fsfodukJoin Date: 2004-04-09Member: 27810Members, NS2 Developer, Constellation, NS2 Playtester, Squad Five Blue, Squad Five Silver, Squad Five Gold, Subnautica Playtester, NS2 Community Developer, Pistachionauts
if you type "profile" in the console and press space when you get a frame stutter when you click on a node its % of total frame time of last couple of frames will be overlayed on the bottom graph
Is this occurring for anyone that does <b>not</b> use Windows 7 64-bit?
Please only speak up if it is a constant issue, not the occasional lag or frame loss, or even load lag. We are talking about every 5-10 seconds of moving, all game.
ok so this happens to me, and it's not a problem with lan game or whatever, although it's more pronounced online.
To test, i ran into spectator mode, looked away from everything, and then turned on profiling.
Results: once every second or so, 4 frames have much higher processing times than normal. This is caused by a function call to ScriptGC_STEP. Also, the processing time for ProfileDisplay::Render tends to be higher in these frames, despite not having anything extra to render.
normal frame: <img src="http://s2.kimag.es/share/68828311.jpg" border="0" class="linked-image" />
the stutter does not start happening until you physically move around (mouselook doesn't produce the stutter). However, once you move, the stutter persists until you respawn/change teams/go to rr.
scriptGC - hmm possibly the lua garbage collector. 5.1 is supposed to be efficient wrt GC, but perhaps they need to tune its behavior.
edit: they have "Profile and optimize OnProcessMove()" on the progress list - the scriptgc call perf will be a part of that analysis I'm sure. hopefully next patch.
and then i also posted a video of the stuttering in FPS thread, playing in a LAN game as a skulk, best place to show the stuttering: in vents moving at high speeds back and forth, it consistently occurs. this lead me to believe it was VRAM access as the next area would be rendered, even within the vent.
heres the video: <a href="http://www.youtube.com/watch?v=8UcM7PW8Yck" target="_blank">http://www.youtube.com/watch?v=8UcM7PW8Yck</a>
goodluck to devs on fixing this one. its been buggin me for months. p.s. great graphs guys, and also good teamwork on isolating it from framerate and latency etc!
Windows 7 Pro 64-bit i7 860 @ 2.8ghz HD5850 @ GPU 775Mhz // Memory 1125Mhz 4gb ram
Lots of stuttering here, can't shoot moving things. Rifle firing sound won't stop sometimes. Playing on 1366x768 and 1920x1080 on worst, medium and high settings I get 40-50fps playing on both and for some odd reason I get 100-120 when spectating the same areas.
I've just tested it on the simple map i was making to learn the editor, its a room with a short corridor, when i walk down it theres a stutter every two seconds, i think its this stuttering that makes the game feel so unplayable, it really needs to be a high priority for the devs.
PlasmaJoin Date: 2003-04-26Member: 15855Members, Constellation, Squad Five Blue
I think there's a console command to turn off the Lua GC? Check the lua script files; I seem to recall turning it off (theres a client and server GC you need to turn off).
From memory I tried turning it off and I still had stutter on occasion, but that was several patches ago closer to alpha, so perhaps its different now.
Regardless though; I think if the GC is responsible for the hitches then it should be done in its own thread. Perhaps the realtime portion of the GC can simply mark things to be free'd/checked and let another thread handle verification before doing a GC sweep etc.
I don't know if its just me being paranoid, but I seem to wonder if the stutter is due to sounds? For example, I notice a stutter when buying a gun for the first time on occasion, or opening a door, etc, which play sounds I have not heard so far.
<!--quoteo(post=1818264:date=Dec 22 2010, 12:01 AM:name=hyperrift)--><div class='quotetop'>QUOTE (hyperrift @ Dec 22 2010, 12:01 AM) <a href="index.php?act=findpost&pid=1818264"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Noticing a theme of GTX cards and windows 7 64 bit.<!--QuoteEnd--></div><!--QuoteEEnd-->
maybe its end user problem
955 BE quad core 4 gigs ram win7 64
gtx 470
no studder problem, just the usual warpy skulks/lag issue do to the unoptimized netcode. this guy that says "you dont know what youre talking about" fails to realize that if youre not synced with the server, this issue can happen. the same thing happens when you get the connection issue msg on the screen. you can still move around the game world but youre not really communicating with the server. this example is a larger case, where youd have to reconnect to the server. i can almost guarantee this is whats happening to him.
<!--quoteo(post=1818590:date=Dec 23 2010, 12:23 AM:name=VeNeM)--><div class='quotetop'>QUOTE (VeNeM @ Dec 23 2010, 12:23 AM) <a href="index.php?act=findpost&pid=1818590"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->maybe its end user problem
955 BE quad core 4 gigs ram win7 64
gtx 470
no studder problem, just the usual warpy skulks/lag issue do to the unoptimized netcode. this guy that says "you dont know what youre talking about" fails to realize that if youre not synced with the server, this issue can happen. the same thing happens when you get the connection issue msg on the screen. you can still move around the game world but youre not really communicating with the server. this example is a larger case, where youd have to reconnect to the server. i can almost guarantee this is whats happening to him.<!--QuoteEnd--></div><!--QuoteEEnd--> I'm really sick of seeing people who don't have the issue not read the thread before commenting blindly. This issue happens both LOCALLY and ONLINE. This therefore means it is nothing to do with latency or netcode. Please, at least attempt to read the thread before you reply again.
<!--quoteo(post=1818568:date=Dec 23 2010, 12:20 AM:name=ironhorse)--><div class='quotetop'>QUOTE (ironhorse @ Dec 23 2010, 12:20 AM) <a href="index.php?act=findpost&pid=1818568"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->i made a bug report about this about 5 months ago:
and then i also posted a video of the stuttering in FPS thread, playing in a LAN game as a skulk, best place to show the stuttering: in vents moving at high speeds back and forth, it consistently occurs. this lead me to believe it was VRAM access as the next area would be rendered, even within the vent.
heres the video: <a href="http://www.youtube.com/watch?v=8UcM7PW8Yck" target="_blank">http://www.youtube.com/watch?v=8UcM7PW8Yck</a>
goodluck to devs on fixing this one. its been buggin me for months. p.s. great graphs guys, and also good teamwork on isolating it from framerate and latency etc!
-Iron<!--QuoteEnd--></div><!--QuoteEEnd-->
I think VRAM can be ruled out. Lowest Details lower VRAM usage heavily as the texture resolution is really low. I don't have any program to monitor VRAM usage (you can do it with NV-cards as far as I know). But it r_stats is correct, I never get over 700MB (620MB max. on Highest Detail and Rockdown - under 200MB with lowest detail) and I still get stuttering no matter what detail level I used. I think something like garbage collection is the obvious answer or some other thing that happens regularly. I really hope the devs find the "leak" and can fix it soon.
Reducing the lag is the most important thing at the moment in my opinion. It degrades game experience really much and is sometimes really frustrating. I'm also aware that the devs know about the problem and are working on it. I only hope they understand that it's really important to fix this issue soon. To be honest, it's really not much fun to play when you lose half of the fights because you can't hit anything warping around like a fade... :S
Kouji_SanSr. Hινε UÏкεεÏεг - EUPT DeputyThe NetherlandsJoin Date: 2003-05-13Member: 16271Members, NS2 Playtester, Squad Five Blue
edited December 2010
<!--quoteo(post=1818595:date=Dec 23 2010, 12:52 AM:name=jamieshepherd)--><div class='quotetop'>QUOTE (jamieshepherd @ Dec 23 2010, 12:52 AM) <a href="index.php?act=findpost&pid=1818595"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Stuff about people not reading...<!--QuoteEnd--></div><!--QuoteEEnd--> You do realize that when you play on a localhost <b>server</b>, Spark actually runs a listenserver alongside the client which you then connect to. Meaning it doesn't completely rule out latency or server performance issues...
The messages you see when starting up a localhost should've been quite obvious...
Connecting... Authenticating...
<!--c1--><div class='codetop'>CODE</div><div class='codemain'><!--ec1-->Or in the console: Connecting to server 127.0.0.1:27015 Client connected(127.0.0.1) Client connected and authenticated on the server Loading 'maps ns2_rockdown' (for example)<!--c2--></div><!--ec2-->
And it is even possible to see this one from time to time when hosting locally!
<!--c1--><div class='codetop'>CODE</div><div class='codemain'><!--ec1-->Disconnected (Timeout) Client disconnected (127.0.0.1) Server shutdown<!--c2--></div><!--ec2-->
FFS we even have latency of 20ms-50ms on a localhost! Of course it is not sure if latency is the main cause, but it is certain it cannot be ruled out as one of the possible causes! Alongside the occlusion stutter, which I have observed and the movement scripts causing issues as Wheeee has found... Ruling things out completely, before having them investigated by the programming gods is not helping at all...
In fact your posts, telling people to "be quiet" and "go away" for "not reading" is actually quite counterproductive...
<!--quoteo(post=1818611:date=Dec 23 2010, 01:44 AM:name=meb2)--><div class='quotetop'>QUOTE (meb2 @ Dec 23 2010, 01:44 AM) <a href="index.php?act=findpost&pid=1818611"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->uhhh I don't think any of you have it as bad as I do in-game... is ns2 doing this for anyone else? I captured a bit of it
<a href="http://www.youtube.com/watch?v=ch4OLz4XXjE" target="_blank">http://www.youtube.com/watch?v=ch4OLz4XXjE</a><!--QuoteEnd--></div><!--QuoteEEnd--> My goodness this is golden lol
<!--quoteo(post=1818613:date=Dec 23 2010, 01:51 AM:name=Kouji_San)--><div class='quotetop'>QUOTE (Kouji_San @ Dec 23 2010, 01:51 AM) <a href="index.php?act=findpost&pid=1818613"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Stuff I have no clue about...<!--QuoteEnd--></div><!--QuoteEEnd--> Oh man reading your posts is cringe worthy, please just exit the thread and don't come back. It is not a latency issue, if it's on my local host. I have 0 ping, that means, 0 latency, that means, it is not a latency issue. I don't think you understand this so let me forward a few articles to you:
<a href="http://en.wikipedia.org/wiki/Localhost" target="_blank">http://en.wikipedia.org/wiki/Localhost</a> - My computer ie. no latency <a href="http://en.wikipedia.org/wiki/Latency_(engineering)" target="_blank">http://en.wikipedia.org/wiki/Latency_(engineering)</a> - The unit we're using to measure speed from server to computer (ie. 0 in this case) <a href="http://en.wikipedia.org/wiki/Idiot" target="_blank">http://en.wikipedia.org/wiki/Idiot</a> - You, it appears.
The reason I don't want you to comment, is because:
1) You don't have the problem, therefore you cannot comment on it since you don't even understand what we're getting 2) You are not a programmer, therefore you have no idea what a) the problem might be or b) any solutions for it 3) You are not helping, this thread was made to highlight how many people are getting the issue, and what it is, for the programmers. Not some random idiot who decides that after reading a few threads about 'occlusion culling' and 'lua' he knows what he's talking about.
Kindly exit now. I will not respond to any more of your ridiculous replies, let's get back on topic.
<!--quoteo(post=1818584:date=Dec 22 2010, 05:10 PM:name=Plasma)--><div class='quotetop'>QUOTE (Plasma @ Dec 22 2010, 05:10 PM) <a href="index.php?act=findpost&pid=1818584"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I think there's a console command to turn off the Lua GC? Check the lua script files; I seem to recall turning it off (theres a client and server GC you need to turn off).
From memory I tried turning it off and I still had stutter on occasion, but that was several patches ago closer to alpha, so perhaps its different now.<!--QuoteEnd--></div><!--QuoteEEnd-->
yea the changegcsettingclient and changegcsettingserver didnt seem to do much when i tried tweaking (you can set how frequent & aggressive the gc run is). but this is difficult to determine without knowledge of the internals & without better instrumentation. still kinda fun to speculate.
<!--quoteo(post=1818633:date=Dec 22 2010, 11:15 PM:name=jamieshepherd)--><div class='quotetop'>QUOTE (jamieshepherd @ Dec 22 2010, 11:15 PM) <a href="index.php?act=findpost&pid=1818633"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->flaming, ad hominems, etc etc<!--QuoteEnd--></div><!--QuoteEEnd-->
I believe that everyone has this problem, and some don't notice it because their framerates are low enough that it masks the effect. Have you tried going in a LAN game and reproducing the effect? For me, the effect is much less noticeable. The graph that eggcake made shows that the effect IS less on a LAN server. This could be due to a variety of things, but I think it's premature to rule out some sort of interaction with the server (although yes, the root cause does seem to be the engine, perhaps there is some sort of unintended recursion in the object movement code).
<!--quoteo(post=1818611:date=Dec 22 2010, 06:44 PM:name=meb2)--><div class='quotetop'>QUOTE (meb2 @ Dec 22 2010, 06:44 PM) <a href="index.php?act=findpost&pid=1818611"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->uhhh I don't think any of you have it as bad as I do in-game... is ns2 doing this for anyone else? I captured a bit of it
Kouji_SanSr. Hινε UÏкεεÏεг - EUPT DeputyThe NetherlandsJoin Date: 2003-05-13Member: 16271Members, NS2 Playtester, Squad Five Blue
edited December 2010
<!--quoteo(post=1818640:date=Dec 23 2010, 03:35 AM:name=tk-421)--><div class='quotetop'>QUOTE (tk-421 @ Dec 23 2010, 03:35 AM) <a href="index.php?act=findpost&pid=1818640"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Dude, come on, chill.
No need to link to Wikipedia articles calling people "idiots". Let's stay on topic, this thread is useful.<!--QuoteEnd--></div><!--QuoteEEnd--> It's what he does, can't seem to talk to him about it either. All remarks made on his ugly posting habits thus far fall on stubbornly deaf ears...
It is kinda funny he doesn't know a listenserver although run in the same process thread as the client, can still add latency, or fps stutter issues in this case, if the listenserver or client has a memory leak or simply can't keep... Also he doesn't seem to realize that latency is not always network related...
I for one get a latency of 20ms in the RR and an average latency of about 65ms (which steadily goes down to 30ms-50ms over time) when running around on Rockdown in a listenserver (FYI jamieshepherd, a localhost 127.0.0.1:27015). C2D 2.2Ghz here. So I find it hard to believe he actually has 0ms, even if his CPU is more powerful. It should at least be >0ms in some areas of maps...
Comments
Please only speak up if it is a constant issue, not the occasional lag or frame loss, or even load lag. We are talking about every 5-10 seconds of moving, all game.
edit: i dont mean net lag - really mean local cpu spikes slowing frame rendering. edit2: exacerbated by dynamically allocated MP data.
To test, i ran into spectator mode, looked away from everything, and then turned on profiling.
Results: once every second or so, 4 frames have much higher processing times than normal. This is caused by a function call to ScriptGC_STEP. Also, the processing time for ProfileDisplay::Render tends to be higher in these frames, despite not having anything extra to render.
normal frame:
<img src="http://s2.kimag.es/share/68828311.jpg" border="0" class="linked-image" />
stutter frame 1:
<img src="http://s2.kimag.es/share/32766910.jpg" border="0" class="linked-image" />
stutter frame 2:
<img src="http://s2.kimag.es/share/67394816.jpg" border="0" class="linked-image" />
stutter frame 3:
<img src="http://s2.kimag.es/share/43130510.jpg" border="0" class="linked-image" />
stutter frame 4:
<img src="http://s2.kimag.es/share/10022302.jpg" border="0" class="linked-image" />
the stutter does not start happening until you physically move around (mouselook doesn't produce the stutter). However, once you move, the stutter persists until you respawn/change teams/go to rr.
edit: they have "Profile and optimize OnProcessMove()" on the progress list - the scriptgc call perf will be a part of that analysis I'm sure. hopefully next patch.
in-game:
<img src="http://s2.kimag.es/share/34212767.jpg" border="0" class="linked-image" />
frame1:
<img src="http://s2.kimag.es/share/97748353.jpg" border="0" class="linked-image" />
frame2:
<img src="http://s2.kimag.es/share/11051214.jpg" border="0" class="linked-image" />
scriptGC_STEP is still taking around 20% of the frametime but there's a lot of other stuff going on. Player::OnProcessMove is doing a lot of work.
unfortunately there was too much stuff so you couldn't see the rest of what was going on.
AMD Phenom IIx4 955 3.20 Ghz
4,00 Ram
ATI Radeon HD 5780 series
----Windows 7 64----
<a href="http://getsatisfaction.com/unknownworlds/topics/has_anyone_noticed_missing_frames_with_a_295gtx" target="_blank">http://getsatisfaction.com/unknownworlds/t...s_with_a_295gtx</a>
and then i also posted a video of the stuttering in FPS thread, playing in a LAN game as a skulk,
best place to show the stuttering: in vents moving at high speeds back and forth, it consistently occurs.
this lead me to believe it was VRAM access as the next area would be rendered, even within the vent.
<a href="http://getsatisfaction.com/unknownworlds/topics/extreme_framerate_issues#reply_3226376" target="_blank">http://getsatisfaction.com/unknownworlds/t...s#reply_3226376</a>
heres the video:
<a href="http://www.youtube.com/watch?v=8UcM7PW8Yck" target="_blank">http://www.youtube.com/watch?v=8UcM7PW8Yck</a>
goodluck to devs on fixing this one. its been buggin me for months.
p.s. great graphs guys, and also good teamwork on isolating it from framerate and latency etc!
-Iron
i7 860 @ 2.8ghz
HD5850 @ GPU 775Mhz // Memory 1125Mhz
4gb ram
Lots of stuttering here, can't shoot moving things.
Rifle firing sound won't stop sometimes.
Playing on 1366x768 and 1920x1080 on worst, medium and high settings I get 40-50fps playing on both and for some odd reason I get 100-120 when spectating the same areas.
From memory I tried turning it off and I still had stutter on occasion, but that was several patches ago closer to alpha, so perhaps its different now.
Regardless though; I think if the GC is responsible for the hitches then it should be done in its own thread. Perhaps the realtime portion of the GC can simply mark things to be free'd/checked and let another thread handle verification before doing a GC sweep etc.
I don't know if its just me being paranoid, but I seem to wonder if the stutter is due to sounds? For example, I notice a stutter when buying a gun for the first time on occasion, or opening a door, etc, which play sounds I have not heard so far.
maybe its end user problem
955 BE quad core
4 gigs ram
win7 64
gtx 470
no studder problem, just the usual warpy skulks/lag issue do to the unoptimized netcode. this guy that says "you dont know what youre talking about" fails to realize that if youre not synced with the server, this issue can happen. the same thing happens when you get the connection issue msg on the screen. you can still move around the game world but youre not really communicating with the server. this example is a larger case, where youd have to reconnect to the server. i can almost guarantee this is whats happening to him.
955 BE quad core
4 gigs ram
win7 64
gtx 470
no studder problem, just the usual warpy skulks/lag issue do to the unoptimized netcode. this guy that says "you dont know what youre talking about" fails to realize that if youre not synced with the server, this issue can happen. the same thing happens when you get the connection issue msg on the screen. you can still move around the game world but youre not really communicating with the server. this example is a larger case, where youd have to reconnect to the server. i can almost guarantee this is whats happening to him.<!--QuoteEnd--></div><!--QuoteEEnd-->
I'm really sick of seeing people who don't have the issue not read the thread before commenting blindly. This issue happens both LOCALLY and ONLINE. This therefore means it is nothing to do with latency or netcode. Please, at least attempt to read the thread before you reply again.
<a href="http://getsatisfaction.com/unknownworlds/topics/has_anyone_noticed_missing_frames_with_a_295gtx" target="_blank">http://getsatisfaction.com/unknownworlds/t...s_with_a_295gtx</a>
and then i also posted a video of the stuttering in FPS thread, playing in a LAN game as a skulk,
best place to show the stuttering: in vents moving at high speeds back and forth, it consistently occurs.
this lead me to believe it was VRAM access as the next area would be rendered, even within the vent.
<a href="http://getsatisfaction.com/unknownworlds/topics/extreme_framerate_issues#reply_3226376" target="_blank">http://getsatisfaction.com/unknownworlds/t...s#reply_3226376</a>
heres the video:
<a href="http://www.youtube.com/watch?v=8UcM7PW8Yck" target="_blank">http://www.youtube.com/watch?v=8UcM7PW8Yck</a>
goodluck to devs on fixing this one. its been buggin me for months.
p.s. great graphs guys, and also good teamwork on isolating it from framerate and latency etc!
-Iron<!--QuoteEnd--></div><!--QuoteEEnd-->
I think VRAM can be ruled out. Lowest Details lower VRAM usage heavily as the texture resolution is really low. I don't have any program to monitor VRAM usage (you can do it with NV-cards as far as I know). But it r_stats is correct, I never get over 700MB (620MB max. on Highest Detail and Rockdown - under 200MB with lowest detail) and I still get stuttering no matter what detail level I used. I think something like garbage collection is the obvious answer or some other thing that happens regularly. I really hope the devs find the "leak" and can fix it soon.
Reducing the lag is the most important thing at the moment in my opinion. It degrades game experience really much and is sometimes really frustrating.
I'm also aware that the devs know about the problem and are working on it. I only hope they understand that it's really important to fix this issue soon. To be honest, it's really not much fun to play when you lose half of the fights because you can't hit anything warping around like a fade... :S
<a href="http://www.youtube.com/watch?v=ch4OLz4XXjE" target="_blank">http://www.youtube.com/watch?v=ch4OLz4XXjE</a>
You do realize that when you play on a localhost <b>server</b>, Spark actually runs a listenserver alongside the client which you then connect to. Meaning it doesn't completely rule out latency or server performance issues...
The messages you see when starting up a localhost should've been quite obvious...
Connecting...
Authenticating...
<!--c1--><div class='codetop'>CODE</div><div class='codemain'><!--ec1-->Or in the console:
Connecting to server 127.0.0.1:27015
Client connected(127.0.0.1)
Client connected and authenticated on the server
Loading 'maps ns2_rockdown' (for example)<!--c2--></div><!--ec2-->
And it is even possible to see this one from time to time when hosting locally!
<!--c1--><div class='codetop'>CODE</div><div class='codemain'><!--ec1-->Disconnected (Timeout)
Client disconnected (127.0.0.1) Server shutdown<!--c2--></div><!--ec2-->
FFS we even have latency of 20ms-50ms on a localhost! Of course it is not sure if latency is the main cause, but it is certain it cannot be ruled out as one of the possible causes! Alongside the occlusion stutter, which I have observed and the movement scripts causing issues as Wheeee has found... Ruling things out completely, before having them investigated by the programming gods is not helping at all...
In fact your posts, telling people to "be quiet" and "go away" for "not reading" is actually quite counterproductive...
@meb2, Skulk rush is a valid tactic :P
lol
<a href="http://www.youtube.com/watch?v=ch4OLz4XXjE" target="_blank">http://www.youtube.com/watch?v=ch4OLz4XXjE</a><!--QuoteEnd--></div><!--QuoteEEnd-->
My goodness this is golden lol
Oh man reading your posts is cringe worthy, please just exit the thread and don't come back. It is not a latency issue, if it's on my local host. I have 0 ping, that means, 0 latency, that means, it is not a latency issue. I don't think you understand this so let me forward a few articles to you:
<a href="http://en.wikipedia.org/wiki/Localhost" target="_blank">http://en.wikipedia.org/wiki/Localhost</a> - My computer ie. no latency
<a href="http://en.wikipedia.org/wiki/Latency_(engineering)" target="_blank">http://en.wikipedia.org/wiki/Latency_(engineering)</a> - The unit we're using to measure speed from server to computer (ie. 0 in this case)
<a href="http://en.wikipedia.org/wiki/Idiot" target="_blank">http://en.wikipedia.org/wiki/Idiot</a> - You, it appears.
The reason I don't want you to comment, is because:
1) You don't have the problem, therefore you cannot comment on it since you don't even understand what we're getting
2) You are not a programmer, therefore you have no idea what a) the problem might be or b) any solutions for it
3) You are not helping, this thread was made to highlight how many people are getting the issue, and what it is, for the programmers. Not some random idiot who decides that after reading a few threads about 'occlusion culling' and 'lua' he knows what he's talking about.
Kindly exit now. I will not respond to any more of your ridiculous replies, let's get back on topic.
No need to link to Wikipedia articles calling people "idiots". Let's stay on topic, this thread is useful.
From memory I tried turning it off and I still had stutter on occasion, but that was several patches ago closer to alpha, so perhaps its different now.<!--QuoteEnd--></div><!--QuoteEEnd-->
yea the changegcsettingclient and changegcsettingserver didnt seem to do much when i tried tweaking (you can set how frequent & aggressive the gc run is). but this is difficult to determine without knowledge of the internals & without better instrumentation. still kinda fun to speculate.
I believe that everyone has this problem, and some don't notice it because their framerates are low enough that it masks the effect. Have you tried going in a LAN game and reproducing the effect? For me, the effect is much less noticeable. The graph that eggcake made shows that the effect IS less on a LAN server. This could be due to a variety of things, but I think it's premature to rule out some sort of interaction with the server (although yes, the root cause does seem to be the engine, perhaps there is some sort of unintended recursion in the object movement code).
<a href="http://www.youtube.com/watch?v=ch4OLz4XXjE" target="_blank">http://www.youtube.com/watch?v=ch4OLz4XXjE</a><!--QuoteEnd--></div><!--QuoteEEnd-->
nice custom models :D
No need to link to Wikipedia articles calling people "idiots". Let's stay on topic, this thread is useful.<!--QuoteEnd--></div><!--QuoteEEnd-->
It's what he does, can't seem to talk to him about it either. All remarks made on his ugly posting habits thus far fall on stubbornly deaf ears...
It is kinda funny he doesn't know a listenserver although run in the same process thread as the client, can still add latency, or fps stutter issues in this case, if the listenserver or client has a memory leak or simply can't keep... Also he doesn't seem to realize that latency is not always network related...
I for one get a latency of 20ms in the RR and an average latency of about 65ms (which steadily goes down to 30ms-50ms over time) when running around on Rockdown in a listenserver (FYI jamieshepherd, a localhost 127.0.0.1:27015). C2D 2.2Ghz here. So I find it hard to believe he actually has 0ms, even if his CPU is more powerful. It should at least be >0ms in some areas of maps...
noob server, skulk didn't know how to bunnyhop ;)