Re: [hlds] Re: Please Explain Latency Measurements in Counter-Strike Source

[Whisper]

--
[ Picked text/plain from multipart/alternative ]
Huh?
 This is a gmail account
 All F1 does in this window is bring up IE Help

 On 7/23/05, James Tucker <[EMAIL PROTECTED]> wrote:
>
> I have generally found that my choke in net_channels shows a
> consistent 0.0 on my chosen rates during these periouds.
>
> Changing to default rates at this time will always produce choke on
> the outbound and on the inbound channels.
>
>
> P.S.
> Want a gmail account, and I'd recommend not using digests - use
> outlooks sorting and archiving fucntions. :)
>
> F1 is a suprisingly useful key, and I'm honestly not trying to be
> rude, it really is!
>
>
>
>
> On 7/21/05, Dabosman <[EMAIL PROTECTED]> wrote:
> > James,
> >
> > I got ya. Well, all I can say maybe is: glitch? And like you said .. If
> > it's infrequent .. And doesn't affect gameplay .. That's about all I
> could
> > chalk it up to .. Unless some firewall software or router is causing
> > incorrect readings .. And I kind of doubt that as well.
> >
> > Any rate, if nothing else, hopefully the 'tickrate' info will help you.
> And
> > with this new update just released, maybe if you used the tickrate 66 (I
> > tried 100 and got worse results) in combo with the 'srcdsfpsboost.exe'
> > proggie, your server would fy. :)
> >
> > P.S. Do you know what a pain it is to edit these 'digest' messages and
> put
> > the quotes in myself? But I guess that's better than getting 30 emails a
> > day from it. And yes, I could setup a 'rule' in Outlook to put them all
> in
> > a folder .. But boy would they add up fast. I guess I'll live with the
> > 'digest' mode. I'd REALLY like if they'd just put this list on a message
> > forum at some point?
> >
> > Kevin
> >
> > James Tucker <[EMAIL PROTECTED]> wrote:
> > >[thank you, but -snip-]
> > >
> > >rate 20; echo rate set to 2mbps
> > >cl_updaterate 60; echo requesting 60 packets per second from the
> server.
> > >cl_cmdrate 60; echo sending 60 packets per second to the server.
> > >
> > >I have seen people running around with cl_cmdrate 0/1/generally under
> 10
> > and this does result in a falsely returned ping on the
> > >scoreboard (or used to, haven't checked today). This however is not
> what I
> > describe. I have seen a ping of 6-8 consistently on
> > >the scoreboard with the above settings and whilst it is not frequent,
> it is
> > impossible, and therefore must be somehow incorrect. >The oddity is also
> > independant of my rate settings, or nearly so, on default rates at times
> > when I am experiencing this, my ping >will simply sit from 5-6 and not
> > deviate for minutes at a time.
> >
> > ___
> > To unsubscribe, edit your list preferences, or view the list archives,
> please visit:
> > http://list.valvesoftware.com/mailman/listinfo/hlds
> >
>
> ___
> To unsubscribe, edit your list preferences, or view the list archives,
> please visit:
> http://list.valvesoftware.com/mailman/listinfo/hlds
>
--

___
To unsubscribe, edit your list preferences, or view the list archives, please 
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds

Re: [hlds] Re: Please Explain Latency Measurements in Counter-Strike Source - also

[James Tucker]

Depending on your client side FPS you may not see a difference between
sv_maxupdaterate 60 and 66. There is a difference however, you can
recieve a packet every tick (you dont need more generally) if you set
sv_maxupdaterate 66, and cl_updaterate 66.

I run, rented on a dual xeon located in Redbus londen:

fps_max 1000
sv_unlag 1
// this is time in seconds, our autokicker kicks at 100ms, so we
shouldn't need more than this
// we leave 20 ms for grace.
sv_maxunlag 0.12
// apparently this provides a reduction in bandwidth useage.
// i believe it's the default anyway, can't remember, but it's 'normal'.
sv_instancebaselines 1
// source uses less that 10 bytes per second, however, map
downloads use more.
sv_maxrate 10
// this minrate is high, but we play competitively so we will probably
drop 56k'ers
// on the ping kicker with this minrate. a reasonable minrate for 56k
support is 4500.
sv_minrate 1
// logic would dictate that non-tick updates are not useful for
accurate game data
// as the gameworld only changes on a tick.
sv_maxupdaterate 66
// as above, we're not interested in 56k support- so this is reasonable.
// this allows cl_interp 0.05 whilst still providing a good (spare)
interpolation time.
sv_minupdaterate 30


And it achieves:

14:20:23 stats
14:20:23 CPU   InOut   Uptime  Users   FPSPlayers
 20.23 5294.75 8822.13 34011  511.17   3
14:20:23 stats
14:20:23 CPU   InOut   Uptime  Users   FPSPlayers
 20.23 5374.05 8851.87 34011  524.95   3
14:20:23 stats
14:20:23 CPU   InOut   Uptime  Users   FPSPlayers
 20.23 5374.05 8925.50 34011  527.04   3
14:20:23 stats
14:20:23 CPU   InOut   Uptime  Users   FPSPlayers
 21.54 5501.40 8952.79 34011  508.02   3
14:20:24 stats
14:20:24 CPU   InOut   Uptime  Users   FPSPlayers
 21.54 5501.40 8955.15 34011  514.33   3

We've been really happy with these settings.

On 7/21/05, Dabosman <[EMAIL PROTECTED]> wrote:
> James,
>
> Probably should have mentioned this before .. But the settings I'm using on
> my server are close to in-line with your '> 18 players' settings in your
> thread you wrote on Steampowered - which I will say is well written and has
> good info for the admin wanting to tweak their server.  The only thing is,
> since my server is a P4 1.8Ghz w/512mb of ram (dedicated with nothing else
> running on it), I decided to set my fps_max to the default .. And I also
> keep my sv_maxrate lower .. Because in my opinion, sv_maxrate 2 is more
> for LAN and really detrimental to server play (lag) when you get around
> 14-16 players on over the Internet.  And yes, the full T1 connection could
> handle it.  :)  Also I actually seemed to notice worse results when upping
> the fps_max when a lot of players got on - although if you have a 3Ghz
> Dual-Processor server, that wouldn't matter I'm sure.  I just believe it's
> not necessary and very helpful to server play from my testing and playing
> with it.
>
> I use the following on my 17 player (16 player and 1 reserve) server:
> -tickrate 66
> sv_maxrate 1
> sv_maxupdaterate 60 (I believe I tried '66' with minimal effects)
> fps_max 300
>
> You might try these and see how your server responds.  Works great on my
> server.
>
> Kevin
>
> ___
> To unsubscribe, edit your list preferences, or view the list archives, please 
> visit:
> http://list.valvesoftware.com/mailman/listinfo/hlds
>

___
To unsubscribe, edit your list preferences, or view the list archives, please 
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds

Re: [hlds] Re: Please Explain Latency Measurements in Counter-Strike Source

[James Tucker]

I have generally found that my choke in net_channels shows a
consistent 0.0 on my chosen rates during these periouds.

Changing to default rates at this time will always produce choke on
the outbound and on the inbound channels.


P.S.
Want a gmail account, and I'd recommend not using digests - use
outlooks sorting and archiving fucntions. :)

F1 is a suprisingly useful key, and I'm honestly not trying to be
rude, it really is!




On 7/21/05, Dabosman <[EMAIL PROTECTED]> wrote:
> James,
>
> I got ya.  Well, all I can say maybe is:  glitch?  And like you said .. If
> it's infrequent .. And doesn't affect gameplay .. That's about all I could
> chalk it up to .. Unless some firewall software or router is causing
> incorrect readings .. And I kind of doubt that as well.
>
> Any rate, if nothing else, hopefully the 'tickrate' info will help you.  And
> with this new update just released, maybe if you used the tickrate 66 (I
> tried 100 and got worse results) in combo with the 'srcdsfpsboost.exe'
> proggie, your server would fy.  :)
>
> P.S.  Do you know what a pain it is to edit these 'digest' messages and put
> the quotes in myself?  But I guess that's better than getting 30 emails a
> day from it.  And yes, I could setup a 'rule' in Outlook to put them all in
> a folder .. But boy would they add up fast.  I guess I'll live with the
> 'digest' mode.  I'd REALLY like if they'd just put this list on a message
> forum at some point?
>
> Kevin
>
> James Tucker <[EMAIL PROTECTED]> wrote:
> >[thank you, but -snip-]
> >
> >rate 20; echo rate set to 2mbps
> >cl_updaterate 60; echo requesting 60 packets per second from the server.
> >cl_cmdrate 60; echo sending 60 packets per second to the server.
> >
> >I have seen people running around with cl_cmdrate 0/1/generally under 10
> and this does result in a falsely returned ping on the
> >scoreboard (or used to, haven't checked today). This however is not what I
> describe. I have seen a ping of 6-8 consistently on
> >the scoreboard with the above settings and whilst it is not frequent, it is
> impossible, and therefore must be somehow incorrect. >The oddity is also
> independant of my rate settings, or nearly so, on default rates at times
> when I am experiencing this, my ping >will simply sit from 5-6 and not
> deviate for minutes at a time.
>
> ___
> To unsubscribe, edit your list preferences, or view the list archives, please 
> visit:
> http://list.valvesoftware.com/mailman/listinfo/hlds
>

___
To unsubscribe, edit your list preferences, or view the list archives, please 
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds

Re: [hlds] Re: Please Explain Latency Measurements in Counter-Strike Source

[Whisper]

--
[ Picked text/plain from multipart/alternative ]
I have changed the server to 66 tickrate and there is very little change to
pings, either on scoreboard or net_graph, and the server is also
pingboosted.
 That is to say, the discrepency still exists
 Cheers

 On 7/22/05, Whisper <[EMAIL PROTECTED]> wrote:
>
> Sorry Kevin I did not mean to come across as defensive, it just seemed you
> were putting a lot of effort in to try to explain things in the most basic
> of terms which causes you write a lot of unnecessary detail, thus I was
> trying to speed up the process a little.
>  I did appreciate the theory that a 66 tickrate may in fact lower the
> net_graph ping, and endevouring to prove or disprove this theory.
>  Cheers
>  On 7/22/05, Dabosman <[EMAIL PROTECTED]> wrote:
> >
> > James,
> >
> > I never questioned your technical expertise - nor did you mine I
> > believe.
> >
> > I'm just confirming what you saw on your scoreboard ... Is the same
> > thing
> > I've seen in-game myself. But I was also pointing out, which I do NOT
> > think
> > you realized, about the tickrate 66 ACTUALLY making the net_graph ping
> > stable in my testing - AND it (in my testing on MY connection and
> > server)
> > have kept that ping right in-line with the scoreboard ping. This, you
> > SPECIFICALLY asked about. And I will still provide proof for this but
> > I'm
> > sure (would hope at least) you believe me. I was not implying one way or
> >
> > the other that you didn't understand tickrate - but you did ask about it
> > in
> > a previous response which did not tell me either way whether you did or
> > didn't.
> >
> > It sounds like what you're wanting is an explanation from Valve as to
> > why
> > these measurements (especially console 'status' ping) do not match each
> > other - for which that I can not give you. I was only trying to confirm
> > what you'd seen as actual fact.
> >
> > But I will say you did mention in an earlier response that you could
> > hardly
> > believe your ping was that low in actuality .. But I say to you from
> > looking
> > at your tracert, it is. Unless your host would allow ICMP though on that
> > end, I could not tell you for 100% certaintity.
> >
> > Again, I obviously can't provide the reason as to why the console
> > 'status'
> > ping is out of whack .. But can provide you a way to keep your ping from
> > net_graph in-line with your scoreboard ping - if that is what you're
> > seeking. AND that would be assuming that this holds true on your server
> > as
> > well .. Since I've only noticed this in my testing. I actually disabled
> > everything (tickrate back to 33 and srcdsfpsboost off) ... And saw the
> > ping
> > on net_graph fluctuate right back up to around 70-90ms (when it's
> > normally
> > 50-60ms steady).
> >
> > It amazes me how people can get so defensive on this mailing list .. And
> > it's probably one reason I haven't bothered to respond to many questions
> > before.
> >
> > Hope this has helped .. As that's all I was trying to do here. If not,
> > good
> > luck in your quest.
> >
> > Cheers.
> >
> > Kevin
> >
> >
> > Whisper <[EMAIL PROTECTED]> wrote:
> > >
> > > Kevin, this
> > > might be useful in getting some idea at what level I am at,
> > > and where I am coming from when I ask what appear to be on
> > > the surface "stupid"
> > > questions
> > >
> > > http://forums.steampowered.com/forums/showthread.php?s=&thread
> > > id=293285
> > > I also have a fair bit of control over the server settings
> > > of the servers I'm taking screenshots of.
> > > I know how many SRCD server processes are on the box and the
> > > physical specs of the server (Its a beast) and how exactly
> > > each of them are configured.
> > > Investigation also led me to find out this particular server
> > > which was supposed to be ping boosted, was in fact not. Job
> > > logged for that thanks Andrew and Dave.
> > > I will have a look at your theory though about 66 tickrate,
> > > as the precise same server I just ramped back up to 66
> > > tickrate and 100 sv_maxupdaterate and 2 sv_maxrate now
> > > that Valve have release this latest patch.
> > > It is still my belief though that the 3 so called Latency
> > > measurments are still extremely out of whack no matter how
> > > the server is configured, amd I for one would like to know why.
> > > Cheers
> >
> >
> >
> > ___
> > To unsubscribe, edit your list preferences, or view the list archives,
> > please visit:
> > http://list.valvesoftware.com/mailman/listinfo/hlds
> >
>
>
--

___
To unsubscribe, edit your list preferences, or view the list archives, please 
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds

Re: [hlds] Re: Please Explain Latency Measurements in Counter-Strike Source - also

[Whisper]

--
[ Picked text/plain from multipart/alternative ]
sv_maxrate won't matter too much if you don't increase sv_maxupdateratre
 Also we found that on >18 player servers, hell most servers if you bump
fps_max up to 600 tickrate 66 & sv_maxupdaterate 100 that you get lots of
choke with sv_maxrate 1.
 You can see it quite easily when you play and measure your incoming
bandwidth on the client on a server setup with maxrate 2, you will get
up to 16KB/s incoming so long as there are enough players and enough action
going on.
 sv_maxrate also seems to be a large factor on CPU usage (I dont understand
why, it just does) turning it down really lightens the CPU load.

 On 7/22/05, Dabosman <[EMAIL PROTECTED]> wrote:
>
> James,
>
> Probably should have mentioned this before .. But the settings I'm using
> on
> my server are close to in-line with your '> 18 players' settings in your
> thread you wrote on Steampowered - which I will say is well written and
> has
> good info for the admin wanting to tweak their server. The only thing is,
> since my server is a P4 1.8Ghz w/512mb of ram (dedicated with nothing else
> running on it), I decided to set my fps_max to the default .. And I also
> keep my sv_maxrate lower .. Because in my opinion, sv_maxrate 2 is
> more
> for LAN and really detrimental to server play (lag) when you get around
> 14-16 players on over the Internet. And yes, the full T1 connection could
> handle it. :) Also I actually seemed to notice worse results when upping
> the fps_max when a lot of players got on - although if you have a 3Ghz
> Dual-Processor server, that wouldn't matter I'm sure. I just believe it's
> not necessary and very helpful to server play from my testing and playing
> with it.
>
> I use the following on my 17 player (16 player and 1 reserve) server:
> -tickrate 66
> sv_maxrate 1
> sv_maxupdaterate 60 (I believe I tried '66' with minimal effects)
> fps_max 300
>
> You might try these and see how your server responds. Works great on my
> server.
>
> Kevin
>
> ___
> To unsubscribe, edit your list preferences, or view the list archives,
> please visit:
> http://list.valvesoftware.com/mailman/listinfo/hlds
>
--

___
To unsubscribe, edit your list preferences, or view the list archives, please 
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds

Re: [hlds] Re: Please Explain Latency Measurements in Counter-Strike Source

[Whisper]

--
[ Picked text/plain from multipart/alternative ]
Sorry Kevin I did not mean to come across as defensive, it just seemed you
were putting a lot of effort in to try to explain things in the most basic
of terms which causes you write a lot of unnecessary detail, thus I was
trying to speed up the process a little.
 I did appreciate the theory that a 66 tickrate may in fact lower the
net_graph ping, and endevouring to prove or disprove this theory.
 Cheers
 On 7/22/05, Dabosman <[EMAIL PROTECTED]> wrote:
>
> James,
>
> I never questioned your technical expertise - nor did you mine I believe.
>
> I'm just confirming what you saw on your scoreboard ... Is the same thing
> I've seen in-game myself. But I was also pointing out, which I do NOT
> think
> you realized, about the tickrate 66 ACTUALLY making the net_graph ping
> stable in my testing - AND it (in my testing on MY connection and server)
> have kept that ping right in-line with the scoreboard ping. This, you
> SPECIFICALLY asked about. And I will still provide proof for this but I'm
> sure (would hope at least) you believe me. I was not implying one way or
> the other that you didn't understand tickrate - but you did ask about it
> in
> a previous response which did not tell me either way whether you did or
> didn't.
>
> It sounds like what you're wanting is an explanation from Valve as to why
> these measurements (especially console 'status' ping) do not match each
> other - for which that I can not give you. I was only trying to confirm
> what you'd seen as actual fact.
>
> But I will say you did mention in an earlier response that you could
> hardly
> believe your ping was that low in actuality .. But I say to you from
> looking
> at your tracert, it is. Unless your host would allow ICMP though on that
> end, I could not tell you for 100% certaintity.
>
> Again, I obviously can't provide the reason as to why the console 'status'
> ping is out of whack .. But can provide you a way to keep your ping from
> net_graph in-line with your scoreboard ping - if that is what you're
> seeking. AND that would be assuming that this holds true on your server as
> well .. Since I've only noticed this in my testing. I actually disabled
> everything (tickrate back to 33 and srcdsfpsboost off) ... And saw the
> ping
> on net_graph fluctuate right back up to around 70-90ms (when it's normally
> 50-60ms steady).
>
> It amazes me how people can get so defensive on this mailing list .. And
> it's probably one reason I haven't bothered to respond to many questions
> before.
>
> Hope this has helped .. As that's all I was trying to do here. If not,
> good
> luck in your quest.
>
> Cheers.
>
> Kevin
>
>
> Whisper <[EMAIL PROTECTED]> wrote:
> >
> > Kevin, this
> > might be useful in getting some idea at what level I am at,
> > and where I am coming from when I ask what appear to be on
> > the surface "stupid"
> > questions
> >
> > http://forums.steampowered.com/forums/showthread.php?s=&thread
> > id=293285
> > I also have a fair bit of control over the server settings
> > of the servers I'm taking screenshots of.
> > I know how many SRCD server processes are on the box and the
> > physical specs of the server (Its a beast) and how exactly
> > each of them are configured.
> > Investigation also led me to find out this particular server
> > which was supposed to be ping boosted, was in fact not. Job
> > logged for that thanks Andrew and Dave.
> > I will have a look at your theory though about 66 tickrate,
> > as the precise same server I just ramped back up to 66
> > tickrate and 100 sv_maxupdaterate and 2 sv_maxrate now
> > that Valve have release this latest patch.
> > It is still my belief though that the 3 so called Latency
> > measurments are still extremely out of whack no matter how
> > the server is configured, amd I for one would like to know why.
> > Cheers
>
>
>
> ___
> To unsubscribe, edit your list preferences, or view the list archives,
> please visit:
> http://list.valvesoftware.com/mailman/listinfo/hlds
>
--

___
To unsubscribe, edit your list preferences, or view the list archives, please 
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds

Re: [hlds] Re: Please Explain Latency Measurements in Counter-Strike Source

[Whisper]

--
[ Picked text/plain from multipart/alternative ]
Kevin, this might be useful in getting some idea at what level I am at, and
where I am coming from when I ask what appear to be on the surface "stupid"
questions
 http://forums.steampowered.com/forums/showthread.php?s=&threadid=293285
 I also have a fair bit of control over the server settings of the servers
I'm taking screenshots of.
 I know how many SRCD server processes are on the box and the physical specs
of the server (Its a beast) and how exactly each of them are configured.
 Investigation also led me to find out this particular server which was
supposed to be ping boosted, was in fact not. Job logged for that thanks
Andrew and Dave.
 I will have a look at your theory though about 66 tickrate, as the precise
same server I just ramped back up to 66 tickrate and 100 sv_maxupdaterate
and 2 sv_maxrate now that Valve have release this latest patch.
 It is still my belief though that the 3 so called Latency measurments are
still extremely out of whack no matter how the server is configured, amd I
for one would like to know why.
 Cheers
 On 7/22/05, Dabosman <[EMAIL PROTECTED]> wrote:
>
> Okay, here is what I see James: Your ping actually may be close to that
> good .. And remember .. From what I've seen in my testing .. The
> scoreboard
> ping will always show the best since it seems to 'average' the ping over
> so
> many samples (again .. 15-30 seconds maybe?). I notice your tracert is
> AROUND 12ms or so before it stops. Generally, if your server is only
> around
> 1-2 hops behind where it's blocked at (which if you saying it's blocked AT
> the server level so I'm sure it is) ... Then your ping probably IS *VERY*
> close to the 11-17ms you are seeing. Generally the hop RIGHT before your
> server is an in-building router .. That then splits off to a switch
> possibly
> that then goes directly to your server. See what I'm talking about with my
> tracert to my server below. Do you host this server or is it at a hosting
> site? If you could somehow enable ICMP, I'm sure you'd see a VERY close
> value to the 12ms that you saw on your graphical tracert. Just for
> reference, I keep anywhere from a 48-58ms ping on my server (depending on
> server load). It's always just a tad higher than your tracert ping .. But
> percentage-wise in your case .. I think yours is right on the money.
>
> 9 47 ms 48 ms 45 ms 69.151.16.250 
> 10 46 ms 45 ms 46 ms 65.69.123.132 
>
> Oh, and to explain that high 'console' ping? They have not been accurate
> at
> ALL in CS:S. I didn't realize you were asking about that as well. In 1.6,
> it was MUCH closer. And this goes along with readings from HLSW (admin
> tool) .. Because when I check the server with it with the 'STATUS' command
> through RCON in HLSW, everyones pings are well over 100 on my server ...
> And
> their ACTUAL in-game pings are usually well under 100 (50-90ms).
>
> And lastly .. Again on the tickrate 33 .. Since this is probably what your
> server is set on .. And I would HIGHLY bet on that if it's 'hosted' to
> save
> CPU cycles since all hosts 'share' boxes unless you pay a high 'dedicated'
> server fee ... It's very common as I said in the previous emails that your
> net_graph ping would be anywhere from 20-30ms higher than your actual
> scoreboard ping. This goes back to what I said about it 'averaging' the
> latency from it .. And I'm still not sure it's totally accurate on that.
> In
> other words, you're going to see the 'low' end of the actual ping most of
> the time on the scoreboard .. But the actual 'server' fluctuated ping on
> the
> net_graph. *IF* .. And big IF here since I'm thinking it's at a hosted
> facility .. You could run the 'srcdsfpsboost.exe' .. To increase the
> server
> frames to above the default 64fps .. *AND* you could run the 'tickrate 66'
> command-line option, I'd be willing to bet your net_graph ping would stay
> right in line with your scoreboard ping .. Assuming the server it's hosted
> on could keep up (I would hope so).
>
> For proof of this, if I get time .. I'll post pics of what I'm talking
> about
> for you. I'll show you my ping fluctuating in net_graph without either of
> my 'tweaks' (srcdsfpsboost and tickrate) ... And then show you WITH them.
> But my wife and I are going on vacation in a couple days .. So it'll
> probably be after the 1st.
>
> Hope this explains a lot. The pics definitely helped me narrow down on
> what
> you were seeing.
>
> Kevin
>
>
>
> Whisper <[EMAIL PROTECTED]> wrote:
> > Kevin here
> > are screenshots that illustrate why I have question about this issue.
> > http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg.jpg
> > http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0001.jpg
> > http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0002.jpg
> > http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0003.jpg
> > http://whisper.ausgamers.com/pictures/sourcepings/de

Re: [hlds] Re: Please Explain Latency Measurements in Counter-Strike Source

[Jesper Sørensen]

Now I might be totally of here, but as far as I can see the scoreboard
ping is the network RTT (packet to server + packet back from server),
and net net_graph is the real RTT (packet to server + server processing
+ packet from server).

So for a 0ms network RTT situation where the server is running at 33
ticks, this would give you a net_graph latency of around 16 (1000/33/2 + 0).
(1000 because that how many ms there are in a second, 33 because that
how many ticks there are pr. second, and 2 because we take the average -
we might come in with a packet just a before a tick is run, or we might
come is just after a tick is run, but on average we come in half way
between the ticks)

Now the client goes totally fubar if the cl_cmdrate / cl_updaterate is
low / high - And the numbers returned have little to no relation to
reality, this is (after looking at the source), due to the fact that
both cl_cmdrate and cl_updaterate are part of the euqations that govern
the behavior of the scoreboard + net_graph latencys.

Best regards

Jesper Soerensen

Whisper wrote:


--
[ Picked text/plain from multipart/alternative ]
Kevin here are screenshots that illustrate why I have question about this
issue.
http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg.jpg
http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0001.jpg
http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0002.jpg
http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0003.jpg
http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0004.jpg
http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0005.jpg
http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0006.jpg
http://whisper.ausgamers.com/pictures/sourcepings/serverping.jpg
If you check the creation times of the pictures you will see that they were
all taken within the same minute.

The Last picture is a http://www.pingplotter.com tracert ping tool set at 1
second intervals that was running whilst I joined the game and took the
screenshots. As you can see, the average ping does not deviate far from the
scoreboard ping. The last hop times out because the server firewalled from
ping packets.
net_graph and stats pings though, do not seem to bear any relation as to my
connections actual latency, and thus far, there has been no adequate
explanation as to why the 3 ping measurements are completely at odds with
each other.
Oh, the server in question is on a link that I can always download at
1MegaByte per Second sustained for 100's of MB's. That is to say, file
servers sitting in the same rack will serve data out to me at that rate.
Hopes this make sense.
Cheers
On 7/21/05, Dabosman <[EMAIL PROTECTED]> wrote:



James,

I got ya. Well, all I can say maybe is: glitch? And like you said .. If
it's infrequent .. And doesn't affect gameplay .. That's about all I could
chalk it up to .. Unless some firewall software or router is causing
incorrect readings .. And I kind of doubt that as well.

Any rate, if nothing else, hopefully the 'tickrate' info will help you.
And
with this new update just released, maybe if you used the tickrate 66 (I
tried 100 and got worse results) in combo with the 'srcdsfpsboost.exe'
proggie, your server would fy. :)

P.S. Do you know what a pain it is to edit these 'digest' messages and put
the quotes in myself? But I guess that's better than getting 30 emails a
day from it. And yes, I could setup a 'rule' in Outlook to put them all in
a folder .. But boy would they add up fast. I guess I'll live with the
'digest' mode. I'd REALLY like if they'd just put this list on a message
forum at some point?

Kevin

James Tucker <[EMAIL PROTECTED]> wrote:



[thank you, but -snip-]

rate 20; echo rate set to 2mbps
cl_updaterate 60; echo requesting 60 packets per second from the server.
cl_cmdrate 60; echo sending 60 packets per second to the server.

I have seen people running around with cl_cmdrate 0/1/generally under 10



and this does result in a falsely returned ping on the



scoreboard (or used to, haven't checked today). This however is not what



I
describe. I have seen a ping of 6-8 consistently on



the scoreboard with the above settings and whilst it is not frequent, it



is
impossible, and therefore must be somehow incorrect. >The oddity is also
independant of my rate settings, or nearly so, on default rates at times
when I am experiencing this, my ping >will simply sit from 5-6 and not
deviate for minutes at a time.

___
To unsubscribe, edit your list preferences, or view the list archives,
please visit:
http://list.valvesoftware.com/mailman/listinfo/hlds




--

___
To unsubscribe, edit your list preferences, or view the list archives, please 
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds

.






___
To unsubscribe, edit your list preferences, or view the list archives, please 
v

Re: [hlds] Re: Please Explain Latency Measurements in Counter-Strike Source

[Whisper]

--
[ Picked text/plain from multipart/alternative ]
 Kevin here are screenshots that illustrate why I have question about this
issue.
 http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg.jpg
http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0001.jpg
http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0002.jpg
http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0003.jpg
http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0004.jpg
http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0005.jpg
http://whisper.ausgamers.com/pictures/sourcepings/de_dust_pcg0006.jpg
http://whisper.ausgamers.com/pictures/sourcepings/serverping.jpg
 If you check the creation times of the pictures you will see that they were
all taken within the same minute.

The Last picture is a http://www.pingplotter.com tracert ping tool set at 1
second intervals that was running whilst I joined the game and took the
screenshots. As you can see, the average ping does not deviate far from the
scoreboard ping. The last hop times out because the server firewalled from
ping packets.
 net_graph and stats pings though, do not seem to bear any relation as to my
connections actual latency, and thus far, there has been no adequate
explanation as to why the 3 ping measurements are completely at odds with
each other.
 Oh, the server in question is on a link that I can always download at
1MegaByte per Second sustained for 100's of MB's. That is to say, file
servers sitting in the same rack will serve data out to me at that rate.
 Hopes this make sense.
 Cheers
On 7/21/05, Dabosman <[EMAIL PROTECTED]> wrote:
>
> James,
>
> I got ya. Well, all I can say maybe is: glitch? And like you said .. If
> it's infrequent .. And doesn't affect gameplay .. That's about all I could
> chalk it up to .. Unless some firewall software or router is causing
> incorrect readings .. And I kind of doubt that as well.
>
> Any rate, if nothing else, hopefully the 'tickrate' info will help you.
> And
> with this new update just released, maybe if you used the tickrate 66 (I
> tried 100 and got worse results) in combo with the 'srcdsfpsboost.exe'
> proggie, your server would fy. :)
>
> P.S. Do you know what a pain it is to edit these 'digest' messages and put
> the quotes in myself? But I guess that's better than getting 30 emails a
> day from it. And yes, I could setup a 'rule' in Outlook to put them all in
> a folder .. But boy would they add up fast. I guess I'll live with the
> 'digest' mode. I'd REALLY like if they'd just put this list on a message
> forum at some point?
>
> Kevin
>
> James Tucker <[EMAIL PROTECTED]> wrote:
> >[thank you, but -snip-]
> >
> >rate 20; echo rate set to 2mbps
> >cl_updaterate 60; echo requesting 60 packets per second from the server.
> >cl_cmdrate 60; echo sending 60 packets per second to the server.
> >
> >I have seen people running around with cl_cmdrate 0/1/generally under 10
> and this does result in a falsely returned ping on the
> >scoreboard (or used to, haven't checked today). This however is not what
> I
> describe. I have seen a ping of 6-8 consistently on
> >the scoreboard with the above settings and whilst it is not frequent, it
> is
> impossible, and therefore must be somehow incorrect. >The oddity is also
> independant of my rate settings, or nearly so, on default rates at times
> when I am experiencing this, my ping >will simply sit from 5-6 and not
> deviate for minutes at a time.
>
> ___
> To unsubscribe, edit your list preferences, or view the list archives,
> please visit:
> http://list.valvesoftware.com/mailman/listinfo/hlds
>
--

___
To unsubscribe, edit your list preferences, or view the list archives, please 
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds

Re: [hlds] Re: Please Explain Latency Measurements in Counter-Strike Source

[Whisper]

--
[ Picked text/plain from multipart/alternative ]
Kevin
 The server runs pure, so its not admins being nasty or anything like that.
I can provide the info as you ask though.

 On 7/21/05, Dabosman <[EMAIL PROTECTED]> wrote:
>
> Tickrate is the number of clock cycles (I believe) that the server
> processes
> (per second?). Someone else could explain much better here I'm sure - but
> basically it increases the amount of processing power it uses to run your
> server - and therefore provides much more stable pings in my testing ..
> And
> is why I now run with '-tickrate 66' on the command-line for my server.
> Hitboxes seem to register much better as well.
>
> Are you absolutely positive your client settings are 2/100/100 - and
> that some unscrupulous admin didn't change them on you (with a client exec
> admin command) without your knowing? Please go verify them for me while in
> CS by typing all three for me (rate, cl_updaterate, cl_cmdrate).
>
> If they are correct, I'd like to see a screenshot of your net_graph data
> along with the scoreboard on that same screen showing the 'strange' ping.
> You can post it on my Counter-strike website. Let me know when you have it
> ready and I'll set you up on there.
>
> Also, have you tried another server to see if it displays the same
> 'strange'
> ping on there? You can try mine if you'd like:
> 65.69.123.132:27015 
>
> Kevin
>
>
> Whisper <[EMAIL PROTECTED]> wrote:
>
> >Here is the things that I don't understand The Server can be a 33 or 66
> tickrate server.
> >
> >Both with default sv_maxupdaterates
> >
> >My clients rates are 2/100/100
> >
> >Yes Im aware of the fudge you can do by lowering your rates to give you a
> very low ping, but in the case I presented initially I >am not doing
> anything to manipulate them from my normal game play.
> >
> >Thanks
>
>
> ___
> To unsubscribe, edit your list preferences, or view the list archives,
> please visit:
> http://list.valvesoftware.com/mailman/listinfo/hlds
>
--

___
To unsubscribe, edit your list preferences, or view the list archives, please 
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds

Re: [hlds] Re: Please Explain Latency Measurements in Counter-Strike Source

[James Tucker]

On 7/20/05, Dabosman <[EMAIL PROTECTED]> wrote:
[thank you, but -snip-]
> Again though, I'm betting your problem with the 10ms or very low ping
> showing on the scoreboard is your cl_updaterate setting.  I found this out
> when I saw someone with a '5' or something on the scoreboard - and after
> some investigating, found out their rates were set very low.  (Also check
> your 'rate' setting - should be at least 8192 for DSL/Cable/etc.)

rate 20; echo rate set to 2mbps
cl_updaterate 60; echo requesting 60 packets per second from the server.
cl_cmdrate 60; echo sending 60 packets per second to the server.

I have seen people running around with cl_cmdrate 0/1/generally under
10 and this does result in a falsely returned ping on the scoreboard
(or used to, haven't checked today). This however is not what I
describe. I have seen a ping of 6-8 consistently on the scoreboard
with the above settings and whilst it is not frequent, it is
impossible, and therefore must be somehow incorrect. The oddity is
also independant of my rate settings, or nearly so, on default rates
at times when I am experiencing this, my ping will simply sit from 5-6
and not deviate for minutes at a time.

> Check it and let me know.
>
> Hope this helps.
>
> Kevin
>
> ___
> To unsubscribe, edit your list preferences, or view the list archives, please 
> visit:
> http://list.valvesoftware.com/mailman/listinfo/hlds
>

___
To unsubscribe, edit your list preferences, or view the list archives, please 
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds

Re: [hlds] Re: Please Explain Latency Measurements in Counter-Strike Source

[Whisper]

--
[ Picked text/plain from multipart/alternative ]
Here is the things that I don't understand
 The Server can be a 33 or 66 tickrate server.
 Both with default sv_maxupdaterates
 My clients rates are 2/100/100
 Yes Im aware of the fudge you can do by lowering your rates to give you a
very low ping, but in the case I presented initially I am not doing anything
to manipulate them from my normal game play.
 Thanks
 On 7/21/05, Dabosman <[EMAIL PROTECTED]> wrote:
>
> James,
>
> I've noticed also that the scoreboard will show a much lower (inaccurate)
> ping if your UPDATE RATE (cl_updaterate) is also set incorrectly (say to 0
> or 1 etc.). Try checking your: cl_updaterate ... And make sure it's at
> around 20 (the default). Or you might put it around 60 since you said your
> cl_cmdrate was 60 earlier - but I personally play with 'cl_cmdrate 40' and
> 'cl_updaterate 30' - which is around 10 above normal on each setting. Not
> really required but just personal preference.
>
> Also, I've noticed that if I DON'T use the '-tickrate 66' setting on my
> server, that the ping in net_graph 3 will consistently fluctuate a decent
> bit higher than my actual 'scoreboard' rate (default tickrate for servers
> is
> '33'). For instance, my ping is normally around 55-60ms for my server. The
> net_graph ping will sometimes fluctuate up to 70-90ms without the tickrate
> 66 setting.
>
> The reason for this is that the scoreboard ping is showing an AVERAGE ping
> rate of what your ping has been over the past x seconds/minutes (I do not
> know what x is but I'm guessing around 30-60 seconds) - whereas the
> net_graph ping rate is showing REAL-TIME ping rate. I've noticed if I *DO*
> set my '-tickrate 66' for my server (on the server command-line when
> starting up .. Not console), that the ping *will* stay consistently around
> 55-60ms consistently on the server - unless there is actual server lag
> (unusual since I'm on a full dedicated T1 line). I should also mention
> though that I use the small 'srcdsfpsboost.exe' running in the background
> at
> all times - that allows the server to go above the default 64fps - up to
> 255fps (I have the fps_max server setting at 300). This seems to allow for
> a very smooth and responsive server in my use. Everyone comments on how
> much faster and fluid the server is when playing.
>
> Again though, I'm betting your problem with the 10ms or very low ping
> showing on the scoreboard is your cl_updaterate setting. I found this out
> when I saw someone with a '5' or something on the scoreboard - and after
> some investigating, found out their rates were set very low. (Also check
> your 'rate' setting - should be at least 8192 for DSL/Cable/etc.)
>
> Check it and let me know.
>
> Hope this helps.
>
> Kevin
>
> ___
> To unsubscribe, edit your list preferences, or view the list archives,
> please visit:
> http://list.valvesoftware.com/mailman/listinfo/hlds
>
--

___
To unsubscribe, edit your list preferences, or view the list archives, please 
visit:
http://list.valvesoftware.com/mailman/listinfo/hlds