Mass Benchmarking Required

The latest development version of the Spring engine is worth using except for the fact that many people experience awful performance. Polling and anecdotes say that the problem is system specific and I don't even think it is directly related to the age of your hardware.



The benchmarking tool is now sufficiently advanced and simple to use that everyone should be able to run a useful benchmark. In short

Run a Benchmark

Everyone Windows user should be able to do this (and Linux users if they can run ZKL). If you complained about anything to do with 91.0 (desync, poor pathfollowing) then run this benchmark. Any benchmark is helpful, not just the ones that have problems with the new engine.

Ignore all those things and used the "How to Benchmark" guide without worrying.

The benchmarks use their own settings which are part of what you download during the guide. They are selected in "Config" and ZKL_default is the settings which people will have upon first running ZKL.

Do you need linux benchmarks too?

If you have ZKL on linux and the benchmarker works there then by all means benchmark on linux. I just was not sure whether everyone on linux would be able to.

Done. Had a problem with calculating averages at the end though.

Done. I think Google you may want to lock the sheets once finished, as someome has been playing games.

Added some more windows7 benchmark (probably can do it in Linux as well as I generally play under Linux, but did not yet try to put zkl there).

What are those numbers (I guess units when tests starts, units when tests ends, but what is speedy_dt) ?...

I think it already becomes apparent that 98 is nearly a factor of 2 slower?

Just ran the tests as well and added the results.

Looks like we need more AMD tests. Nearly all tests were with Intel proecessors and Nvidea cards.

PS, why tests must be run 10 times each?+ they are quite boring to watch. + you can zoom I guess this should affect results and so on.