settings changes might help but best cases are when the latency is exceedingly high already where immediately sending ACKs might actually make a difference rather than every second packet.

The game (server) reacts immediately when it receives the command, that's why buffering or delaying is detrimental to the game experience.

 

Quite honestly, this problem shows up EVERY TIME a Japanese or Korean produced game is localized. The Japanese or Korean game works "just fine" over there amazingly. Yet as soon as it's brought over to the US, the 10x worse latency all over the continent makes the game anywhere from barely playable to unplayable. I've seen this with FFXIV, I've seen it with Archeage, I've seen it with a handful of other games that are long dead like Wizardry Online, it just keeps happening, and the lessons never get learned. Wizardry was almost worse than Mabinogi for hackers destroying the game from the client end. 

 

3 minutes ago, leadeater said:

But this whole "Applications are designed around 200ms", can we please stop that. It's categorically not true, it's not correct for how the TCP stack works in any OS or in the standards itself.

Look, that information came directly out of something I read 10+ years ago.

 

https://docs.microsoft.com/en-US/troubleshoot/windows/win32/data-segment-tcp-winsock

 

3 minutes ago, leadeater said:

Even in Korea and Japan below 10ms is exceedingly good, probably a good few that can get that but I'd doubt less than 8ms. I've personally never had problems competing on ADLS2+ on servers all the way over in Australia on many many games and for MMO's I'd be lucky if there even is an Australia server.

Yeah, well the Australians/SEA players I played with on the game with the servers in NA all had horrific latency that invoking this fix at least let them spam the skills and hope they land. So there's no way for me to know what your experience is like, and vice versa.

[soldier_ph]

On 9/6/2021 at 4:10 PM, linux fanboy said:

bring New Zealand closer to Europe

But but.. New Zealand doesn't exist !

 

 

/s aside

 

 

 

I use ProtonVPN and haven't noticed any differences yet. I mean the connection quality and experience is the same whether or not the VPN is turned on which is why it's just always turned on even on my Home WiFi.

[leadeater]

33 minutes ago, soldier_ph said:

But but.. New Zealand doesn't exist !

Nah that's Australia, but I guess New Zealand is just a state of Australia anyway so we therefore do not exist as you say.

 

P.S. No need to come looking to find out, we're fine with everyone thinking we don't exist lol.

[leadeater]

On 9/11/2021 at 1:27 AM, Kisai said:

Look, that information came directly out of something I read 10+ years ago.

You realize you are using an example of bad configuration and programming and applying it as if it's globally true? I honestly don't know what to tell you if you use these edge cases that are documented and given so programmers don't make the same mistake and act like it's universally the case.

 

As per also Microsoft documentation and literally how the TCP stack works by default in Windows, soon as two delayed ACKs are pending it is sent. You have to make specific mistakes in your code, as documented in your link, to have fixed 200ms issue. Literally read the recommendation section, you actually have to change the defaults to hit these problems.

 

So may I repeat again, 200ms delay is not a thing. Applications are not designed around this because it's not a thing. You have to specifically mess up to actually have the problem and it's easily fixed, if they care to fix it.