Ping Time doubles or halves depending on direction avatar is facing....Why?

[dd Temin]

Lately I have noticed that my indicated Ping time in statistics window varies wildly depending on the direction  my avatar is facing.  I do understand  that FPS  does change significently depending on the direction the avatar is facing. This is normal as it is actually depending  on how many different textures the avatar is seeing in that direction.  For  example facing the "ocean" will  give the highest  FPS readings as there are  very few textures to view in that direction   and conversly; facing a complex highly textured part of the region will indicate much lower FPS....this is expected behavior.

BUT  ,,,,why should PING TIME also be effected by the direction the avatar is facing???

Anyone have any explaination of this?

[Theresa Tennyson]

---

dd Temin wrote:

Lately I have noticed that my indicated

Ping time in statistics window varies wildly depending on the direction  my avatar is facing. 

I do understand  that FPS  does change significently depending on the direction the avatar is facing. This is normal as it is actually depending  on how many different textures the avatar is seeing in that direction.  For  example facing the "ocean" will  give the highest  FPS readings as there are  very few textures to view in that direction   and conversly; facing a complex highly textured part of the region will indicate much lower FPS....this is expected behavior.

BUT  ,,,,why should PING TIME also be effected by the direction the avatar is facing???

Anyone have any explaination of this?

---

A couple of ideas -

1) Are you in an area with multiple connected regions? You need to connect to every region your camera can see into. There may be more regions to look into in the "slow" direction.

ETA - I actually tested this. There will be a substantial spike in sim ping when your view causes you to look into an additional sim. The spike will generally settle down after a while. Also sim ping numbers are actually "snapshots" of the speed at a given time taken at about four-second intervals instead of a continuous figure.

2) Sim ping also includes the internal status of your computer and its connection to the internet. For instance, one user was having problems with extremely high sim ping that nobody else was seeing and it turned out that he was usually simultaneously running several open Firefox browser instances with multiple tabs open. When these Firefox connections were closed the ping returned to normal.

[Nalates Urriah]

You may also see a FPS hit when rotating your avatar, if your viewer is suddenly needing to move new textures into the video card. That happens when you spin the avatar around and see new stuff. The card starts unloading some textures, those behind you, and loading in others now in the field of view. There can be a lot of data to move. The result is decreased viewer performance. Use Ctrl-Shift-3 to watch the texture loading.

The PING in the viewer is not the same ping you get using the OS's ping. If you have both pings running to the same SL region server, you'll see the viewer ping grow/slow while the system ping remains resonably constant. If that happens it is not a connection issue as much as it is the viewer not being able to accurately measure ping while moving video textures around.

A similar thing can happen when the server is busy. 

The viewer ping is a higher level ping based  in viewer and server software. So, the software's ability to run affects ping time. With the OS ping it is a much lower level ping that uses timing from the network cards and thus is less dependent on software perfromance and more accurate for actual communication lag.

[dd Temin]

Thank you Nalates

I think this is the best explaination of what  I am observing. "the viewer not being able to accurately measure ping ...

The statistics window method of measuring "what we normally call Ping time"  is simply not the same kind of measurment as is normally done using a  Ping Time application/command. In a static  mode the two methods  may be  nearly the same.

[Andrew Linden]

Nalates is correct.  The way I would put is this:

The viewer measured ping time includes the time it takes for the viewer to get around to processing the network packets.  Although a packet may have already arrived at the ethernet port and is sitting in the queue, its "arrival time" is measured when it is actually processed, which is done in the same thread as the render work.

[Perrie Juran]

---

Andrew Linden wrote:

Nalates is correct.  The way I would put is this:

The viewer measured ping time includes the time it takes for the viewer to get around to processing the network packets.  Although a packet may have already arrived at the ethernet port and is sitting in the queue, its "arrival time" is measured when it is actually processed, which is done in the same thread as the render work.

---

It is really a shame that I went to update The Wiki with this information but that page is locked, as well as the Knowledge Base which needs updating also.  But at least in the KB we can (sometimes) add comments at the bottom of the page.

 I think I can state unequivocally that ping time is almost universally misunderstood.