Lag in FPS Starter Kit
by Mike DuVall · in General Discussion · 10/18/2006 (6:21 pm) · 15 replies
I think I'm seeing some lag issues when I run the demo FPS Starter kit. I'm running this on a home network with several machines connected to a NetGear router. I start up the FPS Starter kit as a server on one machine and then start it up again on the other machine as a client and join that server.
I can then setup one player to be standing still and have the other player on the other machine run back and forth in front of him. The player does not move smoothly move back and forth, however, and he moves with a jerky kind of motion.
I don't have the exact specs (processor, video card, memory) of these machines at my fingertips, but both machines are good machines with good video cards that run Battlefield 1942 just fine.
Are there any known issues with this FPS Starter Kit demo app that would explain this? Any suggestions on what could be causing the problem?
Thanks,
Mike
I can then setup one player to be standing still and have the other player on the other machine run back and forth in front of him. The player does not move smoothly move back and forth, however, and he moves with a jerky kind of motion.
I don't have the exact specs (processor, video card, memory) of these machines at my fingertips, but both machines are good machines with good video cards that run Battlefield 1942 just fine.
Are there any known issues with this FPS Starter Kit demo app that would explain this? Any suggestions on what could be causing the problem?
Thanks,
Mike
#2
10/19/2006 (4:35 pm)
Check and see if your router firmware is up to date as well.
#3
I don't know if 1942 vs TGE 1.5 is a good comparison..
TGE can smoke 1942 like a cheap cigar..
10/21/2006 (12:23 pm)
What would you call a good video card?I don't know if 1942 vs TGE 1.5 is a good comparison..
TGE can smoke 1942 like a cheap cigar..
#4
For the record, I'm using Torque 1.4.
As an additional test, I tried to do a side-by-side comparison of the FPS Starter kit game with Battefield 1942. My setup included a standalone dedicated server and two client machines. I used exactly the same setup with BF 1942 as I did with Torque. The test consisted of connecting to the server with each client and placing the two players in front of each other. I would then move one of the players back and forth, side-to-side, in front of the other player, while watching the moving player from the other player's screen. I repeated this with both players. With BF 1942, I saw ABSOLUTELY NO GLITCHES of any kind. The movement was silky smooth and I moved the players over and over and over again.
When I do exactly the same test with the Torque demo, I see glitchy movement. The movement will be fine most of the time, but about every 10th pass, the player will suddenly lurch forward. I have the specs of the hardware I'm using listed below. I'd appreciate any insight has on this problem. Am I making some noob mistake here? Is the Starter FPS game not a good template to be using for a real multi-player game? Anyone up for trying to duplicate my experiment? I know the the FPS starter kit has been around a while, but have any of you personally observed the test I'm currently doing with it?
I've head so much about how good the networking code is in Torque and how it can support a HUGE number of players that I have to say I'm very suprised at these results. The side by side comparison I've done with BF 1942 leads me to believe that there is not something fundamentally wrong with my hardware setup. It may be the case that TGE can "smoke 1942 like a cheap cigar", but the only evidence I have says it's the other way around.
HELP!!!!
Hardware Specs:
Client 1:
HP Pavilion
AMD Athlon 64 X2 DualCore Processor 3800+
1.99 Ghz, 1.00 Gb of Ram
Video card: NVIDIA GeForce 7800 GT
Direct X version: 9.0C
Network card: Realtek RTL8139/810x Family Fast Ethernet NIC
Client 2: (this is the weakest machine)
HP Pavilion
Pentium 4 2.2 Ghz
500M Ram
Video card: Radeon 9250
Direct X version: 8.1
Network card: Realtek RTL8139(A)-based Fast Ethernet Adapter
Server: (dedicated, no graphics)
eMachine T6000
AMD 64 Processor 3200+
2.00 Ghz, 1 Gb of Ram
Video Card: Radeon X800 Pro
DirectX version: 9.0c
Network card: VIA Rhine II Fast Ethernet Adapter
Router: Netgear FVS318, running latest firmware
-Mike
10/23/2006 (3:25 pm)
I'm still having problems and I've done a lot more testing. First I went through every machine I'm using and made sure they had the latest video drivers. I also checked the firmware on my router and it is at the latest update.For the record, I'm using Torque 1.4.
As an additional test, I tried to do a side-by-side comparison of the FPS Starter kit game with Battefield 1942. My setup included a standalone dedicated server and two client machines. I used exactly the same setup with BF 1942 as I did with Torque. The test consisted of connecting to the server with each client and placing the two players in front of each other. I would then move one of the players back and forth, side-to-side, in front of the other player, while watching the moving player from the other player's screen. I repeated this with both players. With BF 1942, I saw ABSOLUTELY NO GLITCHES of any kind. The movement was silky smooth and I moved the players over and over and over again.
When I do exactly the same test with the Torque demo, I see glitchy movement. The movement will be fine most of the time, but about every 10th pass, the player will suddenly lurch forward. I have the specs of the hardware I'm using listed below. I'd appreciate any insight has on this problem. Am I making some noob mistake here? Is the Starter FPS game not a good template to be using for a real multi-player game? Anyone up for trying to duplicate my experiment? I know the the FPS starter kit has been around a while, but have any of you personally observed the test I'm currently doing with it?
I've head so much about how good the networking code is in Torque and how it can support a HUGE number of players that I have to say I'm very suprised at these results. The side by side comparison I've done with BF 1942 leads me to believe that there is not something fundamentally wrong with my hardware setup. It may be the case that TGE can "smoke 1942 like a cheap cigar", but the only evidence I have says it's the other way around.
HELP!!!!
Hardware Specs:
Client 1:
HP Pavilion
AMD Athlon 64 X2 DualCore Processor 3800+
1.99 Ghz, 1.00 Gb of Ram
Video card: NVIDIA GeForce 7800 GT
Direct X version: 9.0C
Network card: Realtek RTL8139/810x Family Fast Ethernet NIC
Client 2: (this is the weakest machine)
HP Pavilion
Pentium 4 2.2 Ghz
500M Ram
Video card: Radeon 9250
Direct X version: 8.1
Network card: Realtek RTL8139(A)-based Fast Ethernet Adapter
Server: (dedicated, no graphics)
eMachine T6000
AMD 64 Processor 3200+
2.00 Ghz, 1 Gb of Ram
Video Card: Radeon X800 Pro
DirectX version: 9.0c
Network card: VIA Rhine II Fast Ethernet Adapter
Router: Netgear FVS318, running latest firmware
-Mike
#5
3700+ amd athlon 64
Video card: NVIDIA GeForce 7800 GT
I'm having no problems with the virgin starter kit. I didn't notice any lag at all until I spawned 100+ bots running around shooting each other. (and me) :/
Have you made any modifications to the kit?
10/23/2006 (6:59 pm)
With those specs you shouldn't be having any problems at all. My machine is 1 lower than yours. 3700+ amd athlon 64
Video card: NVIDIA GeForce 7800 GT
I'm having no problems with the virgin starter kit. I didn't notice any lag at all until I spawned 100+ bots running around shooting each other. (and me) :/
Have you made any modifications to the kit?
#6
10/23/2006 (7:24 pm)
Its the dual core issue again.. I wonder when GG is going to release a proper fix.
#7
Peter:
I don't believe this is a dual core issue. I did further testing using two non-dual core machines, and I see the same problem.
For the record, I just downloaded 1.5 and it has the same problem.
Anybody out there willing to try to duplicate my experiment on their machines(1 server and 2 clients)? I'm also seeing the problem with just 2 machines, 1 acting as both server and client and the other as client only.
Am I expecting too much from Torque? With only 2 players running on a 100MB lan, I'm expecting basically no lag or glitching whatsoever (and that's what I see with BF1942). One friend at work suggested a possibility: If Torque is optimized for 128+ players, perhaps that means limiting the bandwidth allowed for each player, even when there are only 2 players. Any thoughts on that? Does Torque always limit the bandwidth for each player in order to achieve the ability to handle 128+ players? I would have thought it could intelligently divy up the available bandwidth evenly among all players rather than just always cap it in case 128 players happened to join.
One more thing: Have I posted this question on the right forum?
10/26/2006 (6:58 pm)
Mike Rowley: In your test, are you networked across more than one machine? It almost sounds like you are running everything on one machine, in which case, I don't think your results are really pertinent to my problem. I'm not seeing issues when I run on one machine. I see issues when I do mulitplayer across a network. And to answer your question, I made no modifications to code. This is straight off of a virgin install.Peter:
I don't believe this is a dual core issue. I did further testing using two non-dual core machines, and I see the same problem.
For the record, I just downloaded 1.5 and it has the same problem.
Anybody out there willing to try to duplicate my experiment on their machines(1 server and 2 clients)? I'm also seeing the problem with just 2 machines, 1 acting as both server and client and the other as client only.
Am I expecting too much from Torque? With only 2 players running on a 100MB lan, I'm expecting basically no lag or glitching whatsoever (and that's what I see with BF1942). One friend at work suggested a possibility: If Torque is optimized for 128+ players, perhaps that means limiting the bandwidth allowed for each player, even when there are only 2 players. Any thoughts on that? Does Torque always limit the bandwidth for each player in order to achieve the ability to handle 128+ players? I would have thought it could intelligently divy up the available bandwidth evenly among all players rather than just always cap it in case 128 players happened to join.
One more thing: Have I posted this question on the right forum?
#8
10/26/2006 (8:17 pm)
Mike, look for a problem with your network. Torque isn't the problem. 
#9
And no, it does not limit bandwidth on 2 players becuase it is networked for 128 (lol).
That's not really optimizing :)
Anyway, some games around here have been running at 64+ and a few at 128+ or even 256+ without problems.
10/27/2006 (3:26 am)
Yup, typically.And no, it does not limit bandwidth on 2 players becuase it is networked for 128 (lol).
That's not really optimizing :)
Anyway, some games around here have been running at 64+ and a few at 128+ or even 256+ without problems.
#10
10/27/2006 (12:21 pm)
Quote:In your test, are you networked across more than one machine? It almost sounds like you are running everything on one machineI'm networking across 3 machines. 2 PCs and 1 mac using the same netgear router you are.
#11
Speeding Mission Loadtimes
I know I change the default settings in ../client/pref.cs in every Torque game, demo, sdk, etc. to the following with some perceived speedup and no perceived ill effects.
$Pref::Net::PacketRateToClient = 32;
$Pref::Net::PacketRateToServer = 32;
$Pref::Net::PacketSize = 450;
10/27/2006 (12:53 pm)
I wonder if the following resource might help.Speeding Mission Loadtimes
I know I change the default settings in ../client/pref.cs in every Torque game, demo, sdk, etc. to the following with some perceived speedup and no perceived ill effects.
$Pref::Net::PacketRateToClient = 32;
$Pref::Net::PacketRateToServer = 32;
$Pref::Net::PacketSize = 450;
#12
www.garagegames.com/mg/forums/result.thread.php?qt=37115
www.garagegames.com/mg/forums/result.thread.php?qt=37170
10/27/2006 (6:36 pm)
It would appear that this is a known problem. The following threads describe the problem I'm seeing almost perfectly. I'm currently pursuing the fixes they have outlined in these threads:www.garagegames.com/mg/forums/result.thread.php?qt=37115
www.garagegames.com/mg/forums/result.thread.php?qt=37170
#13
hate to be blunt, but question all you like...there is something wrong with your setup.
Let's say you do make those changes and everything clears up - well, that's still an indication the problem is with your machine and/or network. You shouldn't have to change anything to get the results you're wanting. I'm guessing you're using windows and somewhere along the line a program or you changed the network settings..
Maybe a cable modem accelerator? A download accelerator? Anything at all that changed your stock windows setup?
Torque networking works, and works very well. It's been used in many products and many different end user configurations over the years.
If it doesn't work as expected, then the problem is on your end.
10/27/2006 (6:58 pm)
I'm going to quote Mr. Zepp from one of those threads:hate to be blunt, but question all you like...there is something wrong with your setup.
Let's say you do make those changes and everything clears up - well, that's still an indication the problem is with your machine and/or network. You shouldn't have to change anything to get the results you're wanting. I'm guessing you're using windows and somewhere along the line a program or you changed the network settings..
Maybe a cable modem accelerator? A download accelerator? Anything at all that changed your stock windows setup?
Torque networking works, and works very well. It's been used in many products and many different end user configurations over the years.
If it doesn't work as expected, then the problem is on your end.
#14
In addition, take a look at these links:
www.garagegames.com/mg/forums/result.thread.php?qt=19886
www.mindcontrol.org/~hplus/pc-timers.html
In addition, Simon Duggan reported that this problem first appeared for him when he switched from 1.3 to 1.4. To me that strongly indicates that some change from 1.3 to 1.4 could be causing this problem.
I'm not an expert on timing issue on PC platforms, but it seems clear to me that precise timing is a very difficult and tricky thing to nail solidly on Windows machines such that it works consistently across all types of different machines with different processors. It also seems clear that there are some known issues with timing in Torque but it's unclear if these issues have been fixed or when they will be fixed. (See thread above: Bug: QueryPerformanceTimer drifting).
Since most of these issues seem to center around PC/Windows timing, if none of the fixes in these links pans out, I think I'll also investigate running the server in Linux and see if that makes a difference.
Regarding your questions about my networking setup, I have tested this with 4 different machines and 2 different routers in my home network, and I also tested with two completely different machines on a completely different network. I see the same results, but to different degrees, in each case.
10/27/2006 (7:25 pm)
David, I respectfully disagree. Did you read all of both threads? A number of people are having similar problems. At the very least I would say that at this time it is unclear what is causing my problem. It could be my specific setup, my specific configuration, or a timing issue in Torque that only manifests itself on certain types of machines. In the 25 years that I have spent developing software, if there is one thing I have learned, it's "never say never". In addition, take a look at these links:
www.garagegames.com/mg/forums/result.thread.php?qt=19886
www.mindcontrol.org/~hplus/pc-timers.html
In addition, Simon Duggan reported that this problem first appeared for him when he switched from 1.3 to 1.4. To me that strongly indicates that some change from 1.3 to 1.4 could be causing this problem.
I'm not an expert on timing issue on PC platforms, but it seems clear to me that precise timing is a very difficult and tricky thing to nail solidly on Windows machines such that it works consistently across all types of different machines with different processors. It also seems clear that there are some known issues with timing in Torque but it's unclear if these issues have been fixed or when they will be fixed. (See thread above: Bug: QueryPerformanceTimer drifting).
Since most of these issues seem to center around PC/Windows timing, if none of the fixes in these links pans out, I think I'll also investigate running the server in Linux and see if that makes a difference.
Regarding your questions about my networking setup, I have tested this with 4 different machines and 2 different routers in my home network, and I also tested with two completely different machines on a completely different network. I see the same results, but to different degrees, in each case.
#15
10/27/2006 (7:38 pm)
Here's a question I don't think has been asked. Are all these machines with issues AMD? I use intel pretty much exclusively. 
Torque Owner Joshua Dallman
Default Studio Name