[TUSA/TX-Gunslinger]

I'd like to add another few small items to Blackdog's excellent list. There is nothing in the checklist that I personally have not been doing - I love his list because it is coherent and one of the only pieces of structured guidance for beta patch setup that exists in our forums here.

I have installed and tested each of the release and beta patches since the beginning - saving and documenting results for each iteration. There have been many improvments since the beginning - but overall frame rates on my system when properly configured average around 42 fps - with min in low 20's - max in the mid 90's.

I got home late last night, so I was quite tired when I ran down the "Blackdogs Checklist". On my first TBD track benchmark - my average frame rate was around 21 fps. After checking my Catalyst, Game settings, configs and system to ensure all settings were consistent with my standard testing. I re-ran the benchmark after rebooting my system. Low frame rates again!

I gave up, went to bed and realized after I woke up, that I'd forgotten to clean out my 1C Softclub/Cache directory. Stupid me. Cleaned the cache, restarted CoD then ran TBD track for the bench. Crap! Back to 21 fps again. Went to work.

After I came home, I restarted my system - checked everything again - recleaned the cache. Still 20-21 fps. Restarted CoD and ran a second time and BANG 42 fps!

Why? **The 1C Softclub\cache has to rebuild itself after you delete the files. So CoD has to be run at least 1 time and closed again for this to happen.

Here are some additional steps:

1. Delete all files in /My Documents/1C SoftClub/il-2 sturmovik cliffs of dover/cache
2. Start CoD
3. Playback "The Black Death" track - you can stop it after a few seconds.
4. Exit CoD
5. Restart CoD
6. Conduct testing

While I'm on this point, maybe it's good that I add this part.

1. Play "The Black Death" Track once all the way through, looking for anomalies or bugs - but don't use this benchmark.
2. Keep the benchmarks from the second and subequent playbacks, and report them together.

Why? On the first track playback, the track is loaded from disk to RAM. You'll find the second playback much more stable than the first (if you conduct enough benchmarks to be statistically significant), because of the random access loadup.

This way, random access storage is removed from the equation, and we are isolating the CPU/RAM/Video sections of each system. This makes the FPS results, much more comparable and non-dependent on the variables of storage, thus giving a better picture of cross-platform relationships and dependencies.

You can always report the first load, sometimes I do - but I keep these results separated.

In summary, if your not making multiple, full track benchmarks and not saving results to files, your results may not be all that usable, and most importantly - you can't see what is really occuring in your system, patch-to-patch.

S!

Gunny