qwertz

Let me answer that for Maximus: Yes, everything else works fine. Here's a screen of the key folder. I have heard that there were issues with the Vopsound 2.1 package, specifically that the package does not include the updated keys, and that using the 2.0 bikey would help - might this be the issue ?

It seems to have disappeared from all sites (gamespy, servertracker, http://arma2.swec.se/server/data/120628). Instead, another server ("EDBA Server") is popping up under the same address (IP Address: 72.249.94.58 Port: 2302). Here's what Gametracker shows: Yesterday: Gametracker Today: Gametracker Anyone knows anything ?

Server is up again, on a different IP

(double post, sorry)

Hey Drew - first of all, after reading my original response again, I feel that it might have been a little harsh - sorry if you have taken any offense. My point is not that the tool is not actually shutting down services. My point is that the promises it makes on the website are marketing mumbo jumbo. Let's see: "It works by defragmenting game directories, temporarily shutting down background processes, cleaning RAM, and intensifying processor performance." Defragmenting game directories: Windows can do that pretty well, as well as any standard defrag program. In any case, games do not tend to write/change a lot of data to/on the HD once installed, so there is no extraordinary level of fragmenting going on that would not be taken care of by your monthly defrag run. Temporarily shutting down background processes: Here's a little test: Do a fresh boot, open your task manager, make sure you enable the columns "CPU Time" and "Memory - Peak Working Set" (-> View -> Select Columns), and "Show Processes from all users". Take a screenshot of your task manager. Keep the task manager open, and start Arma 2. Play for 30 minutes or so. Go back to your taskmanager, and compare both the CPU time and the peak working set with the ones in your screenshot. You will find the following: 1) All your processes together (excluding Arma) will only have consumed a few seconds of your processor time. Thats less than 0.1% of your total processor time. Which is the maximum performance gain you could possibly expect if you would be able to disable all processes other than Arma2.exe (which you can't). 2) If you want to have more detail, right click any of the svchost.exe entries in your task manager, then click "go to services". You will now see all services that are loaded into memory via DLLs through that specific service host process. Don't be suprised if each of the service host processes runs 5-20 services and displays 0 (that is, less than 1 second) total processor time. 3) Also, you will see that the peak working set of any process will not even come close to maxing out your memory (that is, if you have more than 2-3GB physical RAM). Windows automatically gives foreground processes the highest priority both in term of CPU and RAM allocation. "Cleaning RAM" There is no plausible reason why you would want to "clean RAM" (what ever that means, they do not even bother to detail that). RAM is best used by being fully utilized. Free RAM means wasting RAM. In case you will ever run out of RAM (which barely happens on modern rigs with multiple GBs if you are not running specific progs such as VMs or Photo/Video editing in the background while gaming), Windows is pretty good at prioritizing which data gets swapped to disk, which is based upon pretty clever algorithms. "Freeing" or "Cleaning" ram means that a dumb program just brute forces all data that it deems (for whatever reason) unnecessary from RAM to disk. This causes disk writes and costs performance, because Windows will load this again into RAM whenever its own memory manager thinks it needs to be there. The point is, the only effect of "cleaning RAM" is that you see more unused RAM in task manager for a short time and feel good about for the wrong reasons. Intensifying processor performance: I have no idea what the heck this should be. Maybe they mean giving the game a higher process priority, which again is already automatically done by Windows -> snake oil. One doesn't have to test all magic fuel additives on the market that promise 20% more horsepower to know that its snake oil, if you know how a modern engine works. Or, you don't have to buy & read the newest book from the other TV preacher to know that it will not make you do wonders, even if it claims so. That was my whole point. I am not saying that it is impossible that at some point in time some genius creates a tool that can improve windows process performance, but I would really like to see an explanation HOW (which is lacking here) and HOW MUCH (aka benchmarks, which are suspiciously completely lacking here as well). All there is is marketing gibberish.

DUDE, you got to calm down, seriously. I have not attacked you personally, just have my own thoughts about the magic program you posted that claims to to some really amazing stuff ("gives more CPU and RAM to your games" - lets see, how exactly is that supposed to work? ). Your typical 32 bit game (including Arma 2) can in any case only address 2GB in total, no matter how much you "clean up" your RAM. On the other hand, installing some magic program that promises some amazing stuff but does not explain how is negligent at best, at least in my view, and yours might differ. This company seems to got caught with its pants down on unethical business practices, and just because they now claim "It wasn't me" on their own blog does not automatically clear them off the serious evidence presented. But hey, thats just me. Besides the above, lets assume all the software does is disabling some windows services. Ok. Disabling services does not generate any perceivable benefits on any half way modern (and healthy) system, this has been discussed over and over. I have not found a single review of this program that would prove any effect by posting any benchmarks (but I stopped looking on google page 10, I admit). Why is that? It is no surprise, though, because no other program that claimed the same in the past has passed any professional benchmark review neither, so nothing new here. Have a look here or here or here for starters. The program itself claims that its "best part" is that "It's 100% free and safe!". That's at least something. But hey, to each his own. At least - I quote from their website - "Its intuitive interface makes Game Booster the perfect tool for complete computer dummies." -> Enough said. Peace.

Sorry, I am still extremely skeptical. Reading their web sites gets all my red flags up. None of the features listed on their site do anything to your performance (if you count your RAM in Gigabytes and not Megabytes) and have been proven to be irrelevant for quite some time. At best this is BS/snakeoil. If it does something that they don't list on their site, I would be even more worried. The company IOBit is a Chinese company that has been accused of shady business practices in the past: http://www.malwarebytes.org/forums/index.php?showtopic=29681 I my 20 years of being a PC power user, I have yet to see a single program that can boost a computers performance under normal conditions. Usually this is just noob scamming, and the fact that this program does not cost anything and is "advertised" a lot in forums does not make me feel better.

Happy to do it when I find some time. I would need you to post the file list/structure of Arma 1 though. Could you please post the content of the folder for the PBO files on a fresh Arma 1 install?

I guess it's a chicken-egg problem. I know of people that do love ArmA 2 but have stopped playing because there is a lack of public servers that combine PvP with a somewhat coordinated gameplay (which is missing on the crowded berzerk etc. servers). Not everyone has the time to get "serious" and join a clan etc. , which is usually the solution to the problem. The only location in the US that offers all the above is the BMF server as far as I know, thats why I almost exclusively play there (it only gets full after 9-10pm Eastern time, though). I think if Tom_Anger would pull this off, with a little word of mouth, there should be enough ppl that would be happy to jump on board, at least over time. Tom_Anger, if/when you pull this off, let me know if you need help, I don't have too much free time but would be happy to contribute to the funding.

Based upon the graphs here, it is borderline - it is certainly faster than HDD, and looks great on anything 64k and below, but random read is only at 179 IOPS and this might be tight. I guess someone has to try that out and post the results.

I have an i7 @ 4 GHz on a EVGA Classified MB and a GTX280HC. I have tested ArmA 2 on XP32, XP x64, Vista x64, Win7 X64 - all fresh installs with newest drivers. XP32 was leading the pack in terms of performance by quite a margin.

The same that happened to the happy customers that love Win 3.11 so much that they never upgraded to Win 95: They are bitten by the dog named "anachronism". :D

Whilst we don't know for sure and are really only guessing, I think that that there is some evidence that points to a scenario that would scream for 64bit addressing. (note: 64bit PROCESS on 64bit OS) [GUESS MODE ON] As hypothized earlier (and thanks to oktane, somewhat verified in his code review), Arma2.exe does not use system level Windows file caching (which is a place in RAM that buffers reads and writes to the filesystem). The likely reason is that the Windows cache would only cover for a very small fraction of the assets needed by this game, effectively rendering the cache useless (or even counterproductive in some way). The effect is, that Arma2.exe is now forced to stream directly from disk, which in turn (and based on filesystem I/O performance) causes stutter. Of course, any application can create and use its own proprietary buffer implementation (i.e. reserving a large part of available ram and prefetching stuff there). However, as a 32bit application (and without PAE/Largeadressaware), Arma2.exe can only adress 2GB of ram in total for everything including the game code. Therefore, any proprietary prefetching would again only be available for a small fraction of the game assets, with the same problems as described above. Now, with 64 bit, on a 8 GB machine, the engine could take 6GB as a cache, would prefetch all assets it needs for a certain escenario dynamically, and would keep like 1GB for the OS and another 1GB for program code. This would likely remove the bottleneck that we are discussing here. [/GUESS MODE OFF]

Yep, but it seems like the arma engine would specifically benefit from being able to allocate enough ram to cache/prefetch the game assets so it does not have to stream it from disk in real time. This would only work with 64bit. This is different from the effect of 64bit on CPU/GPU/Ram performance. ---------- Post added at 07:24 PM ---------- Previous post was at 07:17 PM ---------- Hi Frag85, received your PM. I have tried to convert if to Office 2003 format, but the main problem (other than the adaption of some functions) is that pre-2007 office versions are limited to 7 nesting levels in formulas - I have used some lengthy formulas to extract/parse the information from the data output of ProcMon. I can try to find more elegant/less nested solutions for this, but this needs some major rework - Excel is really a poor tool for this kind of work, but it is the only tool I am familiar with:-) Will see what I can do - how many people need/want an Office 2003 version ? It seems that this formula is the main culprit - it is extracting the correct PBO name from the ProcMon log, taking into account that path names are different from user to user - if anyone has a simpler solution, please let me know! E:\Steam\steamapps\common\arma 2\AddOns\anims.pbo =IF(ISBLANK('2 - Data Input'!F9),"",IF(COUNTIF($L$2:$BI$2,TRIM(MID('2 - Data Input'!F9,(FIND("*",SUBSTITUTE('2 - Data Input'!F9,"\","*",LEN('2 - Data Input'!F9)-LEN(SUBSTITUTE('2 - Data Input'!F9,"\","")))))+1,100)))=0,"OTHER PBOs",TRIM(MID('2 - Data Input'!F9,(FIND("*",SUBSTITUTE('2 - Data Input'!F9,"\","*",LEN('2 - Data Input'!F9)-LEN(SUBSTITUTE('2 - Data Input'!F9,"\","")))))+1,100))))

Hi Toaster, could you upload the full tool file so I can have a look at the raw data and try to find the reason for this result? It seems like putting only a part of the files on a ramdisk is a hit and miss - maybe we can find the reason for this.

It might be worth a try, but in theory there is a problem: USB sticks tend to suck in IOPS once you are reaching a certain transfer size level. A typical trasfer size pattern in ArmA2 is 80% between 4k and 64k, 17% between 64k and 1M, and 3% over 1M. So whilst you will see a 2x -3x increase in 80% of the times (over a normal HDD), in 20% of the occasions you might be down to less than HDD performance, in extreme case down by 4/5. But again, it is worth a try - it might well be that you can get rid of a substantial level of stutter at the expense of having a small number of extreme stutters and/or increased level loading times. If anyone tests this, could you post the results using the tool in my sig so we better understand whether this helps ?

Hey Shataan, no offense meant, but the answer is very close, if you have a look :) http://forums.bistudio.com/showthread.php?t=88388

I can beat that lol - Mushkin 12GB (1600Mhz) CL 7-7-7-20 1T triple channel :p By the way - too fast for HDTune (talk about "through the roof" performance)...:bounce3:

Because ArmA 2 (and most other games) is 32bit, there is overhead for the emulation that causes a 2% (in theory) and more than 2% (in reality) performance hit. On top of that, ArmA 2 is DX9, which needs again to be emulated by a DX10/DX11 native OS. Don't get me wrong, I love Win7 x64 as my main rig, but I boot into XP32 for ArmA2.

Hi oktane, that's great insight! You might be on the right track here. I guess that the amount and size of data blocks have just gotten too large to be efficiently handled by the Windoze system cache. On top of that, with arma2.exe still being a 32bit app, it can only address 2GB of data including program code and therefore has not much address space left for any user level caching (which would cause the same probs as using the Windoze cache in the first place, thats why I doubt they did this), so it needs to constantly access the data storage volume to stream the content. A way to fix this (besides putting the data on a RAMdisk so that effectively cache=disk) would be a 64bit ArmA on a PC with 6GB+ RAM that comes with its own user-mode cache. Wait a minute - why wait for BIS - Are you a programmer? How about hacking a system-level driver that allocates a lot of system RAM for its own proprietary cache in 64bit, prefetches the stuff and then captures ArmA's unbuffered I/O's and redirects it to that buffer? Sorry if that is BS, I am not a coder :D ---------- Post added at 02:10 AM ---------- Previous post was at 02:06 AM ---------- By the way, new version of the tool is out (see my signature). Just a minor error fix.

There's many products that work. Under 64bit, you don't have to use an advanced product that can use "hidden" RAM, i.e. RAM above the 32bit barrier. Amongst the advanced products, I have only tested Dataram RAMdisk (http://memory.dataram.com/products-and-services/software/ramdisk) and Superspeed Ramdisk Plus (http://www.superspeed.com/desktop/ramdisk.php). The Dataram is free for up to 4GB Ramdisk size, and you can send them an email and ask for a unlock key (I got one for free) to use more than 4GB size. The Superspeed has no free version, but a 15 days trial. All in all, they are both working well, with a few differences: a) Both can support >4GB adressing on 32bit, however, if hidden RAM is enabled, Dataram starts counting at exactly 4096MB (so with 12GB, you can only create a 8GB ramdisk), whilst Superspeed can mix both the RAM below and above the 4GB barrier (so I can create a 8.5GB ramdisk which is what you need for a full ArmA2 install) b) Both have a nice user interface and wizards so its easy to set them up c) SuperSpeed is ca. 30% faster than Dataram (but for ArmA2, both are fast enough by far) d) SuperSpeed costs $60-$90 after 15 days trial Some products I have not tested (yet) myself but I have read good things about: Qsoft RAMdisk: (http://www.ramdisk.tk/) Seems not to be backed by a "real" company but seems to be a good (russian?) product. Prices between $10 and $25 (for the full version, that is - there is a free lite version). It seems that some others here in the forum have tried this one. Romex VSuite Ramdisk: http://www.romexsoftware.com/ is a pretty new product from China but seems to be very feature rich and recieves a lot of good feedback. Prices between $30 and $100. Gavotte RAM disk: (http://www.mydigitallife.info/2007/05/27/free-ramdisk-for-windows-vista-xp-2000-and-2003-server/) This is a "hack" of the original Microsoft ramdisk froma Japanese guy that enables a full feature set and is free. I have heard good things, but this is probably more for people who like to fickle around and use the CMD and RegEdit. Also make sure you download the newest version from the internet (dont forget to scan for malware!) - the original developer website is in japanese (http://www10.atwiki.jp/gavotterd/). It really depends on what you need - some products might have features that you might prefer over others. Important features for me are: 1) Needs to be able to use "hidden" RAM in 32bit (using XP32 for ArmA2) - you don't need that feature on 64bit 2) Should be able to load and save and image of the ramdisk so you don't have to set it up again and copy the game files every time 3) Automatic loading is less important for me When you are in Windows, you can just load the backup image of your ramdisk with the relevant program before launching ArmA2. Then you unload it again (or save it, if you want to keep changes) when you are done. That simple. Some products have automatic loading/unloading during Windows bootup/shutdown, but I have had errors with this functions at times (probably a timeout because it takes a long time to load/save 8.5GB to disk). Last, a word of caution: Always have a backup of any data you are putting on the ramdrive that you want to keep - a RAMdisk is inherently not useful for long term storage, with or without autosave. You should in any case read the websites and descriptions, compare features and use trial versions before making a buying decision - I am afraid there is no product that is "right" for everyone. Have fun!

By the way, in my own tests, I have found that flying around is actually LEAST stressing in terms of I/O - I guess that's because you are high up and staying on a quite low LOD. Most stressing is running/driving around in towns/cities. I have taken the freedom to post your results below - hope that's fine with you.

I think it all depends. I am sure that - compared to a mechanical hard disk - a modern SSD is a huge improvement. If you look at the I/O results using a RAMdrive, though, you will see that ArmA 2 can generate a load of more than 1000 IOPS (I will post some results later). Now have a look at SSD performance: Random I/O maxes out at 300-400 IOPS with these SSDs. So, while 98% of the time ArmA 2 will work in the range of what a SSD can deliver, it is those nasty 2% of times where they don't that can create visible stutter. Here's a (slow) RAMdisk for comparison: And here a Hard Disk: Basically, as a rule of thumb, in random read a RAMdisk is 100x, a SDD is 10x faster than a HDD. But at the end you are quite right, there is a certain level of perfectionism at work here :bounce3:

That's most certainly not an I/O bottleneck. Removing the I/O bottleneck does not increase the frame rate as such (which is limited by your CPU and GPU performance), but increases the "smoothness" of the game, i.e. no stutter and hiccups. In your case, I would think it might be a CPU bottleneck or bad configuration - are you playing on Vista ?

Yep, good find. That's exactly the effect - the faster the disk system, the less the stutter. With a RAMdisk, the stutter is gone completety.