New Beta build 70184 is up! [Archive] - Bohemia Interactive Forums

View Full Version : New Beta build 70184 is up!

SASrecon

May 6 2010, 20:04

http://www.arma2.com/beta-patch.php
Wow it seems BIS are working really hard at this!
No news regarding what they updated yet.

Changelog still says the same
[69714] "enablesimulation" fixed on dedicated server
[69683] VoN volume slider
[69599] Optimized: Texture loading done on background thread.
[68990] Improved: Reduced stutter when looking around, esp. with visibility 5 km or more.
Gonna give it a try a.s.a.p

Oh and the Arma2 beta builds 1.05.6xxxx (http://forums.bistudio.com/showthread.php?t=94740) forum should be changed to .7xxxx to avoid new threads :p

Fox '09

May 6 2010, 20:10

hope this fixes some artifacts I get with my graphics card. Introduced with the no-changelog versions...

thhamm

May 6 2010, 20:21

no-changelog versions...

maybe there aren't any changes?

http://en.wikipedia.org/wiki/Placebo

just kidding. :)

otreblA_SNAKE_[ITA]

May 6 2010, 21:13

hahaha xD

AnimalMother92

May 6 2010, 21:16

Yeah the lack of changelogs is creeping me out, I bet BIS is just messing with us :p

darthmuller_cro

May 6 2010, 21:28

;1623324']hahaha xd

hahaaaaa!!! X2!!

---------- Post added at 09:28 PM ---------- Previous post was at 09:26 PM ----------

;1623324']hahaha xD

Comme on Cratian coast to fish some zubatac.... :)

bravo 6

May 6 2010, 21:29

Jesus.. 3 betas in the same weak..

Am i drunk or what ? :)

Good to see betas poping up "every 30 mins"

Edit: Long live BIS!

MavericK96

May 6 2010, 21:35

Yeah the lack of changelogs is creeping me out, I bet BIS is just messing with us :p

I'm almost inclined to agree with this. :P

But whatever, going from 1.05 to 7xxxx has definitely improved my performance and gameplay experience (i.e. much less crashing) so I'm sure there's some fixes in there somewhere.

darthmuller_cro

May 6 2010, 21:38

Yeah the lack of changelogs is creeping me out, I bet BIS is just messing with us :p

New beta patch 69645 may appear not huge based on the change list, but it contains very significant changes under the hood.

As the changes were made to the engine core, including the core memory allocator, extensive testing is badly needed.

Download at http://www.arma2.com/beta-patch.php or directly from ARMA2_Build_69645.zip, feedback welcome.
__________________

MavericK96

May 6 2010, 21:52

One way to determine if anything has been done is to look at the patch filesizes.

Build 69782: 20,627,181 Bytes
Build 70054: 20,631,368 Bytes
Build 70100: 20,631,946 Bytes
Build 70184: 20,505,860 Bytes

Interesting that each new beta up until now was larger than the last, and now this one is smaller. I haven't compared the actual arma2.exe filesizes, someone else can do that. And there's no real guarantee that the difference in size equates to actual useful changes. But it is interesting, nonetheless. :cool:

thhamm

May 6 2010, 21:57

I'm almost inclined to agree with this. :P

But whatever, going from 1.05 to 7xxxx has definitely improved my performance and gameplay experience (i.e. much less crashing) so I'm sure there's some fixes in there somewhere.

same here btw. (so stop the conspiracy theories, all) :p

MavericK96

May 6 2010, 22:00

Well, whatever they did in this new beta actually (objectively) made the game worse for me. After flying past Chernogorsk only a few times, I got a complete CTD with an "Out of Memory" error (never really gotten that before). I also noticed a lot more texture flashing at a distance, and even sometimes the textures in my cockpit (A-10) would flash white for a brief moment.

Here's the relevant portion of the .RPT:

ErrorMessage: Out of memory (requested 3 KB).
footprint 582916896 KB.
pages 16384 KB.

Link to 8a06571b (Obj-202,80:1819) not released
Link to 8d0709d8 (Obj-225,104:472) not released
Link to 8d27233a (Obj-228,105:826) not released
Link to 8d471f05 (Obj-227,106:1797) not released
Link to 8d471f06 (Obj-227,106:1798) not released
Link to 8d471efe (Obj-227,106:1790) not released
Link to 8d471702 (Obj-226,106:1794) not released
Link to 8f4816fc (Obj-258,122:1788) not released
Link to 8f4816fd (Obj-258,122:1789) not released

That top message was actually displayed to me in a window after the game crashed to desktop.

Thr0tt

May 6 2010, 22:26

Doh. ;)

Rexxenexx

May 6 2010, 22:34

Testing on dserver search "Beta SoCal". BE and Sig checks are enabled, also 250 MaxPing. So far so good running Warefare BE lite.

quicksilver67

May 6 2010, 22:37

=SilveR=

May 6 2010, 22:40

Well, whatever they did in this new beta actually (objectively) made the game worse for me.

Agreed.
Just tested performance after new beta-patch. Comparing to previous betas: in Benchmark 01 average framerate drops from 34 to 26 (everything - VeryHigh, except Normal terrain and AA, Low post-effects). After such result I've decided not to launch Benchmark 02 :)
Btw, 1.05.62017 - in Bench01 my rig (Ci7-920@3500, 3x2048Mb 1666Mhz DDR3, GTX285, Win7U x86) usually shows me 38-40 fps.

MavericK96

May 6 2010, 22:48

quicksilver67

May 6 2010, 22:55

Maybe make sure that your shortcut "starts" in the ArmA2 folder, not the ArmA2/beta folder? That's all I can think of.

Naw, can't be that... the directory isn't setup Arma 2/Arma 2/beta... The beta is in the main Arma 2 directory. I'm back running 69782, and it's working fine. Something else has changed. I'm running an Nvidia 250GTS OC with over a gig of onboard GDDR3 RAM and the latest drivers, so it's not a graphics shortfall on my side. I think it has to do with the way the Nvidia card is responding to their shader changes.

MavericK96

May 6 2010, 23:44

Naw, can't be that... the directory isn't setup Arma 2/Arma 2/beta... The beta is in the main Arma 2 directory. I'm back running 69782, and it's working fine. Something else has changed. I'm running an Nvidia 250GTS OC with over a gig of onboard GDDR3 RAM and the latest drivers, so it's not a graphics shortfall on my side. I think it has to do with the way the Nvidia card is responding to their shader changes.

That doesn't make a whole lot of sense, considering I'm using nVidia as well and I don't get the shader error.

What I meant by the shortcut is (example):

Target: "C:\Program Files\Bohemia Interactive\ArmA 2\beta\arma2.exe" -mod=beta -nosplash -world=empty
Start In: "C:\Program Files\Bohemia Interactive\ArmA 2\"

Make sure it is like that and the "start in" is not in \ArmA2\beta

Rexxenexx

May 7 2010, 00:23

GTX285 no probs here.

MavericK96

May 7 2010, 01:35

Okay, just did some testing with the last 6 betas and 1.05, Benchmark 1 only. Semi-interesting results:

Settings:

All settings at Very High unless otherwise noted.
-Post Processing at Low (I never saw a performance hit from turning it up, but I like the way it looks better on Low)
-Video Memory at Default
-View Distance at 3566

1.05: 49 FPS
Build 63826: 48 FPS
Build 69645: 48 FPS
Build 69782: 48 FPS
Build 70054: 46 FPS
Build 70100: 48 FPS
Build 70184: 45 FPS

Notes:

-While 1.05 seems to give the "best" performance, the stuttering is definitely a LOT worse than in some of the other beta builds. Since Benchmark 1 only measures average FPS, the stuttering doesn't really show up in the results. But believe me, it's a lot better in some of the other builds, specifically 70100.

-The latest beta (70184) definitely messed something up performance-wise since the last beta (70100), since it is a full 3 FPS less and the stuttering seemed noticeably worse.

-Overall, the best beta for me so far is 70100. I've gotten no v-buffer creation error crashes, and the performance stuttering-wise is much improved. I did still get a CTD after playing for a couple hours in MP, but it's still an improvement over the other builds, where I'd be lucky to play an hour without some sort of crash/slowdown/failure in the renderer.

nomdeplume

May 7 2010, 01:40

CarlGustaffa

May 7 2010, 01:54

Suggestion for beta system when OA comes out: Let those who use the beta patches know all the changes (that are kept track of internally). Mark each changelog entry with beta if it is not a change compared to the last stable release, and then just strip away all those lines prior to including it with a new stable release.

Anyways, downloading now...

quicksilver67

May 7 2010, 05:17

That doesn't make a whole lot of sense, considering I'm using nVidia as well and I don't get the shader error.

What I meant by the shortcut is (example):

Target: "C:\Program Files\Bohemia Interactive\ArmA 2\beta\arma2.exe" -mod=beta -nosplash -world=empty
Start In: "C:\Program Files\Bohemia Interactive\ArmA 2\"

Make sure it is like that and the "start in" is not in \ArmA2\beta

Ah you can't launch the beta without using the beta folder. If you try to put the " " anywhere else, you get an error message... or at least, you should. My vanilla 1.05 works perfectly. My beta 69782 works perfectly. I didn't change a thing in my init line for the new beta, because they all go in the same place. I'm reporting that now, for whatever reason, my game does not function while running the last three beta updates. It functioned normally with betas before that. It's a shader error, and I know they played around with the shaders, so the question for BIS is: what did you do? The performance dropped off, and some people (me included) are seeing CTDs.. either in game, or prior to launch. Whatever you did, please DO NOT release it as part of the next official patch. In the meantime, I'll experiment with the betas as they are released to see if they start working for me again.

CarlGustaffa

May 7 2010, 06:20

I'm often using simple radio triggers during mission testing etc. Even without addons I'm now getting all unset radio slots (if my trigger is Radio Alpha, that won't be affected) with **** To TRANSLATE. Through the previous last few betas I always thought this was something that ACE was working on, so I haven't reported it anywhere.

Just started some initial tests, and appear pretty smooth so far. One thing I noticed though, some of the AIs that used to follow me around, now instead follows the leader again. Needs to be checked with others, my testing is a bit vague at this point.

MavericK96

May 7 2010, 06:47

Ah you can't launch the beta without using the beta folder. If you try to put the " " anywhere else, you get an error message... or at least, you should. My vanilla 1.05 works perfectly. My beta 69782 works perfectly. I didn't change a thing in my init line for the new beta, because they all go in the same place. I'm reporting that now, for whatever reason, my game does not function while running the last three beta updates. It functioned normally with betas before that. It's a shader error, and I know they played around with the shaders, so the question for BIS is: what did you do? The performance dropped off, and some people (me included) are seeing CTDs.. either in game, or prior to launch. Whatever you did, please DO NOT release it as part of the next official patch. In the meantime, I'll experiment with the betas as they are released to see if they start working for me again.

Well, I just ran the last 6 betas today for my benchmarking all in a row, only deleting the beta folder and then installing each concurrent beta, not changing the beta shortcut on my desktop at all. So whatever you're getting, it's not solely the fault of ArmA2 or even nVidia.

That said, I agree with you about this new beta. The performance is worse and I got an Out of Memory crash fairly quickly with it. So far the best beta for me is still 70100.

Dead3yez

May 7 2010, 10:11

Would be nice to know what exactly we should be testing here.

Are these memory/cpu/gpu related changes or all of them, or what?

Should we be testing in MP/SP Graphics settings or CPU "stress"? Since CPU is usually the bottleneck then shouldn't these be changes relating to this?

Tommyturbo

May 7 2010, 10:38

+1 for 70100

By far the best performance so far even over Vanilla 1.05

Win 32
C2D extreme 3.4 ghz
4g 667 ram
2x 8800 GTX sli

Dead3yez

May 7 2010, 10:47

-While 1.05 seems to give the "best" performance, the stuttering is definitely a LOT worse than in some of the other beta builds. Since Benchmark 1 only measures average FPS, the stuttering doesn't really show up in the results. But believe me, it's a lot better in some of the other builds, specifically 70100

I agree.
Now, I only tested with AI as that is my main concern. Graphics are not really the bottleneck here with FPS.

1.05
Frames, Time (ms), Min, Max, Avg
6803, 131351, 12, 101, 51.793

69782:
Frames, Time (ms), Min, Max, Avg
6668, 134454, 8, 100, 49.593

70184:
Frames, Time (ms), Min, Max, Avg
6637, 135253, 8, 100, 49.071

This is the mission I used to test.
http://www.filefront.com/16382975/AI_FPS_TEST.utes.pbo

win7 x64
AMD 965
gtx260

bensdale

May 7 2010, 11:50

From CIT:

Quote Suma:

Please, try again with 70184 build and attach a report from it. It will not fix the issue, but there are more diagnostics added in it.

http://dev-heaven.net/issues/9679#note-23

otreblA_SNAKE_[ITA]

May 7 2010, 13:51

hahaaaaa!!! X2!!

---------- Post added at 09:28 PM ---------- Previous post was at 09:26 PM ----------

Comme on Cratian coast to fish some zubatac.... :)

Mh? I Didn't understand what you said XD

"Come on Croatian coast to fish some zubatac", maybe?

Why? ò.ò

darthmuller_cro

May 7 2010, 15:59

I was mad yesterday, lost my nerves ... because laggy multiplayer and some other stuff bothering me for a long, long time. I saw one post of you and you was angry to, so i suggested smarter thing to go fishing or something, that was joke of course

Herr_kalashnikov

May 7 2010, 16:35

the fixes they currently working on are probably to obscure to describe in a change-log. The fact the devs are interested in performance changes makes me think its probably in that area they are working on, atleast i hope so

=KCT=BlackMamba

May 7 2010, 17:36

i believe i get more problems rendering certain tree's/bushes and the gras on chernogorsk..

sometimes they stay ugly like im 2000 mtrs away but i aint...

InFireBaptize

May 7 2010, 17:43

i hate those trees, why so many of them?
please in the next patch, shave of those trees, or even better get rid of some. Structures, buildings need optimization as well. Another problem i'm having are incomplete textures especially on vehicle wheels, they show up first as polygons then they get their texture few seconds after. Is it because of my settings:
PP: disabled.
AA: disabled.
AAA: disabled.
Shadow: High.
VD: 1800
VRAM: very high.
Everything else on normal.

Zipper5

May 7 2010, 17:48

Yes, one thing I've noticed is that LODs are taking much longer to load, especially when you start a mission.

otreblA_SNAKE_[ITA]

May 7 2010, 17:50

Kremator

May 7 2010, 18:16

OK have just tested this build with Armamark2. I got a nice 5091, although I am still getting some microstuttering. If only we could load everything into memory there would be no need for hard drive access.

For information I got 5245 from vanilla 1.05 but there was much more texture popping and more microstuttering.

InFireBaptize

May 7 2010, 18:28

Hard drives are cheap. Setup RAID 0, put games on this RAID and Windows on separate drive, with few hard drive tweaks, operating system tweaks and arma2 settings you'll have stutter free game, and don't forget to defrag.

=KCT=BlackMamba

May 7 2010, 19:10

SaOk

May 7 2010, 19:48

These betas have worked really smooth for me, but big missions like warfare in official campaign have bad fps and the more you play those the more laggy those become. Especially when loading a savegame after restarting the whole game, the even worse oscillateing performance is there. I have been creating a big SP mission "Black Forest" for about year (with about 170 scripts and 100 triggers), but I just cant find a fix for bad perfomance. Even when there is no new unit spawning made (and AI units are dieing), the mission could still become more and more laggy. RPT-file have much errors like this if they lead to somewhere:

C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/EastCenter/Groups/Item10/targetKB/Item5.side: Unknown enum value ERROR
C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/EastCenter/Groups/Item11/targetKB/Item5.side: Unknown enum value ERROR
C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/EastCenter/Groups/Item19/targetKB/Item25.side: Unknown enum value ERROR
C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/WestCenter/Groups/Item27/Waypoints/Item3.combat: Unknown enum value ERROR
C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/GuerrilaCenter/Groups/Item5/targetKB/Item5.side: Unknown enum value ERROR
C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/GuerrilaCenter/Groups/Item15/Waypoints/Item6.combat: Unknown enum value ERROR
Saving undefined enum value -2147483648 / 10, context /EastCenter/Groups/Item10/targetKB/Item5.side
Saving undefined enum value -2147483648 / 10, context /EastCenter/Groups/Item11/targetKB/Item5.side
Saving undefined enum value -2147483648 / 10, context /EastCenter/Groups/Item19/targetKB/Item25.side
Saving undefined enum value -2147483648 / 6, context /WestCenter/Groups/Item27/Waypoints/Item3.combat
Saving undefined enum value -2147483648 / 10, context /GuerrilaCenter/Groups/Item5/targetKB/Item5.side
Saving undefined enum value -2147483648 / 6, context /GuerrilaCenter/Groups/Item15/Waypoints/Item6.combat
Error: post effect with priority 250 already exist.
Error: post effect with priority 450 already exist.
Error: post effect with priority 250 already exist.
Error: post effect with priority 450 already exist.

C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/Vehicles/VisibleFar/Vehicles/Item13/Brain.combatModeMajor: Unknown enum value ERROR
C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/Vehicles/VisibleFar/Vehicles/Item133/Brain.combatModeMajor: Unknown enum value ERROR
C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/EastCenter/Groups/Item10/targetKB/Item5.side: Unknown enum value ERROR
C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/EastCenter/Groups/Item11/targetKB/Item5.side: Unknown enum value ERROR
C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/EastCenter/Groups/Item19/targetKB/Item25.side: Unknown enum value ERROR
C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/WestCenter/Groups/Item27/Waypoints/Item3.combat: Unknown enum value ERROR
C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/GuerrilaCenter/Groups/Item5/targetKB/Item5.side: Unknown enum value ERROR
C:\Users\Santtu\Documents\ArmA 2 Other Profiles\SaOk\Saved\missions\SPBlackForest.Chernar us\continue.ArmA2Save/GuerrilaCenter/Groups/Item15/Waypoints/Item6.combat: Unknown enum value ERROR
Id Veh-1099660 not found
Id Veh-1099661 not found
Id Veh-1099683 not found
Id Veh-1099684 not found
Id Veh-1099708 not found
Id Veh-1099711 not found
Id Veh-1099712 not found
Id Veh-1099713 not found
Id Veh-1099714 not found
Id Veh-1099722 not found
Id Veh-1099747 not found
Id Veh-1099855 not found
Id Veh-1099945 not found
Id Veh-1100014 not found
Id Veh-1100015 not found
Id Veh-1100047 not found
Id Veh-1100048 not found
Id Veh-1100049 not found
Id Veh-1100050 not found
Id Veh-1100051 not found
Id Veh-1100052 not found
Id Veh-1100053 not found
Id Veh-1100054 not found
Id Veh-1100094 not found
Id Veh-1100377 not found
Id Veh-1100378 not found
Id Veh-1100476 not found
Id Veh-1100477 not found
Id Veh-1100478 not found
Id Veh-1100479 not found
Id Veh-1100480 not found
Id Veh-1100481 not found
Id Veh-1100482 not found

Saving undefined enum value -2147483648 / 6, context /Vehicles/VisibleFar/Vehicles/Item13/Brain.combatModeMajor
Saving undefined enum value -2147483648 / 6, context /Vehicles/VisibleFar/Vehicles/Item133/Brain.combatModeMajor
Saving undefined enum value -2147483648 / 10, context /EastCenter/Groups/Item10/targetKB/Item5.side
Saving undefined enum value -2147483648 / 10, context /EastCenter/Groups/Item11/targetKB/Item5.side
Saving undefined enum value -2147483648 / 10, context /EastCenter/Groups/Item19/targetKB/Item25.side
Saving undefined enum value -2147483648 / 6, context /WestCenter/Groups/Item27/Waypoints/Item3.combat
Saving undefined enum value -2147483648 / 10, context /GuerrilaCenter/Groups/Item5/targetKB/Item5.side
Saving undefined enum value -2147483648 / 6, context /GuerrilaCenter/Groups/Item15/Waypoints/Item6.combat
Saving undefined enum value -2147483648 / 6, context /Vehicles/VisibleFar/Vehicles/Item13/Brain.combatModeMajor
Saving undefined enum value -2147483648 / 6, context /Vehicles/VisibleFar/Vehicles/Item133/Brain.combatModeMajor
Saving undefined enum value -2147483648 / 10, context /EastCenter/Groups/Item10/targetKB/Item5.side
Saving undefined enum value -2147483648 / 10, context /EastCenter/Groups/Item11/targetKB/Item5.side
Saving undefined enum value -2147483648 / 10, context /EastCenter/Groups/Item19/targetKB/Item25.side
Saving undefined enum value -2147483648 / 6, context /WestCenter/Groups/Item27/Waypoints/Item3.combat
Saving undefined enum value -2147483648 / 10, context /GuerrilaCenter/Groups/Item5/targetKB/Item5.side
Saving undefined enum value -2147483648 / 6, context /GuerrilaCenter/Groups/Item15/Waypoints/Item6.combat

There is much "Id Veh-1082388 not found"-style errors like 100 of them in row from 1082388 to 1089695.

Edit: I made two bug reports (with sample missions) to dev-heaven about buggy FPS loss with AI.

Deleted units bring only about 1/3 of FPS back:
http://dev-heaven.net/issues/10544
A dead AI unit takes as much FPS away as an alive AI unit:
http://dev-heaven.net/issues/10542

Protegimus

May 7 2010, 20:20

bensdale: thanks for the heads up about the CIT. Great to see BIS so actively following up on difficult to diagnose problems, plus the insight into what else is going on with the betas.

There appear to be changes to the shaders between beta builds 63826 through 70100, haven't compared 70184 yet. 70054 stands out with noticeably lower level of detail on trees.

[70100 seems to give the best results - amended to: 70100 has noticeably more stutter than current release and previous betas, this is immediately obvious in the demo worlds. Also, anti-aliasing looks worse in 70100].

70184 I observed momentarily missing cockpit texture for AH-64 and other bits of vehicles missing/popping in.

-- Amended 09.05.2010--
Now that I've had more time on 70184 I see it is quite slow to load the higher detail LOD's. Overall this isn't such a bad thing as it helps reduce some of the LOD thrashing that we typically see - some hysterisis in the switching algorithm seems to be needed - but the orange bushes/trees in particular look a bit blobby at medium distance.
70184 also has stutter about once every 3s, just not to the same degree as 70100. Running from a RAM drive or a hard disk has no impact on this.
In summary, I the reduction in stutter makes 70184 easier to live with than 70100.
--

From profiling the various betas I have observed the following:
Performance bottleneck appears to be the main execution thread - this continuously saturates a single core from a multi-core CPU.
The thread appears to be doing setup of graphical data, as the only time its utilisation falls is between loading screens when there is little graphical rendering going on. The graphical work is accounting for around 58% of the workload.
Performance scales directly with MHz too - the higher I push the core speed the better overall performance is, utilisation of the core stays at > 98% irrespective of clock speed, hence the deduction that this is a bottleneck.

Secondary thread appears to be doing the data streaming. I say this as it spends a lot of time in the Windows kernel and I understand this is typical of threads doing lots of I/O.

Remaining threads of the 23 total are all < 1%.

70184 looks to have slightly higher utilisation of the secondary I/O thread, around 0.5% above 70100.

Until BIS are able to move some [more] of the workload out of the main thread, I believe we are going to stay CPU limited. I expect this isn't easy however, or I've no doubt it would be done already.

This is all based purely on observation of course, I have no internal information or debug symbols for the software, however it may help get a better understanding of what is limiting performance.

Following feedback from the developers about how the wide variety of PC hardware causes headaches, a suggestion to BIS is to give up to date recommendations on preferred hardware.
A lot of guys are spending a lot of money to improve their experience with ArmA, not all of it bearing fruit.
If BIS could state we developed or tested with CPU A on motherboard based on chipset B with graphics card C or D and hard disk/RAM drive/SSD E with sound card F running O/S G with the following recommended in game settings(!) with some information, then I expect many would follow the guidelines to ensure the best performance and reduce headaches.

Protegimus

froggyluv

May 7 2010, 20:53

Still no luck here

I've tried all the recent patches briefly and mentioned that they literally bring my computer to a slideshow or alt/cntrl/delete esacpe.

Specs: e8500 stock 3.2ghz
ATI 4870 512mb stock settings/latest drivers
8 gigs 800 DDR2 Ram
250 Sata HD just defragged
Vista 64

ArmaMark Results with recent Beta:

Test1: 8.6;10.8;11.9;11.1;14.4 = 1140 took 8 minutes 27 seconds

Test2: 8.5;13.9;17.5;11.2;23 = 1487 took 7 minutes 31 seconds

ArmaMark Results 1.05:

Test1: 22.3;34;26.5;35.7;13.8 = 2652 took 2 Minutes 38 seconds

Test2: 27.3;34;29.7;42.7;28.2 = 3359 took 2 minutes 20 seconds

All settings on normal/AA disabled. Beta folder was deleted before installing.

Am I doing something wrong here? The performance is just awful with 2-3 second delays for mouse input and the title screen stutters every frame as well as music. Do I need to tweak a config or add some parameter as it seems many people only have minor performance issues but this has been my case with every recent patch.

:(

JW Custom

May 7 2010, 21:09

70184 I observed momentarily missing cockpit texture for AH-64 and other bits of vehicles missing/popping in. Not that much time on 70184 though, so they are initial impressions.

I have experienced that flashing cockpit in air vehicles since very first version of ArmA 2.

Rexxenexx

May 7 2010, 21:22

If their was a tree density setting it would have to be set at mission start only since they are objects with locations that everyone has to be able to see and interact with. That would cut down on performance hit. But I think trees are part of the map itself so a map maker would have to make a new "light" version of say Cherno. Right??

quicksilver67

May 7 2010, 23:05

Feedback deleted. Nice.

MavericK96

May 7 2010, 23:28

ArmaMark Results with recent Beta:

Test1: 8.6;10.8;11.9;11.1;14.4 = 1140 took 8 minutes 27 seconds

Test2: 8.5;13.9;17.5;11.2;23 = 1487 took 7 minutes 31 seconds

ArmaMark Results 1.05:

Test1: 22.3;34;26.5;35.7;13.8 = 2652 took 2 Minutes 38 seconds

Test2: 27.3;34;29.7;42.7;28.2 = 3359 took 2 minutes 20 seconds

All settings on normal/AA disabled. Beta folder was deleted before installing.

I can't comment on 70184, because I got crashes with it so I reverted back to 70100, but I just did some ArmAMark with 1.05 and 70100, and the results are pretty significant:

1920x1200, All settings at Very High, except Post Processing at Low and Video Memory at Default

First runs only:

1.05 Vanilla: 57.6 Build 70100: 56.4
36.8 37.9
33.3 35.9
47.9 53.2
13.6 45.5
Score: 3783.46 4577.32

Obviously, 70100 is doing something right over 1.05. Especially in Test 5, where I'm guessing the new feature that eliminates grass at high altitudes is at work here.

It's also interesting to note that I ran ArmAMark in 1.05 before I got my SSD, and my results were 2728.19, and that wasn't even with everything on Very High.

KeyCat

May 8 2010, 04:53

FWIW, after a quick spin with build 70184 here are my experience...

The stuttering seems to improved a bit, it's not as smooth as in 1.05 final but better than in build 70100 on my rig. Now it's playable if I set Terrain Detail = Low but still quite evident stuttering if I set my Terrain Detail = High/Very High (wich is what I use without problems in 1.05 final).

Windows XP Pro
Intel C2D E6850 @ 3.41 GHz
2 GB RAM @ 1066 MHz
2 x Seagate ST3250620AS SATA-II 250 GB HDD
GeForce GTX 260 896 MB with drivers 191.07 WHQL

As previous this simple test was done on Utes with me as US infantry running in the grass alongside the runnway, no other units/scripts/addons running and frametrate is 30+ FPS.

Hope to get some more time testing later this weekend...

/KC

Rexxenexx

May 8 2010, 06:00

MavericK96

May 8 2010, 07:59

All settings Very High except PP=Low Shadows=Normal Res=1920X1200

v1.05 = 1508
beta build70100 = 1981
beta build70184 = 1917

Build 70100 is very very fast at loading the textures on the airstrip flyby. Both 1.05 and 70184 didn't load the strips texture until the end after pausing a bit. Hope that helps.

Yeah, 70100 is definitely the best one out so far for me, subjectively (stutter) and objectively (FPS/crashing). Still has some issues but is by far an improvement over 1.05 and 70184.

Thr0tt

May 8 2010, 09:57

I have experienced that flashing cockpit in air vehicles since very first version of ArmA 2.

Me too, I thought I had just got shot down as the transition between good and damaged is not the smoothest but not its a sort of flash and everything is fine. It did appear to happen when light was changing i.e. flying towards sun then turning.

Univ

May 9 2010, 06:47

With all the 70000 beta builds I experience *heavy* stuttering after a while of gameplay (1h of warfare for ex). FPS suddenly drops from 30 to 5.
Win7 64bits and ati 4850.

Alex72

May 9 2010, 07:15

Am I doing something wrong here? The performance is just awful with 2-3 second delays for mouse input and the title screen stutters every frame as well as music. Do I need to tweak a config or add some parameter as it seems many people only have minor performance issues but this has been my case with every recent patch.

:(
It does sounds like somethings wrong on your end. Never saw such a big difference in the betas. Are you sure you start the beta's properly? I remember i had the same problem like you at some point in time and it was me starting it wrong. Do you start the "launch arma2 beta" shortcut? And do you have the startup parameter in that shortcut "-mod=beta"?

=KCT=BlackMamba

May 9 2010, 08:13

the only problem im strugling with is the brown trees/bushes and grass...
when im playing a single player mission sometimes its realy good but when the armored vehicles come up, the trees and grass start looking like im 2000 mtrs away.
when they leave or are taken down the grass and trees start looking bether again.

Wolfrug

May 9 2010, 20:23

Still getting the same Pixel shading error with this supposedly fixed version of the beta. ("Error compiling pixel shader PSSpecularAlpha:0" I have not gotten this error with any other beta (and I've been on board for more than a dozen) except these last two, EVER. My init. line is as follows:

"D:\ArmA 2\beta\arma2.exe" -mod=beta

I am now getting this same error when trying to start the latest beta (did not have it with any of the previous ones). GeForce 8600 GT 512 MB, Win XP 32 bit, latest NVidia drivers. Shortcut the same unchanged one I've been using for ages; vanilla (no mods). What could be the problem?

Edit: here's my arma.rpt for the event:

================================================== ===================
== K:\games\Arma 2\beta\arma2.exe
== "K:\games\Arma 2\beta\arma2.exe" -nosplash -world=Chernarus
================================================== ===================
Exe timestamp: 2010/05/09 19:08:58
Current time: 2010/05/09 23:39:24

Cannot register unknown string STR_INCOMPATIBLE_LOAD_GAME_ATTEMPT
Error: 80004005 in D3DXCompileShader while compiling the PSSpecularAlpha:0 pixel shader (w:\c\poseidon\lib\d3d9\ShaderSources_3_0\PSBasic. hlsl)
error X3501: 'PSSpecularAlpha': entrypoint not found

ErrorMessage: Error compiling pixel shader PSSpecularAlpha:0

Regards,
Wolfrug

Yoma

May 9 2010, 21:23

I am now getting this same error when trying to start the latest beta (did not have it with any of the previous ones). GeForce 8600 GT 512 MB, Win XP 32 bit, latest NVidia drivers. Shortcut the same unchanged one I've been using for ages; vanilla (no mods). What could be the problem?

Edit: here's my arma.rpt for the event:
Regards,
Wolfrug

Most likely your "start in" folder is not set correctly.

NoRailgunner

May 9 2010, 21:52

Got following arma2.rpt lines:

Performance warning: Very large search for 376454 (>300 m)
Performance warning: Search for 35315000# 376454: sloup_vn.p3d was very large (390 m)
Action::Process - No target
Bad action name agonyCargo - not found in CargoActions
Unknown action Surrender
Unknown action GetOver
Error: cannot load sound 'ca\dubbing\global\radio\...
AI is still not able to break from contact and to move quickly to another place or get in a vehicle.
What about making the "Disengage" command more stricter for AI?

AI medics are still sadistic and let wounded die - even if they are within 0-5m close range.
What about getting the AI next to wounded units do cover them + medic?
In that way the AI medics could use a more direct way and heal wounded instead of going around them.

What about making the AI move faster:
if there are no enemys within a certain range
and
if enemys are retreating/running away?

Himmelsfeuer

May 9 2010, 22:38

Im signing that and hope strongly this points get fixed BEFORE Arrowhead..!

Geraldus

May 10 2010, 00:19

Steakslim

May 10 2010, 01:14

I got a single crash with this latest beta after some hours of gameplay in the campaign right after an autosave without showing any errors. It was like i "alt + tab" cause the game was still active in task manager, but not showing really.

If you can still hear audio in the game, doing the ctrl+shift+ -NUM and type Flush.

It "sometimes" gives you your screen back.

nomdeplume

May 10 2010, 01:53

Wolfrug

May 10 2010, 07:11

You're missing the -mod=beta from your shortcut.

D'oh! :D Yeah, that fixed it. I don't know when or how I removed that, actually, stupid me. Starts now. Never mind!

Regards,
Wolfrug

CarlGustaffa

May 10 2010, 08:15

Is safe mode (setBehaviour) acting "weird" for any of you guys? In safe, my guys (in my squad) run walk and run around like maniacs, like their trying to find their formation point but never get there. SetFormation doesn't seem to do anything either.

Can be a change in the system to show how safe they feel compared to aware mode, but at the airport at Utes, I'm getting a lot of these errors in my rpt:

Out of path-planning region for B 1-1-A:2 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:4 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:4 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:4 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:4 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:2 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3878.8,3600.0, node type
Out of path-planning region for B 1-1-A:5 at 3880.0,3600.0, node type
Out of path-planning region for B 1-1-A:3 at 3880.0,3600.0, node type

It may take a while for the rpt to start filling up. Was running @CBA at the time though.

nomdeplume

May 10 2010, 10:20

Formations don't apply in Safe/Careless mode, they always move in column formation. Has been that way since forever.

=KCT=BlackMamba

May 10 2010, 16:56

HAA!! new beta

ftp://downloads.bistudio.com/arma2.com/update/beta/ARMA2_Build_70256.zip

megagoth1702

Aug 10 2010, 16:50

Question - AI vision affected by light sources.

Can someone explain this with a simple example? "If you do X, Y will happen" - somethin like that.

Thanks in advance!

MadDogX

Aug 10 2010, 18:42

Question - AI vision affected by light sources.

Can someone explain this with a simple example? "If you do X, Y will happen" - somethin like that.

Thanks in advance!
If you stand under a lit street lamp at night, a nearby AI can see you.

This wasn't the case before.

TRexian

Aug 10 2010, 18:45

I wonder if there's a way to measure the amount of light now, and have a Splinter Cell type device? :)

CarlGustaffa

Aug 10 2010, 20:15

If you create a #lightpoint object, AIs vision will be affected. If you create a flare, AIs vision does not appear to be affected.

megagoth1702

Aug 11 2010, 00:41

If you stand under a lit street lamp at night, a nearby AI can see you.

This wasn't the case before.

Roger , thanks.

MadDogX

Aug 11 2010, 13:32

I wonder if there's a way to measure the amount of light now, and have a Splinter Cell type device? :)
That would be a pretty cool command. Call it illumination or something and pass a 3D coordinate, like this:

_lightVal = illumination [x,y,z]

Return value should be a float number from 0 to 1, where 0 is completely dark and 1 is fully lit. I guess during the day it would always return 1, unless it could somehow take shadows into account. (doubtful)

inf25th

Aug 13 2010, 19:15

I was having Out of Memory and CTD problems on Multi-player games. I was instructed by BI in another thread to install the latest beta build which was ARMA2_OA_Build_72418. This fixed the Memory and CTD problem but now I am having a problem with the game freezing for about one second every 3 or 4 seconds. It will run for about 3 or 4 seconds and then just freeze for about 1 second over and over. I tried installing the latest ARMA2_OA_Build_72716 but I still have the same problem.. fps during the 3 or 4 seconds of run time is 50/60 fps and during the freeze it drops to 29/30 fps.

Game installed --->ARMA II OA - CO

My Rig
Alienware Aurora
CPU: Intel(R) Core(TM) i7 CPU 975 @ 3.33GHz
Video: NVIDIA GeForce GTX 260
Video_AdapterRAM: 1792MB
12 GB Ram - DDR3 (533 MHz) 1068MHz
Win 7 (64)
25 meg cable internet (just tested at 27 meg)

When I run the game without the beta build, it runs very smooth but of course I have the Memory and CTD issue.

Any Ideas??

=KCT=BlackMamba

Aug 14 2010, 05:56

maybe u can try the new beta 72716

LOG:
[72701] New: getWeaponCargo, getMagazineCargo, getBackpackCargo script commands
[72698] Experimental: TCMalloc used as a memory allocator.

maybe this solves your stuttering ?

inf25th

Aug 14 2010, 16:13

I tried installing the latest ARMA2_OA_Build_72716 but I still have the same problem..?

I already tried that.