Page 3 of 3 FirstFirst 1 2 3
Results 21 to 23 of 23

Thread: The official RATE thread

  1. Default

    Quote Originally Posted by ManBearPig <ibis> View Post
    He puts out on the first date, i have first hand experience.
    What a whore.

  2. Default

    As some of you may recall one of the lesser known rate changes I always recommended was cl_interp .02 when on a 100tick server. This is because valve said that the ideal value should be 2/cl_updaterate. However the cl_interp cvar apparently is no longer used and is replaced by cl_interp_ratio this value is the number from which cl_updaterate is divided by in order to achieve the interp. However, the default value is 1! So in other words your defult interp is off by 100% from its ideal value in poor conditions. These poor conditions occur when you receive choke. Setting the value to 1 is best for a client with a perfect connection and fps that never goes below cl_updaterate. Therefore anyone playing on any zombie server that wants good hit detection should aim for a setting of 2 while a pub player wants a setting of 1. Ideally players should change this setting based off current server choke conditions however they may wish to use a setting of 1.5 to achieve good middle ground.

    For those that wonder what this does here is the easiest explanation/example of how it works in the real world:
    As we know in source a moving player will see a non moving player first. Assuming all things such as ping ect constant, the non moving player will see them with a delay of cl_interp. Therefore the lower the interp the less the effect of netcode related lag and thus better hit registration. Ideally you would want interp at 0 however this is impossible to achieve. While having a low interp is good for canceling out the problem with netcode it is bad for accurately predicting what is going on. A low interp assumes that you have a high quality connection with the server. Ideally an interp of 0 could be achieved if the server and client were one in the same and false data was never propagated within the system. With the internet or even on lan the effects of packet loss or choke mean that the client must predict what it does not know. The interp setting tells the client how long of a period it needs data to predict for. Therefore at a setting of 1 for cl_interp_ratio it assumes that the client is able to get a cmd packet every time the server generates one assuming that the server tick=cl_cmdrate. If the server generates less ticks than expected or if your client receives less than it was supposed to you will receive choke relative to this amount. As a result your computer will not accurately display where a player really is on the server. If you used a cl_interp_ratio of 2 you allow yourself to lose 50% of the cmd packets and still generate an image that is equivalent to having no loss and a cl_interp_ratio of 1. However the image will be rendered on your screen in 50% the time of if cl_interp_ratio was 1.

    So you can have a low interp that gives you less data to predict where clients are or you can have a high interp that gives you more data to predict where they are. You only need the prediction to compensate for inconsistency between the data on the server and the data on your client.

    For those that want raw data with no explanation:
    An interp of .02 =20ms delay before rendered on your screen
    An interp of .01=10ms
    and .1=100ms

    100tick server produces a cmd packet every 10ms because it calculates player location 100 times per second.
    If you miss a CMD packet at 100tick with cl_updaterate 101 and cl_cmdrate 101 and cl_interp_ratio 1:
    There should be a 50% chance of incorrect data
    If cl_interp_ratio was 2 then:
    There should be a 25% chance of incorrect data
    *Those predictions are based only on the way that 1 packet should effect the clients ability to predict.

    The cost of the 100% increase in consistency by using 2 instead of 1 is a 100% increase in the delay of the image, so in the settings given above a setting of 2 vs 1 means that the player with 2 sees the image 10ms after the player using setting 1 except that their image is twice as accurate.



  3. Default

    I was under the impression that in CSS the default interp value was .02?

    I know in 1.6 default is .01 though. I remember in the old CAL days it was illegal to change the interp value, though. Something to do with the cal.cfg you had to execute.

    Anyway, good info, Zero!


    4) Use admin privileges sparingly and appropriately.


Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •