Spring RTS Engine

The bug that has been causing a lot of late game stuttering issues, especially with many players and high CPU loads has been fixed.

The engine build 96.0.1-g382e626 CRNfix is just Spring 96.0 with this patch applied.

There are no other changes to the engine from 96.0, only 3 lines of code got changed.

All [ACE] autohosts already have this engine version as default, and I highly recommend playing large team games on servers which host this engine version.

Thanks to everyone!

Much wubb to teh Beherith!

Caldari Navy Raven?
What was actually fixed?

gajop wrote:Caldari Navy Raven?
What was actually fixed?

ClientReadNet

http://springrts.com/mantis/view.php?id=4290

https://github.com/spring/spring/commit ... 898d839a0c

Going to test in about an hour.

FLOZi wrote: ClientReadNet

http://springrts.com/mantis/view.php?id=4290

Nice behe! Simple fix, however it's impressive how you managed to investigate the cause.

Haven't tested yet but good stuff, thanks behe. Anyone done any benchmarking yet?

woot, will this make it into 97?

My apologies to everyone involved, but it turns out I made a mistake while debugging, and this fix only affects stutter issues while watching replays

Meh, it happens Still a good catch.

IIRC aren't replays used in the comparisons of the violation and the spring engine? So it is possible this could rectify much of the bullshit frequently bantered around here.

Either way, it is a fix, I will happily take it behe! Does this effect the resume of a match when you have a crash/got booted or no idea if it would effect resuming a match?

IIRC aren't replays used in the comparisons of the violation and the spring engine? So it is possible this could rectify much of the ignorant stuff frequently bantered around here.

I would guess.

However this just proves that replays shouldnt be used to test engine performance.

Behe its a fix and we will take it, no worries

I 100% agree scifi. That is NOT a good test but people were using it all the same.

OK np, really !

Thx for trying behe!

Replays are useful for benchmarking if and only if they are correctly designed for the task and the engines being compared are similar enough.

Smoth: This CRNfix is nice but does not affect any benchmarking results. Stutter and simulation speed are not even related. I'd recommend you to be more careful in using the word ignorance around here.

ok. hang on while I edit my post.

zer_ver wrote:Replays are useful for benchmarking if and only if they are correctly designed for the task and the engines being compared are similar enough.

Smoth: This CRNfix is nice but does not affect any benchmarking results. Stutter and simulation speed are not even related. I'd recommend you to be more careful in using the word ignorance around here.

Except that in this case it was relevant. Because each netmsg_synccheck forced an extra draw frame, thereby taking away time from doing sim frames.

Yeah, less than 2 fps per second, but it does not count as Sim time so none of my benchmarks would be affected. Also, MT engine would not gain anything from this fix.

Why don't you just release your source and stop this silly shit?