dnk
Member-
Content Count
578 -
Joined
-
Last visited
-
Medals
Everything posted by dnk
-
That makes no sense. It's like saying "getting 5 MPG means a car is fast". I hope you can see the fallacy. ---------- Post added at 01:56 PM ---------- Previous post was at 01:55 PM ---------- Well, my prior comments plus a third.
-
Unfortunately, I can't really dig into it as my CPU is not bottlenecking things in any way, given the disparity in CPU and GPU on my machine. Or is it...Specs: Settings: Here's mine in Girna: FPS avg: 22FPS Same setup, now in Agia Marina: Again, trying to mimic original movements, but no AI: And now for A2:CO+ACE2 Setup: Results: My impression: On my system, the game is more multicore-friendly than I said previously. I guess I need to do more testing, but this is about perfect. The GPU is fairly maxed out throughout, but that's to be expected. Oddly, it seems that the more the disk is reading, the better the performance, but that's only with 2 examples, so another might have different results. Generally, that happens when in a city, where much of the long-range geometry/objects are obscured by buildings yet lots of textures/models need to loaded and the view shifts around a lot (at longer ranges, you tend to move your head around less when engaged). Obviously, with no AI things run more smoothly (about 45% more smoothly). Even so, inactive or unengaged AI have a smaller footprint on performance than actively engaged ones. I see no GPU unterutilization without AI, but about 10-15% underutilization with 30-70 AI. This is fairly minor, imo, but it seems with better cards it's worse. It might not just be the AI routines and pathfinding that are causing the "underutilization", but all those bullet trajectories and FX particles. All I can say for sure is that all my cores are being used equally. Lastly, though the sample size of tests is small, A2:CO is clearly utilizing the GPU far less than A3:Alpha, and the FPS in A3 without AI ends up almost equalling that of a far more bare Takistan with AI. To claim the devs haven't made any progress in the engine/code is, at least here, totally incorrect. Stop using 6-cores as a bench for how "multicore" the game is. Quad/dual is far more important for the game. Besides, BF has no AI basically.
-
This this this this this! Good response. Thank you for your contributions to the thread. Being the one who mentioned the release date, I will have to agree to this. Sticking with 32-bit on such a massive streaming game is fairly short-sighted. Whatever defenses I've made for the development, this is something that really needed to be fixed. With multicore support, they can at least do a half job (and quantifiably I think they have) and get tangible results, which is why I've been more forgiving, but here that's impossible, and the LAA solution was already in place and doesn't cut it for 2013 anyway. Well, first Arma has always been a CPU-heavy game. There is just so much geometry to deal with on 3km+ draw distances, plus having 50-150 AI which actually have to do things more complicated than following simple scripts, you can't get the same results as other shooters with vastly smaller draw distances, map sizes, and AI that have a lot less to "think about".And, I'm getting tired of explaining this to you, but the devs have admitted there's an issue. In this thread. And what do you want them to do exactly? They're busy, you know, designing a game. They're not PR spokesmen. Your sense of entitlement is bewildering: "I need devs tell me every day my concerns are valid!"
-
Jumping would just end up killing immersion for me. The current "jump over" animation is much improved and faster than A2's, and I'm happy with it, plus it does allow you to "climb up" low obstacles. All we need is for something to climb up slightly higher obstacles.
-
The tools I used for OA aren't working anymore for A3. How are you guys unpacking these? Link to program? Thanks. FIXED - see next post k, that first one worked. I got BI Tools 2.5.1 now also, but when I try to use binmake or cfcconvert, it doesn't work. So... how to edit .bin files?
-
k, that first one worked. I got BI Tools 2.5.1 now also, but when I try to use binmake or cfcconvert, it doesn't work. So... how to edit .bin files?
-
60 on what rigs? What graphical settings? By all means, A3 can be a 60FPS game right now. And let's remember this game isn't coming out for another 6-12-18 months. The average hardware is going to be better by the time it hits its final release, as is the code (even if it's just fixing a few loose ends and light optimization, this combo will get 30-40FPS results). I mean, I can get up to 45 or so in A3 I think on lowest settings if I wanted to play an ugly game with only a few AI not in Agia. I chose 15-40 as an acceptable compromise between appearance and performance (it's not too much compromise at that). I think most casual players (read: not the gaming elitists complaining on these threads) with casual gaming rigs (not the ones complaining about "under-utilization"), which are sporting mid-level GPUs, 2-4-core 2-3GHz CPUs, and 4-8GB of RAM aren't going to mind picking up a brand new title and getting 30-40FPS. This is the market, not tet5uo's need for superfluid motion. Looking over Steam's stats again, between a 650ti and 560 is the average card. That's enough for 30-40FPS, or sacrificing quality for 40-60FPS. The vast majority of gamers don't want to pay a lot for their PCs, as the stats show. Only 20% own high-end GPUs (GTX660+). Only 3% have 6 or 8 cores. Only 12% have 3.3GHz+ CPUs. The only thing people have a lot of is RAM, and guess what? It's the value buy these days. And it makes sense because at least 80% of gamers are value gamers, and even the remaining 20% have a considerable amount that are borderline.
-
Getting rid of the Grass Layer: Yes / No as well as an Idea
dnk replied to instagoat's topic in ARMA 3 - BETA DISCUSSION
Ground is camoflaging, the sinking is in the engine, I don't think it can be disabled through mods. If it could, it probably would've by someone by now. -
I've only found aimingspeed and aimingaccuracy to have any effect. 0.1/0.3 is perfect for those two, with 0.15/0.45 being for a "good" AI.
-
@white 30FPS minimum in what settings and missions? I can manage that with a 3.3GHz 4x CPU and moderate settings. I'm actually fairly amazed at how well the game runs on my system, given I get ~40FPS in A2:CO-ACE2 with similar setups. Really, the game is a good improvement, and I've had only a 25% reduction in performance. Kudos to the devs! Regarding the multicore usage, well I would say about 40% of the processing is done on 1 core, with the other 60% spread out on the other three. Perhaps it's closer to 50-50; I haven't bothered to be exact in my findings. Either way, that stuff being sent to the other cores isn't exactly insignificant. That's pretty multicore, and just because it's clouds and physx doesn't make it less important than AI or other engine components. It's still processing being offloaded onto other cores that otherwise would have to compete for the first core's resources. If you compare a perfectly monocore process to a perfectly quadcore one and place this in the middle, it's about 40% of perfectly quadcore, 24% of 6-core, and 18% of 8-core. I'm not sure why they'd need to optimize/code a game for 6- and 8-core CPUs, though, when 4-core is pretty much the standard and required specs. http://store.steampowered.com/hwsurvey/cpus/ 6- and 8- core CPUs just aren't popular. 90% of Steam users are on 2- and 4- core computers. Why should BIS care about the top -literally- 3% of users? --- Note that I agree that the devs should be making it more multicore, and agree that the AI routines should be individually threaded and able to be placed onto other cores. I'm just disagreeing with what I see as hyperbole because (your) superdupercuttingedge 8 core wasn't what the devs designed this game for. People with moderate rigs are going to get acceptable performance, and that's the vast bulk of their clientele. I'm sure A4, or perhaps an expansion, will work on full multicore handling for 6/8 cores once that actually becomes a commonality in the marketplace. Right now, it's an extreme fringe, and a smaller development studio can't really cater to it. These guys don't have Crytek budgets, and they can't release top-end software and rewrite the engine every 3 years from scratch. They have to balance their resources, and likely sorting out the AI, adding new features, and updating the graphics are all far more important than satisfying elite gamers' high end purchases. If this is an apologia or excuse-making, so be it. I'm tired of hearing people with $1500 rigs bemoaning the fact that BIS can't keep up with the trends of $60/100M budget games. I'm sure there's much that can be done (and with luck WILL) before the release to improve performance, especially for MP, but if the game maxes out at 50FPS for the best rigs and plays at 40FPS for midrange rigs, I think the devs will be happy since 90% of their customers will also be happy with that (the silent majority, as it were). I'll sure be happy with that.
-
Even the original STALKER did rain well enough using bumpmaps and gloss factor and the environment map (that small texture that tells you what reflections look like, whatever it's called). Basically, set the gloss factor too high and get a sort of gray gradient for that map and things start to look really wet without any performance cost that I'm aware of. I would be amazed if BIS couldn't get that into the renderer.
-
In the future apparently human beings are not subject to inertia or weight.
dnk replied to pd3's topic in ARMA 3 - GENERAL
No, it isn't. It's actually a bit slower for some of the paces. There's a video in this thread I made comparing the two games. The results:Crouch: 4% faster in A3 Run: 5% slower in A3 Tactical Pace: 30% slower in A3 Sprint: 5% faster in A3 -
And, yeah, I really love the changes to movement overall, though they could fine tune some things (like having the tacpace button switch into it for both walking and running instead of just running). They just need to add some restraint to the turning speed is all.
-
Getting rid of the Grass Layer: Yes / No as well as an Idea
dnk replied to instagoat's topic in ARMA 3 - BETA DISCUSSION
We have a long topic on this already, with a good solution that's gotten a ton of upvotes and is currently "assigned"/"open" on the tracker. This is somewhat similar to yours, but adds in a blending texture that looks like grass to make the model harder to pick out as its geometry does not so easily give it away. http://forums.bistudio.com/showthread.php?148861-Rendering-grass-at-long-distances-My-thoughts-about-it The OP's idea to use an alpha texture is so simple and light on performance (as is yours). Here it is in pictures: http://s1.directupload.net/images/130309/uaklkkkf.jpg (266 kB) And here are my two additions to the topic: http://forums.bistudio.com/showthread.php?148861-Rendering-grass-at-long-distances-My-thoughts-about-it&p=2335978&viewfull=1#post2335978 http://forums.bistudio.com/showthread.php?148861-Rendering-grass-at-long-distances-My-thoughts-about-it&p=2336692&viewfull=1#post2336692 And here's the tracker: http://feedback.arma3.com/view.php?id=3505 (4th place in upvotes, btw) ADDITIONALLY, SQB_SMA has release this addon for the midrange textures that makes spotting people a LOT harder at range by giving a bunch of similarly-shaded background noise. I think it needs to be toned down a bit, but here's a pic for reference: https://dl.dropbox.com/u/106806270/arma3%20midtex/Ultra%20Midtex/4/ultra4.png (1640 kB) Try and spot people in that! Here's the comparison post in the thread: http://forums.bistudio.com/showthread.php?150645-midrange-terrain-texture-replacement&p=2352520&viewfull=1#post2352520 -
So, I see we have old modders being included into the dev team, but are we going to see the mods already being made incorporated into the beta/final? I mean, there are a lot of good things being done already that I think really improve things, especially regarding midrange textures and, of course, the old A2 mods like ACRE and 3rd person view change and such. Any chance of seeing these ingame at release if they're top-notch, or is the dev team going to do it all by themselves and just take mods for their inspirational/copyable content? Obviously, I'm sort of asking the dev team specifically here, but I think it'd be good for the community as a whole to know, hence not going through PM. Also, a "yes" could help generate even more modding, which is always a good thing, and get current modders to step their game up even more, also always a good thing.
-
midrange terrain texture replacement
dnk replied to bad benson's topic in ARMA 3 - ADDONS & MODS: COMPLETE
That's great! Especially for concealment from humans. Only problem is that it's hard to enforce such mods on players, so eh. Regardless, these ultras look great. I think there's a little too much contrast - too black - so maybe toning it down by 33% or so and blending the dark patches a bit more would help smooth it a little, though don't loose the contrasty look altogether! Just subdue it a bit. My 2c, but keep up the good work! -
Arma 3 Alpha lacking Arma 2's immersion and fear factor.
dnk replied to fixmygame's topic in ARMA 3 - GENERAL
Point was that there has NEVER been a river in an ArmA release because it doesn't seem to be in the engine's abilities, which is a massive limitation. I know I've seen ponds above sea level, so it might be doable, but apparently no one's thought it important enough or good looking enough to implement. -
Can you look into your crystal ball and tell me what I'm going to do in 6 months, too?Now, regarding the psycho: Except it does have multicore support, and my own quad-core i5 is seeing at least 50% of each core used, I think. That's not "support for multiple cores"? Interesting usage of language. He goes on to again literally claim there's only 1 core being used. This is just factually inaccurate, and guess what? Quoting people with huge factual inaccuracies as though they're the truth makes you look like an ignorant ranting fool too. To let the community give feedback and help with development and bug testing? And to get some money and ensure sales while they have a high media profile, yes, but anyone considering purchasing at this point should be doing so solely for that first reason. That's what an "alpha" is? Someone (including guusert) needs a dictionary. The game literally says this almost word-for-word during the loading screen. It's a sale, like a half-off preorder where you get to start playing the new content as it comes out. It's half price. Let's remember that: half price, on a huge sale, and you get immediate access to the game instead of waiting another 6-12 months for the final release. Where else in the world do you get to buy things 50% off and use them immediately and then get to return them? Try returning sale items that you've used for a week+ at any clothing store or electronics store and see what the results are, especially with software, which basically all stores have a non-return policy on.These are the sorts of comments only children make because anyone who has actually lived on their own for an extended period knows that you don't just get to return anything you buy under any conditions. They also know to look for whether something is returnable or not before buying if they think they might need/want to return it. They also do research to see if what they're buying is something they really want before purchasing. Sane adults do not just go out and buy, say, a TV on sale that's non-returnable, take it home and install it for 1 week just to see if it works well. Why does this work differently with the Internet?
-
Ambience and Weapon sound out of balance a bit?
dnk replied to BorisLozac's topic in ARMA 3 - BETA DISCUSSION
Heh, guns are still too quiet compared to ambiance to be realistic, but okay a slider would be nice. -
Important with clouds if you're in an air battle (not ArmA's strong suit at all, but still).
-
Arma 3 Alpha lacking Arma 2's immersion and fear factor.
dnk replied to fixmygame's topic in ARMA 3 - GENERAL
Eh, I just miss Sahrani. Such geographic/cultural range was nice, being able to transition from this Mediterranean type climate to a middle easternish dry one to a large port city to swiss alps and Canadian rockies all in one map... Now we get more realistic monothemed ones. Eh. Progress? On that note, WHERE ARE THE RIVERS? Such a pet peeve of mine to have a milsim without rivers, brooks, streams, etc, just mountains and dry gorges since forever. I do wish we had more cities, but I think the full map should be more interesting. -
Those new midranges look real tight! They fit the landscape so well. Still not really loving the long-distance, though. I think BIS just needs to decrease the contrast/alteration a bit, but like a -20% tweak, so I'm going to be hard to please on that since I don't really want a full redo :)
-
Agree, I certainly have been critical and pushed for changes. I wouldn't ever suggest doing otherwise. But the fact remains that BIS can basically release the A2 engine v1.5 with minimal improvements to performance (though I think the gameplay, physics, graphics, and so on have all been well improved) and a lot of people are going to end up buying it because, really, what else are they going to play like A3? I'm going to preorder it as soon as my PayPal funds come through despite knowing full in advance that there are these issues (though I assume they will be at least partially fixed) because, really, what else am I going to play? FYI, this has been getting a TON of press on "PC Gamer", and that plus all the DayZ hype means we're getting a lot more... for the lack of a better word... "dumb 13-yo kids" coming in here now, posting their idiocy 10-20 times, and leaving to go read their webcomics and do their homework, never to return.
-
Hmmm, I may have made a mistake. I think running in A2 is 19.5s, not 15.0. I mixed up clips. Doh. That does make the run speed for A3 more than marginally faster, so my video is incorrect. That said, you guys do know there's a tactical pace in A2 also, right? Bring up your scope while in running mode, it's halfway between run and walk. It might be faster here, but I forgot to check it. A3 tactical pace = 45s A2 tactical pace = 31.5s Difference = 70% as fast
-
There is a tracker for this. Footsteps don't get heard on certain terrains.