Ping latency

[david 921]

I'm curious what other people are getting for latency when they ping their wireless access point with their Pro1 phones.  I'm generally getting 20+ ms.  My old Relay 4G is getting what I would expect, which is 1-2 ms.  A Pixel 1 I tested got about 4-5 ms (I think it is running android 10).

I'm curious if it is android that has gotten more stingy over the years (trying to save power) or if their is something about the Pro1 that is holding back, or a combination of the two.  But either way, I'd be interested in what other people get.  I'm doing this through terminal programs and getting the same results with multiple apps.  Also, this is a strong wifi connection, so that's not the issue.  It is very consistent, which makes me think it is purposefully being throttled by the OS or a driver.

[_DW_ 628]

43 minutes ago, david said:

I'm curious what other people are getting for latency when they ping their wireless access point with their Pro1 phones.  I'm generally getting 20+ ms.  My old Relay 4G is getting what I would expect, which is 1-2 ms.  A Pixel 1 I tested got about 4-5 ms (I think it is running android 10).

I'm curious if it is android that has gotten more stingy over the years (trying to save power) or if their is something about the Pro1 that is holding back, or a combination of the two.  But either way, I'd be interested in what other people get.  I'm doing this through terminal programs and getting the same results with multiple apps.  Also, this is a strong wifi connection, so that's not the issue.  It is very consistent, which makes me think it is purposefully being throttled by the OS or a driver.

There is a power saving mode to it in newer versions remember reading about it.  Not sure if you can alter it's settings through root or not.

[Erik 1,487]

I suspect it's the server that your device is picking up to do the test with, unless if you do the test from your home network somehow. Maybe check if the battery manager is on?

On our work wifi, we have a relatively impressive ping result. 

 

[david 921]

5 hours ago, Erik said:

I suspect it's the server that your device is picking up to do the test with, unless if you do the test from your home network somehow. Maybe check if the battery manager is on?

On our work wifi, we have a relatively impressive ping result. 

 

Was that test done over wifi or cellular?

I am talking about over wifi, by running the ping command, in a terminal shell, pinging the address of the wireless access point 6 feet away.  

I checked that "battery saver" was off. I also went into battery optimization and turned it off for the terminal app, but that didn't make a difference.

If you got 2ms ping over wifi, to a remote speedtest server, then i would question the results.  You would have at least one hop to get to the wireless access point and then one or more hops to get to your internet provider's gateway to the internet, etc.  Even over cellular, that seems too good to be true.  The speedtest server must be ultra close to you.

I would be curious what you can get by running the ping command to that same server, if you are able to find the ip address.  But running ping to your local wireless access point would replicate what I am doing.

[Erik 1,487]

11 minutes ago, david said:

Was that test done over wifi or cellular?

I am talking about over wifi, by running the ping command, in a terminal shell, pinging the address of the wireless access point 6 feet away.  

I checked that "battery saver" was off. I also went into battery optimization and turned it off for the terminal app, but that didn't make a difference.

If you got 2ms ping over wifi, to a remote speedtest server, then i would question the results.  You would have at least one hop to get to the wireless access point and then one or more hops to get to your internet provider's gateway to the internet, etc.  Even over cellular, that seems too good to be true.  The speedtest server must be ultra close to you.

I would be curious what you can get by running the ping command to that same server, if you are able to find the ip address.  But running ping to your local wireless access point would replicate what I am doing.

It was done over wifi in a WeWork office, which admittedly sometimes we get 800 MBPS up/down. The router is an enterprise grade Cisco Meraki. But I get similar sub 5ms pings with most fibre optic ISPs, in and around the London area.

[Hook 2,945]

If it's of any use, here is what I get oinging the WiFI router gateway next to my desk <--I'm not an IT guy and am not very technical and may or may not be using the terminology correctly.  😉

 

[3zet 100]

10 minutes ago, Hook said:

next to my desk

That does not mean you will have good signal. Read about wave propagation.

[Hook 2,945]

1 minute ago, 3zet said:

That does not mean you will have good signal. Read about wave propagation.

I'm not sure of your point since I wasn't making a claim of either good or bad signal.  IRL, everything works so I'm happy.  I was just providing David with ping data from another Pro 1, for whatever it might be worth to him.

[EskeRahn 5,427]

My results in termux are similar to what @Hook sees, note that the eleventh is down to 2.38ms though

[EskeRahn 5,427]

....BUT I see much better ping numbers in speedtest

 

[david 921]

Yes, this test is for wifi, with the ping command, in a terminal shell, to its access point.  Speedtest must be measuring in some weird way, although fiber optic definitely has lower latency.  Are you on fiber optic, @EskeRahn?

To your point @EskeRahn, yes, it sometimes reports faster pings.  That is another reason I am thinking something is holding it back the other 99% of the time.

[EskeRahn 5,427]

14 minutes ago, david said:

Yes, this test is for wifi, with the ping command, in a terminal shell, to its access point.  Speedtest must be measuring in some weird way, although fiber optic definitely has lower latency.  Are you on fiber optic, @EskeRahn?

To your point @EskeRahn, yes, it sometimes reports faster pings.  That is another reason I am thinking something is holding it back the other 99% of the time.

Yup, fiber, but above I'm pinging the WiFi access point on the inside. So the fiber is not in play, pure Wifi.

I think it is the other way round, Termux emulation or communiction with the phone adds a delay.

From within firefox Fast.com show similar results as speedtest prg

 

[_DW_ 628]

I get 22ms from my pro1 to my wifi router was from sailfish (needed SU wtf lol) and that was about 2 meters away

PC gets 3ms wired

Laptop gets 4ms wifi (and that was upstairs did the phone test near it as it got over a second upstairs)

 

[EskeRahn 5,427]

Just for compare, if I let Termux ping the Pro1 itself, it gets fast results, so it is not a delay from Termux reaching out to the local interface.

[david 921]

I did the same test with localhost.  It is something within the network stack that is slowing down when it is accessing outside the phone.  I even set my old phone as a hotspot and got the same results.  However, if I set the Pro1 as a hotspot and connect to it from my old phone, I get *under* 1 ms pings, consistently.  

I downloaded a couple ping apps and they get the same results pinging the wireless access point.  

For those of you who are getting fast pings reported in speed test apps, who are on fiber, can you try pinging google.com or some other known endpoint on the internet?  Preferably close to you.  When I do that, I get about 20+ more ms on the Pro1 than on my old phone, due to this extra time to get to the router over wifi.  

[EskeRahn 5,427]

20 minutes ago, david said:

 

For those of you who are getting fast pings reported in speed test apps, who are on fiber, can you try pinging google.com or some other known endpoint on the internet?  Preferably close to you.  When I do that, I get about 20+ more ms on the Pro1 than on my old phone, due to this extra time to get to the router over wifi.  

Please read above, Pinging the ACCESS POINT from Termux gives the same about 20ms results

So it does not matter what the connection-type is on the other side of the access point, it could only add to it.

[david 921]

17 minutes ago, EskeRahn said:

Please read above, Pinging the ACCESS POINT from Termux gives the same about 20ms results

So it does not matter what the connection-type is on the other side of the access point, it could only add to it.

Well, in theory, yes, but it could be something in the software that is looking at LAN addresses and adding latency for some unknown reason.  That was the only reason for asking for that test.

I did another test.  I plugged the phone into my laptop and enabled USB Tethering.  I then pinged the phone from the laptop and got 1-5 ms times.  I pinged the laptop from the phone and got the same.  So the network stack by itself is not wholly doing this.  And the terminal application is not doing this.  The common variables seem to be wifi to an address outside the phone.  I have tried 2.4 and 5.8 ghz wifi and both exhibit the same latency issue.

If anyone else wants to test USB Tethering, you have to go into USB Preferences and pick "USB tethering" once you are connected to the computer.  The easiest way to find the IP address of both sides is to either:

A)  Go into the network adapter in Windows and right click on the RNDIS adapter and pick "status" and click the "Details..." button.  This will give you the ipv4 address of something like 192.168.42.x.  It will also give you the phone's ip address on this tethered network as the IPv4 Default Gateway, which again will be a 192.168.42.x address, I'm guessing.  42 is the subnet it has been consistently using for me anyway.

NOTE:  To get to the network adapter screen, in Windows 7, go to the start menu, type in "network connections".  Pick "View Network Connections" from the list.  You will get something like this:

When you right click on that and pick Status or double click on it, you will get this:

After clicking on Details..., you will get this:

or

B)  Go into a command window and type "ipconfig /all".  Find the RNDIS entry and it will have the same information as option (A) above.

Edited January 30, 2020 by david

[EskeRahn 5,427]

6 minutes ago, david said:

I'm guessing.  42 is the subnet it has been consistently using for me anyway.

But 42 is the answer to ..... 🤣

[david 921]

7 minutes ago, EskeRahn said:

But 42 is the answer to ..... 🤣

:-)  It certainly is the answer to lowering ping latency from the Pro1! LOL

[david 921]

More testing...

Just in case it was only a problem with ICMP (the protocol ping uses), I found a way to test TCP latency.  By using the app PingTools Network Utilities, and changing the Ping feature's settings to use TCP, I get the same results -- 20+ ms typical latency (TCP pinging port 80 on my router).  It also has an option to test with HTTP, which gives slightly higher latency.

Testing over the USB tethered network to my laptop gives the normal, low latency pings again (using the TCP test in PingTools).

The main reason why all of this is important is because if there is added latency, web pages will load slower than they otherwise could.  Unlike a file download, where 20 ms won't matter, a web page with 100 objects (the average, modern web page has over 100 objects to download, each though a separate request) will take 2+ seconds longer to load with that added 20+ ms latency on top.  If the delay to the router is only 5 ms, then the added delay is only half a second. 

To be clear, this is *added* delay, on top of the latency introduced by the hops to the internet server in question.  So if the latency from the router to the website is 30 ms, then a web page could take a total of 5 seconds to load with 20 ms to from the phone to the router, instead of 3.5 seconds with 5 ms from the phone to the router.  It isn't the end of the world, but if this can be fixed, it will just make web pages feel more snappy on the Pro1.

Oh, and for those people who are into interactive games, the latency may have an effect on how the game perceives their response time.

 

[_DW_ 628]

Something I didn't point out above properly.

My laptop in same location as pro1 can get 4ms ping times but the pro1 struggles and gets ping rates all over some over 1 second this is opposite side of house through walls and upstairs. 

When near the wifi router it gets about 22ms consistent.

So this could be an antenna power issue.

Edited January 30, 2020 by _DW_

[david 921]

This may or may not be related to this issue:

https://source.android.com/devices/tech/connect/wifi-low-latency

[david 921]

55 minutes ago, _DW_ said:

Something I didn't point out above properly.

My laptop in same location as pro1 can get 4ms ping times but the pro1 struggles and gets ping rates all over some over 1 second this is opposite side of house through walls and upstairs. 

When near the wifi router it gets about 22ms consistent.

So this could be an antenna power issue.

Yes, I would say there are two issues.  I still get the 20+ms pings directly by the router, with full wifi signal.  I have noticed, and other people have mentioned, that the Pro1's wifi range isn't as great as other phones or devices.  It seems like that is a separate issue.  It is normal to get higher pings and lower data transfer rates when a wifi connection is weak.

I think it would be good to use a separate topic for the Wifi signal strength issue, where people can report their testing and results.  I've created one here:

If you could report your testing and results in that topic, @_DW_, that would be great.  I haven't done any thorough testing of that, but I did notice it possibly being lower.

[SchattengestaIt 559]

Just for more data a test with a different phone: Motorola G3

Direct router test: phone ~9ms, laptop 1-2ms

Google.com: phone ~29ms, laptop ~20-21ms

No optic fiber but a copper line... (max 2.77Mbit/s, 350kB/s)

 

According this data termux probably needs more time to process ping requests (which would fit with the results of other apps).

Maybe a videochat over the home network with front cameras facing each other (or mirrors). This would show how fast we get shown new pictures, but also depends on the apps you use and you'd also need a comparison to understand the results.

When do you really need good pings on your home network?

Edited January 31, 2020 by SchattengestaIt

[david 921]

54 minutes ago, SchattengestaIt said:

Just for more data a test with a different phone: Motorola G3

Direct router test: phone ~9ms, laptop 1-2ms

Google.com: phone ~29ms, laptop ~20-21ms

No optic fiber but a copper line... (max 2.77Mbit/s, 350kB/s)

 

According this data termux probably needs more time to process ping requests (which would fit with the results of other apps).

What I get out of your test is that your G3 has 9 ms of latency between it and your router.  But what we are talking about is that the Pro1 has 20+ ms of latency between it and its wireless gateway/router.  This happens whether it is in a terminal app being run as a command or in a graphic android app that implements Ping.

If you are saying that possibly the Pro1 isn't giving enough time to the application for doing  network processing (or to the drivers), then that could certainly be a possible cause.  The only problem with that is that @EskeRahn and I have tested pinging localhost from the phones and we get very fast pings doing that.  I would think if it was the application itself being throttled/starved by the OS, then we wouldn't see that behavior. It makes me think it is something in the drivers, but that only affects the external IP interfaces, specifically wifi.  I was also able to get low pings over a USB tethered network connection.

Do you have a Pro1?  If so, try that to your router and you should see a larger latency than your G3.

Quote

When do you really need good pings on your home network?

You don't.  You need them when you are accessing websites that have 100 objects on a page.  Your G3 will add approximately 0.9 seconds to the page load time (above the latency on the internet, depending on how far away the server is and how many hops it goes through).  The Pro1 will add approximately 2+ seconds to the page load time.  If the page load time would be 3 seconds with an ethernet connection to your router, then with the G3 it will be about 3.9 seconds vs 5+ seconds with the Pro1.  It is just a matter of pages loading faster if they can fix the source of the latency issue.

Some people play games that are highly latency dependent with other people online.  Latency can affect game play. 

You also brought up a good use case, which is video/audio chats.  Those are latency dependent.  But those protocols are such that they will handle this type of lag without too much trouble.  If it gets above 150 ms total (from end to end), then that could start to be a problem.  This 20+ ms latency shouldn't cause any issues with video chatting.

The other reason for looking into something like this is that more than one issue might be found/solved if something is found to be configured in a sub-optimal way.  There have been some other wifi issues reported (other than latency), so it is a good idea for fxtec to look into them (which they are).

To be clear, this issue isn't being brought up to say that the Pro1 is a bad phone because it has this latency issue.  It is being brought up so that they can make a very good phone even better.  Compared to some of the other issues, this one is fairly minor, by itself.

I don't have my cellular enabled yet on the phone.  If this affects latency over cellular too, then that would be another reason to look into it.  Latency in 4G isn't all that great, so if the phone were to add onto that, it could cause problems.  One of the benefits of 5G is very low latency (in theory), to improve upon this issue.  But since we'll all be using 4G for a while, all we can do is make it work as well as it can on the phone side.

Edited January 31, 2020 by david