Is there a standard demo so it's possible to FPS benchmark the game between builds, etc? Every build I need to create a new demo for benchmark purposes.
There is no good way to do this, and the best way is hacky. Here is the best way, because it will get the data the developers need to do something about bad performance.
1. Open a listen server with the map Kodiak or Summit
2. In console enter "sv_cheats 1"
3. In console enter "scenload xl"
3. In console enter "p_logall"
4. Run a loop around the map as fast as you can.
5. In console enter "P_endlog"
6. Find the *.plog file in the same place as your log.txt : %appdata%/Natural selection 2/
7. Send the *.plog file to either Ironhorse or Mclglaspie if there is a problem. WITHOUT THIS *.plog FILE NOTHING CAN BE FIXED. Keep the *.plog under 5 minutes so that is it's manageable to be looked at.
This is how Matso used to benchmark ns2 during development. "Scenload xl" loads up the map with a predetermined set of structures all over the map, as if in late game conditions. The *.plog file is a collection of data about how ns2 is performing. This is 100% necessary for the developers to actually be able to fix anything with performance. The *.plog file tells them what needs to be fixed.
Beyond that, the rest of the testing procedure is up to you. Feel free to use fraps to capture fps data such as min, max, average fps. Just make sure you keep your testing consistent between builds.
For something standardish probably best bet is the run Cr4zy put together, see here under "Settings used" and "How to reproduce"
He had a consistent testing procedure, which was good, but it would have been better if he used sceneload xl. It is standard. But of all things, make sure you make plog files as I described because that is how the developers know what to fix.
Instead of saying "there is a problem" like cr4zy did. Get the data and say "There is a problem, here is the data."
But of all things, make sure you make plog files as I described because that is how the developers know what to fix.
If you're trying to do something diagnostic for feedback sure, but not if all you're doing is benchmarking under game-realistic conditions for your own self
Comments
1. Open a listen server with the map Kodiak or Summit
2. In console enter "sv_cheats 1"
3. In console enter "scenload xl"
3. In console enter "p_logall"
4. Run a loop around the map as fast as you can.
5. In console enter "P_endlog"
6. Find the *.plog file in the same place as your log.txt : %appdata%/Natural selection 2/
7. Send the *.plog file to either Ironhorse or Mclglaspie if there is a problem. WITHOUT THIS *.plog FILE NOTHING CAN BE FIXED. Keep the *.plog under 5 minutes so that is it's manageable to be looked at.
This is how Matso used to benchmark ns2 during development. "Scenload xl" loads up the map with a predetermined set of structures all over the map, as if in late game conditions. The *.plog file is a collection of data about how ns2 is performing. This is 100% necessary for the developers to actually be able to fix anything with performance. The *.plog file tells them what needs to be fixed.
Beyond that, the rest of the testing procedure is up to you. Feel free to use fraps to capture fps data such as min, max, average fps. Just make sure you keep your testing consistent between builds.
He had a consistent testing procedure, which was good, but it would have been better if he used sceneload xl. It is standard. But of all things, make sure you make plog files as I described because that is how the developers know what to fix.
Instead of saying "there is a problem" like cr4zy did. Get the data and say "There is a problem, here is the data."
If you're trying to do something diagnostic for feedback sure, but not if all you're doing is benchmarking under game-realistic conditions for your own self