SMP client restarts and slower performance?

[APOLLO]

Anyone notice when restarting the SMP2 client that performance drops significantly? I was wondering if I was mistaken but have seen this too many times now that I know it's no my imagination. PPD drops and TPFs increase. This isn't the case only with crashes, regular shutdowns and restarts also seem to lower performance thereafter.

Another thing that makes me wonder about all this is the archaic message I always see displayed after a restart that's really a throwback to the uni-processor clients:

Looking at optimizations...
Working with standard loops on this execution.
Previous termination of core was improper.

 

[Mr. Pedantic]

The standard methods is what is causing the slowdown. Use -forceasm as a flag in the config to force the assembly optimizations.

 

[APOLLO]

The standard methods is what is causing the slowdown. Use -forceasm as a flag in the config to force the assembly optimizations.

OK, with the SMP1 client there was never a slowdown and adding that flag didn't have an effect. Stanford must have changed things when they updated the client. I'll put that flag in the shortcut. Worst that can happen is nothing changes. Thanks for the suggestion.

 

[SmokeRngs]

I've seen PPD drop before after shutting a client down and restarting it on A3 work units due to the bonus. Since there is no way for the work unit to complete quicker due to the "lost time" while the client was down, the PPD will always drop. Well, unless the system was under a heavy load from some other software beforehand.

I've never seen a noticeable frame time increase after shutting down and restarting the client if all things were basically equal before and after.

I know you're not supposed to need the -forceasm flag anymore, but I still do it on all my clients no matter what. I've seen the same standard loops thing before even with the -forceasm flag. No matter what, frame times seem to be the same for me.