Login and multiplayer play

[ADukes]

How about reducing the graphic card settings in the Nvidia/ATI control centers?

[Olszowy]

Tried lowering the card settings and made sure it was optimized for performance. All three machines also replaced gcards as part of this with nvidia geforce 600 GT series cards. Performance was not improved by the new cards... I was thinking there might be an issue between screen resolution and what SOW set the default to on install. It set mine to my current size without issue during install. Hmmm, have to go one by one to see if anything helps in that regard.
I know with certainty all three started having issues the day we installed 1.5 as we played a game that evening after walking everyone through install and making sure everyone was updated. That first game ran sluggish taking us three hours to get through 2 hours of game time. I missed the next Thursday game but was told it was just as bad. This Thursday I was on again and it was just as bad. That's when we went into troubleshooting mode. All three comps involved used the original 1.5 file from Matrix. I did not using the 1.5rev file as the first version could not set to D drive. That is the only difference. Last night I recommended the machines involved do a total wipe of all files, mods, zips, regedit folders, shut down, and then reinstall. Still verifying it occurred and then will test.
It has got to be something simple.

[Olszowy]

Well, after doing everything that was suggested and thought of we tried to play Thursday evening and multiplayer was unplayable. OOB was below 40,000, but it makes no difference if it is 10000 or 100000, couple machines still stubbornly stay below 5FPS. It was not like this with 1.3 or 1.4. Same machine, same OOBs, same scenarios, but something is causing so much lag that we were running 2 to 2 1/2 minutes real time for every minute of game time. Units would move a short distance, itck tick tick tick, move, tick tick tick. Completely unplayable. So much for multiplayer and that is why I play.

[Jim]

This is quite frustrating because we certainly did not see anything like this in our testing of MP before we released the final build to Matrix. We would have delayed the release to fix an issue of this magnitude.

Are you able to identify which machines are slowing everyone down? If the slow systems drop off, what FPS are you getting from the rest of the group?

-JIm

[Garnier]

@Olszowy

Can each player run the same scenario in singleplayer with all the same settings and get good FPS, but when you do it multiplayer, you get horrible FPS?

Try just two people. Have each one start the scenario in singleplayer and watch performance. Then have those two people play MP with each other and watch it.


I assume you've tried having everyone hit T three times in game to make sure sprite rendering isn't the issue.

[Olszowy]

I know exactly what two machines are causing the problem. Yes, both machines in question were tested in single player, MP, mods, graphics, OOBs, stock scenarios, etc, trees on and off. I know exactly what the FPS is under each circumstance. What I do not understand is why prior to 1.5 both those machines ran fine. We had no lag issues whatsoever even with OOBs running a combined total of 160,000 and decent graphics on. The 1.5 install killed performance, no doubt about it, but I have no idea why. Nothing seems to improve it. Both machines drop to 4-5FPS in MP and lag everyone down. Sp performance is poor as well and yes we did clean reinstalls.
As soon as we kick both machines from MP performance goes back up above 10, into high teens, or better and all is fine. My machine smokes along at 30-50 or higher depending on the map(high as 115 on cville maps), but cannot overcome the 4-5 from those two machines. I hate to tell them that they can no longer play especially when I KNOW they both worked fine the week before the 1.5 patch.
The Thursday before 1.5 we had five corps at 10000 strong in a line running from Nicodemous Hill through the West Wood past the East wood and extending another half mile or so versus at least 7 CSA divisions and it ran fine. The next week 1.5 came out and this happened. I just don't get it. I know SOW was not built for large fights like this in mind, but even with a mere 37000 last night FPS was 4-5 one those two machines. Is this intentional piece of design? Is this NSD's way of saying you must stay at div and lower?

[Garnier]

@Olszowy

I take it that even with everything turned off, in singleplayer, these two machines run it at 4-5 FPS?

And if so, have you tried pausing the game in singleplayer on one of these machines while it's getting 4-5 FPS? If pausing the game brings FPS back up, we know it's the gameplay/AI calculations. If pausing does not fix it, then it's the graphics.

[Olszowy]

@Olszowy

Can each player run the same scenario in singleplayer with all the same settings and get good FPS, but when you do it multiplayer, you get horrible FPS?

Try just two people. Have each one start the scenario in singleplayer and watch performance. Then have those two people play MP with each other and watch it.


I assume you've tried having everyone hit T three times in game to make sure sprite rendering isn't the issue.

The best FPS single player or MP on either machine post 1.5 is no more than 15. Lowering graphic setting and turning trees off gain 2-4 FPS. At game start FPS will be 11-12. The second a corps or division sized unit begins to move ANYWHERE on the map or on either side FPS drops like a rock. After 3-4 units move FPS stays at 4-6. For me I will lose 1 FPS per corps sized unit as it starts moving, but as units stop and get into combat FPS goes up. Not so with those two machines.

Stats on one the machines is an AMD 9950 Quad Core 2.61, 4 Gig Ram / 2 Gig Video Nvidia, the other one was posted above. The one posted previous is the worst of the lot and causing the worst lag. Machines are on different ISP and performance is well below what it should be in SP as well.

[Olszowy]

I have not tried pausing the game to see if FPS goes back up on those machines, but will find out. I tried it on my machine with the saved copy of the game we played last night and I went from 45-168. Least it will narrow down what we are looking for.

[Jim]

The inverse test for graphics limitations is to point the camera at bare grass at minimum altitude. If the FPS goes up then the issue is graphics. One of our testers used to have a system that was graphics limited. For MP testing he would point the camera down and just peek up every now and then to give new orders. From the specs listed above, graphics *should* not be limiting but odd things do happen. We need to narrow this down as much as possible to give Norb the best clues possible as to where in the code this issue might have come from.


-Jim