Check my rig in my signature. I was not able to play arma 2 on very high, had to set object details to normal or low, cause i laged alot with it. Now i'm able to play arma 2 on very high, with min 30 fps in forest areas, and 40-50 cities/outside in fields.
What i did: fresh windows 7 x64 reinstall, windows update right after install, reboot. After that, installed all latest drivers for my motherboard/GPU/soundcard. Reboot - amd vision centre - GPU overdrive - gpu power to +20%, abit overclock to 945mhz without voltage increase. Litle bit d3 settings tweak. Catalyst A.I very high, AF 16x etc. Now started arma 2, and was lucky to play arma 2 on very high again!
If someone expecting bad performance, even he has good pc - try all i did, it might help. Don't forget to backup all your important documents before u reinstall windows ofcourse
I hope it will help anybody so it did to me!
Did anyone try benchmarking the game with different RAM speeds and timings?
I am thinking of upgrading my DDR3-1333 to DDR3-1866
If it isn't my GPU/CPU/Disk all is left is RAM right?
I tried running ArmA 2 OA on 3 cores on my 2500k it ran with the same performance as with 4 cores @ 4.4ghz speed :/
I now have two sticks of DDR3-1333 CL9
I'm not completely sure for the purposes of ArmA2/OA, but from what I've seen of RAM benchmarks the difference between 1333 and 1600 or even 1866 is very minimal. I am guessing you would not see much difference at all.
How much RAM do you have total?
---------- Post added at 00:34 ---------- Previous post was at 00:30 ----------
It might help if we knew more about the servers you play on and if you've tried any different ones, i.e. if it happens all the time or just on some servers. Also, what is your internet connection speed/type?
Core i7 920 @ 3.995 GHz, HT off
12 GB OCZ DDR3-1600
2x Intel X25-M 80 GB SSD
Windows 7 Pro x64
1920x1080 w/ View Distance at ~3600
Video Memory at Default
MSAA Very High, AToC=0, SMAA Ultra
Post Processing at Very Low
All other settings at Very High
I am very sure I am not bottlenecked by my hard drive. 10MB/s doesn't create a bottleneck on a 7200RPM drive.
Since everyone is using custom memory allocators I was just thinking memory modules may cause a difference.
But has anyone every benchmarked ArmA 2 on different memory speeds? No because nobody thought of it. They did however benchmark every memory allocator for every beta.
Maybe I need Quad Channel memory (all slots filled)
I am saving up for 32GB RAM. This may be overkill. But I need it anyway for virtualization software.
Maverick you may be able to underclock your RAM to 1333 and test at those speeds as well. If you are willing to do that.
Do 5 runs of the benchmark to make sure everything is loaded so we have the least possible slowdowns of your hard drive.
I will test different latencies as soon as I have my new BIOS chip for my bricked motherboard.
instead of overclocking your ram, you could lower the timings, i have changed from 9-9-9-24 to 8-7-8-21, and got +0.1-0.2 points in windows 7 bench thingy. Works fine with standard 1.5 volts.
As you can see, the average performance gain going from 1600 to 1833 is something like 1.4%.
You are more than welcome to run some tests and get back to us, though.
I will sure do the RAM benchmarks and let you guys know the outcome. I will make a detailed Excel document about my results !
Does anyone here know why RAM drives aren't recommended ?
Dwarden just commented on that in the thread about RAMDrives, he said it's because CPU overhead is pretty high with a RAMdrive, so really you are not going to gain much due to that. Theoretically a RAMDrive is much faster than SSD, but in practice...not so much (in my experience, anyway).
Also, it's not really an "assumption", since I did post an actual RAM benchmark. But you're right, I don't think anyone has specifically tested it in ArmA2, so I look forward to your results.
according to my source it is possible we might have a RAM bottleneck if this is the situation in ArmA 2 so I hope I will find out tomorrow.. This might explain why our cores are never fully used :/Yes, RAM I/O (or the FSB) often becomes the bottleneck.
There are plenty of other possibilities though. Do the two threads use variables *next to* each others?
If you have an array a, and thread 0 accesses a, thread 1 accesses a and so on, that will hurt performance. Because the CPU caches don't operate with single bytes, but with cache lines (typically 32 byte per line, which corresponds to 8 ints or floats, or 4 doubles)
So if this is the case, and the two threads access data a few bytes away from each others at the same time, they'll have to move that cache line from one core to the other, and back, and forth again, and back. (Since it may not exist in both cores' caches at the same time.
Finally, you may want to use 3 or 4 threads, in order to ensure that there's always a thread ready to run, even if one gets blocked. You generally need slightly more threads than you have cores for best performance.
Without knowing more about how your program works, it's impossible to say what's holding you back.
Another related, but simpler explanation might be that the singlethreaded version just gets better cache locality. It doesn't get as many cache misses as the multithreaded version, for whatever reason. Again, impossible to say without knowing more about your program.
A third option might be that the greater bandwidth usage means your program is seeing relatively higher latencies (because there are more pending requests that have to be served before *your* request returns data, which causes the CPU's to stall and have nothing to do for some of the time. That might be possible to fix by rearranging your code a bit to reduce dependencies between instructions.
Btw, don't run Sandra with your program running. It's meant to profile your system *alone*. Anything you get while the CPU is busy with other processes is going to be highly skewed and inaccurate.
There's no way to determine how much RAM bandwidth is being used at any instant in time. The reason being, to do that, you have to keep track of everything that happens for a few hundred nanoseconds, which would take so much CPU time, it'd skew the results badly.
Also explains why RAMDrives will do more bad than good.