Engine Testing - 96.0.1-443-g9f44dbc (25. May 2014)

[abma]

release candidate #3 for 97.0!

recent tests revealed no major bugs, so we/i plan to release it as 97.0 next weekend. (very likely 24.5. ~midnight)

Changes

only fixes since rc2:

• fix some ATI related crashes/bugs, shadows should work again
• fix underscore char shown as block

To make the next release post more interesting i need your help. i want to add screenshots & maybe a video to the release post, current screenshots/text snippets are:
http://springrts.com/wiki/ReleaseNotes:97.0


Changes in Detail:
changelog.txt (is currently the same as for rc1)
git diff last rc to HEAD

no breaking changes known atm since 96.0


Download
See the download page for how to get it.


Bugs

If you find a bugs, please report to Mantis.

Please attach infolog.txt as file, if you crash!

current release (known) blocking bugs no major bug.

[Beherith]

Me and a few other people have had trouble with hosting games. We have our ports forwarded in our firewalls, other applications work fine and we have spring.exe allowed through our firewalls. What info can we give or what tests/logs should we post to help debug this issue?

I would like to request that the developers wait a little bit, because this hindered our ability to run online proper tests.

[Jools]

release candidate #3 for 97.0!

recent tests revealed no major bugs, so we/i plan to release it as 97.0 next weekend.

That makes me curious. How do you test the releases?

[abma]

That makes me curious. How do you test the releases?

play! as much as possible in different combinations of ai/games/maps/players.

[Silentwings]

Could the changelog be updated to include all recent changes please? It's hard to search for bugs without that.

[abma]

all changes expect this one: http://springrts.com/mantis/view.php?id=4395 ( LogSections entries must end with a comma or the last entry is not recognized) were gfx related/regressions... if you want more details, use the link "git diff last rc to HEAD".

[Silentwings]

Awesome, thanks.

[Google_Frog]

I have seen desync with this version and the sim stutter problem is still being investigated. Anarchid has investigated the stutter. It turns out that the problem actually exists, occurs for some people and persists with LuaUI disabled. The current theory is that stutter is caused by overheating CPU because this engine seems to run significantly hotter than 91.0.

[Jools]

Why not compare it to 96.0?

[cleanrock]

About high cpu usage / heat:
jk decided to keep threads running for some time (5ms) after having nothing to do, he says it improves performance.
I dislike burning cpu needlessly so I added and use http://springrts.com/wiki/Springsetting ... adSpinTime to be able to avoid this, try:
WorkerThreadSpinTime = 0
WorkerThreadCount = <ur_number_of_cores>

About desync:
I beleive there have been much desync lately with zk and 91.0 so this might just as well be a recent zk change causing desync.

[Google_Frog]

Can they set WorkerThreadSpinTime without needing to set WorkerThreadCount? Also would it be reasonable to try WorkerThreadCount = 1 as a test?

I do not think that the desync is a result of a recent ZK change. You said yourself that some terrain related desync was fixed somewhere around 95.0. I just think that people are better at reporting desyncs or there are more games. I tried large scale reverts upon hearing of the desync and discovered that the desyncs occurred with earlier versions of ZK as well. I think that ZK just uses more features of the engine and has the largest bug reporting user base.

I don't even think desync can be the fault of a game. Games consist of data along side some lua states which should be set up such that they are identical on all computers. If the engine fails to do that then it is an engine problem. If it is impossible to properly isolate a lua state then there should at least be information available on what to not do in lua.

I'm not comparing to 96.0 because it has worse performance than the latest dev engine.

Edit: The desync is not a release blocker for my purposes. From my limited sample it does not seem to occur more frequently than in 91.0 and it is better to use an actively developed and broken engine than one which is similarly broken but has no chance of being fixed.

[Google_Frog]

I have got one person who has the problem to test with WorkerThreadSpinTime = 0. They say that it fixes the stutter problem.

[cleanrock]

Can they set WorkerThreadSpinTime without needing to set WorkerThreadCount? Also would it be reasonable to try WorkerThreadCount = 1 as a test?

WorkerThreadSpinTime and WorkerThreadCount are independent.
If WorkerThreadCount is not set it will default to CORES-1 iirc.
Side note: if you set WorkerThreadCount higher than the number of cores you have spring will crash (i tried it once some time ago).
Only using WorkerThreadCount=1 (or maybe 0) may be a way to see if the thread pool cause stutter, i doubt it though.

[gajop]

[quote="cleanrock"]
Side note: if you set WorkerThreadCount higher than the number of cores you have spring will crash (i tried it once some time ago).
[/quote]
That sounds like a bug..

[malric]

if you set WorkerThreadCount higher than the number of cores you have spring will crash (i tried it once some time ago).

Checked with the latest release candidate on Arch Linux 64bit and the crash still happens. I agree with what Gajop said, this is a bug.

[Masure]

Fixes the AMD/ATI crash for the settings :
3DTrees = 1
AdvMapShading = 1
AdvUnitShading = 1
ReflectiveWater = 4
GroundDecals = 1
Shadows = 1

Thanks

[varikonniemi]

Seems to be working well on ubuntu 14.04 after one remembers to install sdl2 :)

[Jools]

Works well wth xta and windows now as well.

[abma]

next test engine, release candidate for 98.0:

http://springrts.com/phpbb/viewtopic.php?f=12&t=32230

97.0 sadly contained a crashbug at start, so please test harder