ArmA2 / OA (low) performance issues

[MavericK96 0]

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.

I honestly think it's because the difference would be so minuscule that it's not worth the effort. Have a look at this review:

http://www.tomshardware.com/reviews/core-i7-870-1156,2482-9.html

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. :cool:

[RadicalAtHeart 11]

I honestly think it's because the difference would be so minuscule that it's not worth the effort. Have a look at this review:

http://www.tomshardware.com/reviews/core-i7-870-1156,2482-9.html

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. :cool:

Instead of making asumptions we can at least try benchmarking the difference. At least I will when I have my new BIOS chip :)

I will sure do the RAM benchmarks and let you guys know the outcome. :D I will make a detailed Excel document about my results :)!

Does anyone here know why RAM drives aren't recommended ?

[MavericK96 0]

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. :D

[RadicalAtHeart 11]

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. :D

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[0], thread 1 accesses a[1] 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.

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 :/

Also explains why RAMDrives will do more bad than good.

[valnwt 11]

I recently tried alot of things written here and, well didnt really see any results. Ive got the same problem as many, cpu stuck at 50-70%,Gpu at 40-50% (this regardless the ingame setting), i do have one question, regarding the Resource Monitor in windows 7. While running Arma 2 benchmark 2 and looking att Ram usage, which section shows application ram usage, Commit(KB), Working Set (KB), Shareable (KB) or Private (Kb)?(this in Memory tab)However, non of them show more usage on arma2oa.exe than 1.5 gig, shouldnt it be 2gig? It really feels like something is keeping arma from using more resources since the fps go low but usage dont.

Some numbers:

Benchmark 2 = 17 fps

Benchmark 1 = 53 fps// give or take 5 fps depending on settings

One more thin to point out, only one of my four cores run around 80-90% rest are at 30-40%(if not lower).

My rig:

CPU i7 960 (no HT)

GPU Gtx 560 ti

24 gig ram

120 gig SSD

[RadicalAtHeart 11]

BENCHMARK RAM

I used BENCHMARK02 from ArmA 2 because it's most intensive benchmark

i5-2500k @ 4.4ghz

8GB RAM

HD6970

TEST 1 - 1333mhz 9-9-9-24-1N - 6.75ns - AVG 13.4 FPS

TEST 2 - 1333mhz 8-8-8-21-1N - 6.00ns - AVG 14.2 FPS (-12% latency, +6% improvement over test 1)

TEST 3 - 1333mhz 7-7-7-18-1N??? still waiting for it to become stable

Going back to 1066 mhz gave me a huge performance drop to 12.9FPS with the same latency 6.75ns

We can conclude this game is dependant on RAM speed as well.

Every time I did 10 runs to ensure accuracy.

IN PROGRESS

Edited February 20, 2012 by RadicalAtHeart

[MavericK96 0]

+/-2 FPS is not a "huge" performance drop IMO. Still, interesting results, I'd be interested to see more if anyone has faster ram like 1866 or 2000.

[RadicalAtHeart 11]

+/-2 FPS is not a "huge" performance drop IMO. Still, interesting results, I'd be interested to see more if anyone has faster ram like 1866 or 2000.

I also noticed a performance drop testing DDR3-1066 speeds. It was about 0.5FPS less when using the same calculated latency ~6.75*2 ns

At such a low FPS, these results are still impressive in my opinion. Even they are only 0.5FPS difference. We talk about a very low average FPS. I think we should keep that in mind. I ran every benchmark 10 times without alt tabbing out or restarting the game to be sure the game was loaded properly.

But I mainly think this game lacks proper multithreading. One core using 80% and others 20%.

I also did not notice a difference adding -exThreads=7 maybe because it is default already.

[MavericK96 0]

Yeah, I agree that something is limiting performance, and it is possibly a threading issue. I often get low GPU and CPU utilization in conjunction with low framerates, and there's no real explanation as to why. For me the two biggest contributors seem to be when there is a LOT of foliage and when there is a LOT of AI. One or both of these factors can cause aberrant utilization of both GPU and CPU.

I posted some information in this thread which is somewhat related:

http://forums.bistudio.com/showthread.php?127236-Super-low-GPU-usage-on-MP-only

[RadicalAtHeart 11]

On a i5-2500k @4.3ghz i report performance improvement using -noCB switch together with -cpuCount=4 -maxmem=2047

Benchmark02: 17fps

vs

Benchmark02: 19fps

So it is possible the game runs faster this way on cpu's with higher freqs

[MavericK96 0]

That is interesting, I'd like to see more testing with that -noCB parameter. Never even heard of it until now, actually. Lol.

[denco 16]

Im just about to give up on this game. My current system specs are

-i7 2600k 3.4 HT on, tried 4.2 no diffrence ( again no diffrence with HT off )

-Sapphire HD 7970

-RAM : 8GB Corsair Vengance

-MOTHERBOARD Maximus IV gene-z

Game settings:

-Everything Very High, memory on Default, PP on Normal and AA on normal, vsync on, viewdistance 3000. No mods

-Benchmark 2: 20fps HT on

-Benchmark 2: 19fps HT off

-Benchmark 2: 14fps HT on radeon HD 6850 ???????????????

-Benchmark 2:13fps with ACE, Warfx, JSRS.

Now the game does run a lot better then when I had 6850 but to be fair I expected more. No matter what you put in it doesn't help a lot. I hope they will optimize arma 3 better or else I'll be stuck with everything on very low and viewdistance on 500 m and I'm not being sarcastic. Other games run great and I'm really happy with this card ( its a beast ) but arma I guess is just being arma.

[valnwt 11]

What does your Resource Monitor (plus GPU usage) read during the benchmark run?

[denco 16]

LOL

GPU usage was only 24%

CPU Usage is as follows:

-Core #0 Thread #0:-71.4%

-Core #0 Thread #1: -69.7%

-Core #1 Thread #0: -63.8%

-Core #1 Thread #1: -55.1%

-Core #2 Thread #0: -75.3%

-Core #2 Thread #1: -48.0%

-Core #3 Thread #0: -72.3%

-Core #3 Thread #1: -41,7%

-Total CPU usage: 41.7%

The monitoring was taken with MSI afterburner for GPU and HWiNFO64 for CPU.

[MavericK96 0]

There should definitely have been a difference between 3.4 and 4.2 GHz. When I overclocked from 2.66 to 3.8 there was a huge difference.

Hyperthreading probably isn't going to help much, though. You could try setting cpuCount=8 though if you have it enabled and see if that makes any difference.

But yeah there is a definite issue with the game itself in regards to low CPU and GPU usage sometimes, along with poor framerates. Still trying to figure out what causes it.

[denco 16]

No I only increased turbo boost to 4.2 but since my cpu is not used 100% I don't think Turbo boost is used in this game ( Of course, if a game isn't asking for more performance, Turbo Boost doesn't deliver it ). My normal cpu runs at 3.4 and when it needs more cpu power it turns on boost to get more power. My rig is build for overcloaking but I need to get a more powerful PSU before I get down to any serious or rather real overcloaking besides games are working fine for now so I'm not thinking about this yet. I love this game but sometimes it pisses me off and Like I said I already given up on this I just hope they optimize arma 3 not better but a lot better.

[valnwt 11]

There should definitely have been a difference between 3.4 and 4.2 GHz. When I overclocked from 2.66 to 3.8 there was a huge difference.

Hyperthreading probably isn't going to help much, though. You could try setting cpuCount=8 though if you have it enabled and see if that makes any difference.

But yeah there is a definite issue with the game itself in regards to low CPU and GPU usage sometimes, along with poor framerates. Still trying to figure out what causes it.

Ok, so you got a fps gain by OC your CPU. Then i ask, as earlier, did you do any cpu usage test during benchmark 2(this when your cpu wasnt OC)? Also just really curious, whats your CPU usage when OC?

[NeuroFunker 11]

i can't test my core i5 2500k now, but my CPU usage droped around 7%-10% after i clocked it to 4,3ghz from 3,3mhz.

[MavericK96 0]

Plus the fact that there is no reason to defrag an SSD because access from any "part" of the drive is always just as fast as from any other "part".

EDIT: Whoops, not sure where the post that I was replying to went... :confused:

[Leopardi 0]

Is GTX 460 really not enough to render the leaf trees or is there some trick to prevent FPS dropping by about 50% when one of the trees is on the screen?

Edited February 26, 2012 by Leopardi

[denco 16]

Is GTX 460 really not enough to render the leaf trees or is there some trick to prevent FPS dropping by about 50% when one of the trees is on the screen?

There is a mod on armaholic that replaces all leaf trees with pine trees but I can't remember exactly where so search for it.

[domokun 515]

[/color]

There is a mod on armaholic that replaces all leaf trees with pine trees but I can't remember exactly where so search for it.

You're right Denco, something's not quite right there (you should be getting much better performance).

Try these tried and testing recommendations.

We'd be interested to know how you get on...

[denco 16]

[/color]

You're right Denco, something's not quite right there (you should be getting much better performance).

Try these tried and testing recommendations.

We'd be interested to know how you get on...

I got good frames If I have viewdistance set to 3000 and In MP when I'm playing with my squad or if I'm playing in the mission editor but I get a lot of bad fps in Red Harvest campaign sometimes specialy when there is a lot of action it goes bellow 25 but I think I should be getting more I mean 2600k is one of the good processors today and 7970 is currently the fastest single gpu out there. I'm still experimenting what could be wrong but I suspect that this games is not optimized in some parts.

In MP I have viewdistance set to 5000 and on some islands like takistan to 8000 and it works good.

Yesterday I played around in the editor and later I found that my gpu was used 99% and my cpu was used ( shit I forgot ) at some point but when I run Benchmark 2 My cpu is used 50% and GPU is used 24% like the game doesn't take what it need's?

Anyway as long as its smooth in MP I won't complain too much but I really like the Harvest Red Campaign and would like to play it with viewdistance set to 5000 but sadly I think this is just a nice little dream I have.

[domokun 515]

I got good frames If I have viewdistance set to 3000 and In MP when I'm playing with my squad or if I'm playing in the mission editor but I get a lot of bad fps in Red Harvest campaign sometimes specialy when there is a lot of action it goes bellow 25 but I think I should be getting more I mean 2600k is one of the good processors today and 7970 is currently the fastest single gpu out there. I'm still experimenting what could be wrong but I suspect that this games is not optimized in some parts.

In MP I have viewdistance set to 5000 and on some islands like takistan to 8000 and it works good.

Yesterday I played around in the editor and later I found that my gpu was used 99% and my cpu was used ( shit I forgot ) at some point but when I run Benchmark 2 My cpu is used 50% and GPU is used 24% like the game doesn't take what it need's?

Anyway as long as its smooth in MP I won't complain too much but I really like the Harvest Red Campaign and would like to play it with viewdistance set to 5000 but sadly I think this is just a nice little dream I have.

No doubt that View Distance is has significant impact on framerates but so does AA, Post-Processing and AToC. If VS is paramount to you, have you tried lowering AA, PP and/or disabling AToC? Also try increasing the Shadows to High or Very. OA does indeed require some thought when it comes to configuration. Also the in-built campaign is known as a performance-killer so it is rather optimistic to expect to run an 8K VD smoothly. Even a 5k VD at smooth framerates is tricky.

[denco 16]

No doubt that View Distance is has significant impact on framerates but so does AA, Post-Processing and AToC. If VS is paramount to you, have you tried lowering AA, PP and/or disabling AToC? Also try increasing the Shadows to High or Very. OA does indeed require some thought when it comes to configuration. Also the in-built campaign is known as a performance-killer so it is rather optimistic to expect to run an 8K VD smoothly. Even a 5k VD at smooth framerates is tricky.

-AA is set to Very High ( I'm playing at 1024 x 1280 resolution and I hate jaggines ), PP and World detail is set to normal ATOC is off and Memory is set to Default. Everything else is set to very high.

-Correct me If I'm wrong but VD is mostly dependent on cpu right?