Ambal

No, this is just a small addition to that patch I'll see what I can do to provide a git branch for everyone interested...

Hi everyone. Want to present you another performance optimization patch. Idea is very old, but benefits are huge. Currently when player stands on map, he stands in map cell too. Length of cell side is 133 yards. Also player has 8 cells around him. So current algorithms for cell object search (for example AOE spells) must look for mobs in 9(!) cells. Lets say, our AOE damage affects only targets in 30 yards radius. So we have a huge save in CPU power because with so small radius we would never need to search all 9 cells, maximum 4 of them This patch affect alot of places: chat whispering, visibility calculations, AOE target search, mob assistance, area packet sending etc... P.S. Current cell walk algorithm in CellImpl.h was taken from Trinity 1, I don't know who author is. But next versions will include more generic calculations from mangos dev team, which will work with any cell sizes and radiuses. Have fun Update #1: Updated cell search algorithm, cell size reduced to 66 yards. Also fixed a crash with fishing... Update #2: Another improvements to cell search algorithm. Also packet sending improved to scale with visibility distances. New visible distance options included into server config. MAX_NUMBER_OF_CELLS set to 16 <- this is experimental value. Report any problems with it asap. Update #3: Fixed some compilation issues with GCC. Restored possibility to use 'reload config' command to reinitialize visibility distance for each map w/o need to restart server. Update #4: Fixed VisitCircle() algorithm. Now it is more precise. Update #5: Updated cell coord range checking. Patch supports revs 8182+. Update #6: Updated grid unload algorithm. Added more security checks for grid/cell coords range checking... Update #7: Fixed a bug with visibility distance in dungeons/BG/Arenas - it was always the same as on continents until you used .reload config command... Patch 1.14: http://filebeam.com/6cc48d245ee183b5e5df7fca8894f9cd Patch 1.13[rev 8277+]: http://filebeam.com/0f565f9494b0869942a4ab47fd1e715c Patch 1.13: http://filebeam.com/d34652aada593787748884a7e26b9972 Patch 1.12: http://filebeam.com/39ba2fc9a850e686fd3a907135397f15 Patch 1.10: http://filebeam.com/2c9fb21627cb9ee7c639fad863223ffd Patch 1.09: http://filebeam.com/8b637b44f86831d3bca7669735f071c6 Patch 1.08: http://filebeam.com/c47df9de78907324e7346ce804588623 Patch 1.01: http://filebeam.com/a4e5d66735a33db959dbc9d4c8f9cac2 Patch 1.00: http://filebeam.com/e1389b2727e0e021f23d1f4a64e11563

Any hint what causes this crash? :huh: Or this is a new one that you didn't have earlier? Do you use Derex's mtmaps patch and do you have similar crashes? P.S. Infinity said holy truth about having 2700 ppl on dual-core CPU with ~160-170% utilization, I trust him because he tested my GetMap patch for more than 3 weeks. That Relocate patch is really awesome, we (me and SilverIce) just found a huge bottleneck in server performance. I have to add, that Trinity 1 used similar technique as we found out later, but it was much simpler.

You should turn VMAPS off if you want to get max performance from your server

Well, this is almost fixed Currently I and SilverIce are working on patch, which will allow core to scale like hell with mtmaps... Stay tuned

You wouldn't save that much traffic by setting compression level to '9'. Use '1', mangos will definitely find use for free CPU resources

Patch updated to rev 8103 since part of a patch already in GIT. First post updated.

Updated patch for rev 8099+. First post updated

Why would this patch need a GIT branch if it'll get into GIT repo in few days? :rolleyes:

You can apply mtmaps patch w/o any problems with this one

divisorxx, thank you very much for your response I'm glad that patch is working. Have you noticed any stability improvements?

Yes, just rename .txt file into .patch/.diff. Filebeam doesn't allow patch/diff files to be hosted so I had to share a txt one. I will not make any git branches just because on getmangos.ru noone reported any bugs with latest 1.10 version... Also if you use KAPATEJIb binaries, this patch is already there.

It would be really nice to get patch feedback finally after almost a month of development

Well, I'd say 'prepare to be banned' :rolleyes: Mangos is Mangos. If you want to call things in your way - go and organize your own project

Anticheat helped to track and fix mega-bug in delayed teleports code. Big thanks to KAPATEJIb and Wowka321 for bug reports First post updated.

Fixed AchievementMgr.cpp compilation. Added additional IsInWorld() checks into WorldSession::HandleMovementOpcodes(). First post updated.

Latest patch for rev 8071 uploaded.

huh, I doubt I can help with such an old rev Tomorrow I'll upload a new patch version where I suppose these bugs will be fixed. Thanks P.S. Wowka321, those callstacks are very useful

This small bug in this function was unnoticed in clean sources just because it is called from very specific BG's code. But with new map system patch this func is really a PITA if not fixed.

Thank you. But next time, please, post patches inside appropriate forum section so they can be noticed easily

This one should be moved to "Rejected patches", because it doesn't solve a problem as it should be solved and was reverted in repository.

Great news Latest patch uploaded and first post updated. Have fun

Thank you, similar patch was suggested on getmangos.ru

Patch updated for 8039 rev. Crash with 'Deserter' spell cast should be fixed

Thanks, Wowka321, I'm glad that fix worked I saw a crashdump - server died when player casted spell 'Deserter' while being far teleported e.g. not-in-world (NIW). I'll suggest a fix in nearest time.