Jump to content
Sign in to follow this  
Alabatross

When will the nVidia Arma 3 optimized drivers come?

Recommended Posts

Since its finally released I'm really looking forward to that 30% performance increase they add, any word on this or is it something nVidia does on their own without BIS?

Share this post


Link to post
Share on other sites

Useless. When BI start optimizing their own engine we will start to see the biggest fps increase with or without the latest drivers.

Share this post


Link to post
Share on other sites

Arma 2 was a trainwreck in performance optimization so I didn't have high hopes for A3, its a shame really.

All during the alpha if you complained about the game not running well and the first excuse you get is "get a better card/cpu" or "its alpha" which is the only thing I hate about the PC community. You can never tell who's fault it is, but in this case I'd have to point the finger at BIS because we're at release and nothing has really changed. I have no issues in planet side 2, GTA 4, BF3, Red Orchestra so its obvious their rendering pipeline has some flaws in it (most likely what someone has already said, its too CPU heavy)

Planet Side 2 at launch ran like crap, but even the devs acknowledged it was too CPU heavy and over the next 3 months shifted more on the GPU and now I can run the game on ultra just fine at 60 fps. (I used to have to use a modified medium setting ini file and I only got 38)

If I run it on Ultra in MP I get 25-35 fps. If I run it on lowest in MP I get 25-35 fps, which is the exact same thing that happened in Arma 2 for me

Edited by Alabatross

Share this post


Link to post
Share on other sites

The answer you will get a lot on these forums is:

A) your computer is CPU bottlenecked

B) your computer is GPU bottlenecked

The solution we are actually looking for are:

A) Devs are working on optimizing engine

B) Nvidia is working with BIS to optimize drivers for A3.

But very few people on this forum will actually want to come out and say yes BIS and Nvidia need to get together to work out this problem so that users can get a solution. Instead they will revert to the CPU/GPU bottleneck answer that gives the developers a free pass. I'm not saying it is easy to optimize for a game like Arma but I'm tired of seeing people reporting legitimate issues with performance and getting bs answers excusing the developers from responsibility for their product.

You can come here and post a spec with some of the best hardware available on the market but someone will always chime in and say "oh, but it doesn't matter that you have a $2000 set of video cards, you are CPU bottlenecked". And even if you respond with the fact that you have a $1000 6 core CPU, they'd rather argue how you need an even more powerful CPU to get any performance increases. I'm as much a fan of BIS and the Arma series as anyone else on this forum but I always get annoyed by these kinds of excuses the community likes to indulge in.

Share this post


Link to post
Share on other sites

The game only uses about 65% of my CPU at the most (i5 2500K), a similar thing is happening for most people. This suggests to me that it's completely a fault with the games optimisation.

---------- Post added at 21:28 ---------- Previous post was at 21:15 ----------

I find it funny how when Arma 3 is released is when I STOP playing it, so far it has been a bigger let down than BF3 was at the time that came out and I never thought any other game would be a bigger let down than that. For it to be the sequel to the game I like the most and have spent by far the most amount of time playing, that's pretty disappointing. Maybe if the thing ran well that would be a bit of an improvement and maybe enough to make me consider playing it ever again.

Edited by clydefrog

Share this post


Link to post
Share on other sites

Man i'm tired of people saying, that they have poor performance, everybody knows the game is cpu heavy, Dev knows, we know, but solutions... buy the better proc you can afford...I don't know if people notice, but arma have a view of 2km (average player setting max 15 or 12...) and all other game have 500 m on average, some like crysis (looks good, but have a render of 30 meter....)I think the problem is that, and when you put arma and other side by side, other have a much little vision range even in the bigger maps in bf, cod crysis etc is a joke the render distance, also arma have to manage a LOT of simulation/Ai no other game have ballistics like arma, penetration diversion ricochet etc...and the Ai of arma, is one of the best..(I think)

Share this post


Link to post
Share on other sites
Man i'm tired of people saying, that they have poor performance, everybody knows the game is cpu heavy, Dev knows, we know, but solutions... buy the better proc you can afford...I don't know if people notice, but arma have a view of 2km (average player setting max 15 or 12...) and all other game have 500 m on average, some like crysis (looks good, but have a render of 30 meter....)I think the problem is that, and when you put arma and other side by side, other have a much little vision range even in the bigger maps in bf, cod crysis etc is a joke the render distance, also arma have to manage a LOT of simulation/Ai no other game have ballistics like arma, penetration diversion ricochet etc...and the Ai of arma, is one of the best..(I think)

It's nothing to do with view distance, I bet most people are playing with about 2000 which is low and still getting poor framerate on multiplayer games. On arrowhead I can put view distance up to about 5 or 6000 and still get decent framerate. Telling people with new i5's and i7's and whatever AMD's to buy new processors because that's their problem? You're talking complete rubbish. If the problem was just that the game is CPU heavy it would be maxing out peoples CPU's wouldn't it, not using just over half of what's available to it. It's a CPU intensive game that doesn't know how to use a CPU intensively (and especially multicore ones) which equals bad performance.

Edited by clydefrog

Share this post


Link to post
Share on other sites

the cpu isn't limited, the game uses fuck all of it, it's a poorly optimized turd.

Went from a Core I7 950 @3.8 to a haswell 4770k @4.3 and my FPS is still under 30.... When you look at the usage it's not even 40% and most of the cores/threads are untouched.

Edited by scaramoosh

Share this post


Link to post
Share on other sites

*somehow clicking back on the page reposted my old post

Share this post


Link to post
Share on other sites
Your CPU is used by its 30% .. and your GPU as well.. drivers won't help to improve anything.

I understand everyone would like the game to run smoother, better, best. In reality, there is no spell that could be used, and many days or weeks of hard engineering work lead into neglible performance increase.

As I read all over again and again posts that seems to say people believe measuring how much of their multicore CPU or GPU is utilized (often by some very generic means) I can only recommend to try to read and understand problems associated with concurency in game engine architecture there is still some good read avaialble here: http://www.bistudio.com/english/company/developers-blog/91-real-virtuality-going-multicore

So yes, it is correct that main bottleneck is singlethreaded performance of the main thread for Arma series, yet it does not have any simple solution (and I do not think Arma is the only game that fails to really benefit from anything beyond dual core very well). Depedning on scenario complexity, the game can scale well on two cores, to some level to four cores but benefit is less visible. With GPU: I do not know, why you simply do not set the GPU settings higher to get best from your PC? If you have powerful GPU, there are many options you can maximize without getting any real penalty on the CPU side. Generally speaking, rendering distance and even object detail has some serious impact on the CPU as well (also affects simulation), yet most of the other graphical settings are solely or primarily on the GPU side.

I am not trying to start any flame war and not denying there is not room for optimizations or that it is not important priority, I am mostly tired to read all over again something that just is more myth than reality and that simply uses completely false metrics. All I can do is quote Suma:

It is important not to lose the sights from the goal, which is the performance increase. All other things are secondary. One example of wrong metrics is a concurrency level. Concurrency level tells us how much are the additional cores used. This factor is very easy to measure (you can do it in default system task manager), and that is probably why many hard core end users and reviewers are interested about it. Often you can see phrases like "Game XXXX is using quad cores very well, because when you watch CPU usage in task manager, you see all cores are running 100 %". It is very easy to create a trivial program which will make "full use of all cores" - all you need to do it to spawn a few threads and make them spin in an infinite loop. Concurrency is not a goal, only a mean. It is required, but not sufficient. Real life scenarios are more intricated then idle loops, but the principle is the same: using CPU does not mean you get any benefit from using it. In many cases the overhead of going "threaded" is so high that even when two cores are running 100 %, the performance improvement is very small, say about 20 % from single core, and the difference between quad and dual is even smaller.
Edited by Maruk

Share this post


Link to post
Share on other sites

So yes, it is correct that main bottleneck is singlethreaded performance of the main thread for Arma series, yet it does not have any simple solution (and I do not think Arma is the only game that fails to really benefit from anything beyond dual core very well). Depedning on scenario complexity, the game can scale well on two cores, to some level to four cores but benefit is less visible.

In this case why do the system specs state that an Intel Core i5-2300 or AMD Phenom II X4 940 is recommended when really it sounds like there isn't much point in buying such a processor for this game.

Share this post


Link to post
Share on other sites
In this case why do the system specs state that an Intel Core i5-2300 or AMD Phenom II X4 940 is recommended when really it sounds like there isn't much point in buying such a processor for this game.

You misread what I am saying. If I can try to answer the question: Quad core is better than dual core for Arma 3 generally speaking. But that does not mean the benefit is 2x higher fps over dual core. One of the benefits you may not be able to measure at all (certainly not by using system CPU monitor) is that AI gets more CPU power for its calculations depending on the complexity of the scenario. Maybe if you try to read the blog, you can understand nature of the problem a little bit more. It is not trivial problem, does not have any trivial solution but most importantly: a lot of users seem to miss what the real goals are and that their way of assessing it has no real value.

Share this post


Link to post
Share on other sites
You misread what I am saying. If I can try to answer the question: Quad core is better than dual core for Arma 3 generally speaking. But that does not mean the benefit is 2x higher fps over dual core.

I know that and know that's not at all how multicore works. You said the game "fails to really benefit from anything beyond dual core very well" and "the game can scale well on two cores, to some level to four cores but benefit is less visible.". So I'm asking, well what's the point in buying a quad core CPU to play Arma then if you don't yet have one? Will it improve performance even though it can't really benefit from the extra cores available? Is it just because these CPU's have considerably better technology than most dual core processors which makes them faster and better for the game in general?

Edited by clydefrog

Share this post


Link to post
Share on other sites

Unless you want your AI to get some more room to breath in more complex scenarios, I think there is no big benefit from quad cover over dual core CPU in Arma 3 at this stage but I did not play on dual core system for long time so I can be wrong and I do not know exact specifics about how physx scales in multicore etc. Also, I think small benefit is still benefit so certainly better to have quad core if you can choose. And anyway, based on my random browsing of the performance related threads on the forums I nearly seem to be playing different game on different hardware anyway /frequently on my quad core laptop where I can play on battery mode using the integrated intel HD4000 GPU and I am fine, the game runs better for me than Arma 2/.

And there is another aspect that can be important in Arma: memory bandwith, CPU cache etc. And I guess these are other problems very hard to measure. Arma 3 code is very large so I would not understimate impact of the memory on the performace but I am only quessing here (but my laptop with Core i7-3610QM is so much faster than any previous laptops I used in the past and the memory bandwith got improved dramatically with this generation of mobile CPUs)

Edited by Maruk

Share this post


Link to post
Share on other sites

And have you tried going on any public multiplayer servers? As this seems to be where people are experiencing the worst of performance issues where they get huge GPU usage drops which results in low frame rate down to around 20fps or less even with mid to high-end hardware. I have played Arma 2 Combined Ops for 2 years and I have never really experienced this on similar missions with many players and AI. It has happened before as performance can get worse the longer a mission goes on if it is not optimised well, but I have not seen it happen on the scale it happens in Arma 3.

Edited by clydefrog

Share this post


Link to post
Share on other sites
I understand everyone would like the game to run smoother, better, best. In reality, there is no spell that could be used, and many days or weeks of hard engineering work lead into neglible performance increase.

As I read all over again and again posts that seems to say people believe measuring how much of their multicore CPU or GPU is utilized (often by some very generic means) I can only recommend to try to read and understand problems associated with concurency in game engine architecture there is still some good read avaialble here: http://www.bistudio.com/english/company/developers-blog/91-real-virtuality-going-multicore

So yes, it is correct that main bottleneck is singlethreaded performance of the main thread for Arma series, yet it does not have any simple solution (and I do not think Arma is the only game that fails to really benefit from anything beyond dual core very well). Depedning on scenario complexity, the game can scale well on two cores, to some level to four cores but benefit is less visible. With GPU: I do not know, why you simply do not set the GPU settings higher to get best from your PC? If you have powerful GPU, there are many options you can maximize without getting any real penalty on the CPU side. Generally speaking, rendering distance and even object detail has some serious impact on the CPU as well (also affects simulation), yet most of the other graphical settings are solely or primarily on the GPU side.

I am not trying to start any flame war and not denying there is not room for optimizations or that it is not important priority, I am mostly tired to read all over again something that just is more myth than reality and that simply uses completely false metrics. All I can do is quote Suma:

I don't mean to sound rude, but this sounds like a cop out to me. I realize there are hard but solvable issue's with concurrency and that parallel processing comes with it's own issue's and problems and it's not a linear performance increase per core, but to me it sounds like you are using those problems as the basis for why you don't work on improving thread concurrency within your own program.

Basically that you don't see why your company and your employee's should work towards the future, like better parallel processing and concurrency, to improve what little they can about the game and the engine when you are comfortable with the performance we see right now and consider it "Too hard" to fix.

This is how it sounds, it may not be exactly what you mean but it's what you are implying with what you are writing. I am very realistic in my expectations for proper parallel programming within the RV engine. I don't expect every core to be screaming at 100% usage and that 100% usage across all cores means that the engine is properly utilizing them. But seeing 28% CPU usage in just about any situation in ArmA 3 and watching that usage then DROP as the scenario performance degrades even more, does not suggest that the game is utilizing the hardware very effectively. In fact a situation like that wreaks of some sort of thread contention or other serious bottleneck.

One thing that "I" am tired of reading is the so called explanation that everything is either the Customers hardware or the mission they are playing's fault, even when it's an empty mission with no scripts or they're overclocked to the max and have top of the line hardware and getting 20 fps, as being the justification for why the game has the performance that it does. Hearing that fixing the problem is just too hard and that you are uninterested in pursuing it is just as tiring as well.

Share this post


Link to post
Share on other sites

I think performance in MP won't change a lot unless they move to a server-client architecture like DayZ so the simulation will be clalculated server-side and not server AND client side. Arma's current client-server-client architecutre is simply outdated and is probably the biggest impact on performance in MP... unfortunately BI didn't realize that soon enough (only after DayZ)....

Share this post


Link to post
Share on other sites

Of course if the main thread is stuck for whatever reason, you are going to see drop in utilization of all other cores (and GPU). In any case, why you try to read between the lines and come to conclusion that we are not interested and will not be trying hard to improve the game performance is beyond me. But in reality, multicore is not solution to all problems and it is easily possible, that right now there could be some other problem in mp that leads people into building their own theories. So as always, problem has to be identified, reproduced and as such particular bug can be addressed. All I wanted to say is that users measuring their CPU and GPU usage are focusing attention in false direction. Focusing on game fps itself is much more valid approach for example but more critical are occasional significant drops in performance than average fps.

That was my main point, not trying to defend our "lazy and incompetent" engineers (as I see them called on the forums often: and I strongly disagree with this view btw) or hide behind nice words. But on a side note: I have been tempted few times to ask engineers to spawn all cores with dummy loops to please this type of audience. The problem is that it benefits no one. So on more constructive note: there easily could be some problem (or many different problems) in multiplayer causing serious performance degration and that certainly is high prioirty to track down and fix asap for us. It is just there is no magical multicore optimization that could out of sudden improve everything available as some may seem to hope for when they see some of their cores are nearly idle most of the time. Most important is this: there is no single "general performance problem of CPU or GPU utlization", there are only specific problems (and these may easily be completely different in mp, sp, different depending on users hw etc.) that need to be fixed.

Edited by Maruk

Share this post


Link to post
Share on other sites
Of course if the main thread is stuck for whatever reason, you are going to see drop in utilization of all other cores (and GPU). In any case, why you try to read between the lines and come to conclusion that we are not interested and will not be trying hard to improve the game performance is beyond me. But in reality, multicore is not solution to all problems and it is easily possible, that right now there could be some other problem in mp that leads people into building their own theories. So as always, problem has to be identified, reproduced and as such particular bug can be addressed. All I wanted to say is that users measuring their CPU and GPU usage are focusing attention in false direction. Focusing on game fps itself is much more valid approach for example but more critical are occasional significant drops in performance than average fps.

That was my main point, not trying to defend our "lazy and incompetent" engineers (as I see them called on the forums often: and I strongly disagree with this view btw) or hide behind nice words. But on a side note: I have been tempted few times to ask engineers to spawn all cores with dummy loops to please this type of audience. The problem is that it benefits no one. So on more constructive note: there easily could be some problem (or many different problems) in multiplayer causing serious performance degration and that certainly is high prioirty to track down and fix asap for us. It is just there is no magical multicore optimization that could out of sudden improve everything available as some may seem to hope for when they see some of their cores are nearly idle most of the time. Most important is this: there is no single "general performance problem of CPU or GPU utlization", there are only specific problems (and these may easily be completely different in mp, sp, different depending on users hw etc.) that need to be fixed.

I just want to point out that I never called your engineers lazy or incompetent. I have respect for you and BI as a whole. I may not agree with your priorities or your viewpoints on issue's, but that doesn't mean that I don't respect the work that you do. Please don't try to lump me into some troll group.

Performance overall is important, not just "occasional significant drops". Sure a significant drop might indicate a problem, and I'm not trying to argue or say that it's not, but if your software runs terrible or badly all of the time with poor hardware utilization constantly, that also does indicate a problem. As for utilization drops, they happen A LOT.

What is this fascination with spawning dummy loops to spike usage to 100%? I've said time and time again that I don't expect 100% core usage, and I rarely if ever see people say "Why does ArmA 3 not use 100% cpu usage?". What I see a lot of though is "Why is my CPU usage so low?", I.E. why is it lower than watching a Youtube video, when my Performance is so bad. That is a valid question and a concern. I honestly would like an answer to it, seeing as how it's been a problem since all the way back in ArmA 2.

I'm not reading between the lines as far as coming to the conclusion that you think that better multi-threading and better parallel processing would not help performance. You yourself keep saying that Multicore is not the solution, yet Intel and AMD are busy putting out multicore processors with more and more core's on them. I never said you were not interested in or trying hard to improve performance. But what you are telling me is basically that I need a 10Ghz single or dual core cpu in order to run ArmA, and that's just something that's not going to happen, probably ever.

There are core issue's at play here, that have been in play for a very long time even since before ArmA 2 and I feel like the general consensus is to sweep it under the rug rather than fix it. Judging from your responses, in that you don't feel that how modern technology is moving into multicore CPU's and parallel processing is of benefit to this engine, I don't know if these issue's will ever be fixed because I don't actually believe that you see them as issue's or accept the reality of them.

Share this post


Link to post
Share on other sites
Man i'm tired of people saying, that they have poor performance, everybody knows the game is cpu heavy, Dev knows, we know, but solutions... buy the better proc you can afford...I don't know if people notice, but arma have a view of 2km (average player setting max 15 or 12...) and all other game have 500 m on average, some like crysis (looks good, but have a render of 30 meter....)I think the problem is that, and when you put arma and other side by side, other have a much little vision range even in the bigger maps in bf, cod crysis etc is a joke the render distance, also arma have to manage a LOT of simulation/Ai no other game have ballistics like arma, penetration diversion ricochet etc...and the Ai of arma, is one of the best..(I think)

The problem is that the good cpu's still dont add up in some cases. And that is not the solution, i am on the high end of the recommended requirements yet i am at 10 - 15 FPS. Don't talk about something if you cannot offer something constructive.

Share this post


Link to post
Share on other sites

CPU - 25% usage, GPU - 45% usage. i7 3770k @ 4.5ghz, GTX 780 Dir CU II = 30 fps @ 1080p. Screw this alpha stage game, such a waste of money.

Share this post


Link to post
Share on other sites
CPU - 25% usage, GPU - 45% usage. i7 3770k @ 4.5ghz, GTX 780 Dir CU II = 30 fps @ 1080p. Screw this alpha stage game, such a waste of money.

Errr not the most constructive post in an otherwise very interesting thread. Kind of embarrassing to read actually given how Maruk is reaching out to us. But anyway, I really appreciate Maruk talking about this, it is very interesting.

I do bet though that Mr theOPO was playing online when he experienced that performance. I think this is the biggest puzzle for me, how performance behaves online, not just for Arma3 but Arma2 also. The way the server seems to dictate the Client FPS. I think this is what baffles most people and creates a lot of the "my CPU is not being used threads". Can any improvements be made in this area?

Will any of the improvements from DayZ's multiplayer Architecture be able to translate over to Arma3 at all when they have it working? Or can that never happen?

Thanks again Maruk for trying to answer some of these questions.

Share this post


Link to post
Share on other sites
Man i'm tired of people saying, that they have poor performance, everybody knows the game is cpu heavy, Dev knows, we know, but solutions... buy the better proc you can afford...I don't know if people notice, but arma have a view of 2km (average player setting max 15 or 12...) and all other game have 500 m on average, some like crysis (looks good, but have a render of 30 meter....)I think the problem is that, and when you put arma and other side by side, other have a much little vision range even in the bigger maps in bf, cod crysis etc is a joke the render distance, also arma have to manage a LOT of simulation/Ai no other game have ballistics like arma, penetration diversion ricochet etc...and the Ai of arma, is one of the best..(I think)

Anyone with any basic knowledge of LOD object streaming will say view distance does not matter (as the main culprit), especially when 85% of the island is farm land.

Like I said. In multiplayer I get 25 fps on lowest AND 25 fps on ultra.

I'm on an i7 quad at 2.8ghz with a GTX 560 TI

there is absolutely no reason I shouldn't be able to play at 60 fps even if I tone some settings down, but thats IMPOSSIBLE, its cpu heavy regardless for some other reason. I never buy games expecting to max them because I have a semi-dated card, but I'm damn well sure I can get it to look good and run at 60 fps. Thats not the case with Arma

Edited by Alabatross

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
Sign in to follow this  

×