bflyblue

Hi Dave, It's wired. Up until now I just assumed it might have been my internet connection. Thing is TCP retransmits should take care of any missing packets, so unless the connection hangs for a while, I'm not sure why CMSF would display the network paused message. Looking at the TCP trace, there are regular packets flowing at the time (server->client) and the client is ACK'ing the packets, so it's not a problem with the transport layer. Other than ACKs, the client last sent actual data 30 secs before the problem occurs. I'm not sure of CMSF expects keepalives at the application layer, but appears the client just stops sending application data even though both client and server are alive, and the TCP connection still established.

I've been trying to get a friend of mine into CMSF by playing the various demos multiplayer, and we've never completed an entire mission because of network issues. I'm sad to say I'm having the exact same problems with the 1.20 demo, it goes into network pause and never recovers. I've even captured all the network traffic of our last game on my firewall to try trace any network issues. Right before the hang, the server sends the client data, and gets ACK packets back from the client for that range of the TCP stream, but the server never sends any more data after that during the network pause. I don't know how to do any more troubleshooting on my own - it seems like its not the network itself, but something to do with how CMSF does TCP multiplayer. I'd be happy to send the devs my packet dump if that will help at all? How many people actually have working TCP multiplayer?

I do a lot of OpenGL development (on Linux) so I know exactly how frustrating it is for developers when you code things to spec, and get different results not only on different hardware, but between different drivers on exactly the same hardware. API's are there for a reason, but it feels like there are so many work-around in graphics drivers these days to get some games to work that it gets in the way of adhering to the spec for other applications. The 8800 does things very differently in hardware compared to previous generations and requires far more complicated drivers. I get the feeling Nvidia rushed things, and it's going to take a long time for the new drivers to stabilise. My 5c..

I tried the balanced setting in the Nvidia panel, forcing off most of the options, and forcing on AF and AA and the numbers stay basically the same (turning off thread optimisation helps the most as I'm using dual core AMD). Moving around a little doesn't seem to do much to the frame rates, but if I pan around 360 degrees, or even look straight up/down and return to the starting view, the speeds drops back down to what it was before I alt-tab out.

I didn't notice a big difference from 163.44 to 163.67 (8800GTX on WinXP) except tabbing out on the latter temporarily gave me a boost from 17 fps to 29 fps. As soon as you start panning the camera around a little this dropped back down to 17 until you tab out again, which seems to confirm the texture memory problem.

I've had some weird path finding issues with buildings too, especially after ordering a squad onto the roof. If you then plot a waypoint on the ground or another building, they sometimes get 'stuck' inside the building and repeated cancel my waypoint (this is in RT mode). Sometimes they simply run through the air from one building to the next across the street!

I've also removed the AMD Dual Core Optimizer and have played several missions since then without it locking up

Also hangs for me and requires and "end task". 1) Battlefront version 1.01 2) Athlon 64 X2 5300+ 3) 2Gb RAM 4) Nvidia 8800 GT I've tried both: - x32 XP SP2 with/without MS hotfix (did improve performance); Forceware 158.22 & 162.18 - x64 Vista; Forceware 158.24 & 162.22 I found turning 'Thread Optimization' to off in the Nvidia settings helped performance.