@Kalabalana: I have no problem finding servers that can hold the tick rate of 30 even in end game. Given you don't play on servers that are > 18 players. Also the player-updates are decoupled from the server tick-rate. A higher tick rate does not improve hit reg, while a lower tick rate can cause problems. But again, this doesn't happen on servers with a reasonable player count.
Also keep in mind, that NS2 is not comparable to any kind of shooter. What does the server need to calculate in BF3? Player-Positions, -Directions and Status. Vehicles. Projectiles. CapturePoints. Thats it. NS2 has much more game logic than that. Unmanned vehicles don't create more load than dropped weapons. Manned vehicles don't create more load as the player already does. But in NS2 you got so many dynamic things, like buildings and logic connected to this buildings, that it needs more CPU-power to calculate it. On top of that comes the LUA that isn't the fastest.
If you look into this game in more detail, you will see, that it has simply more demand than the default shooter. It is heavier to calculate.
This is all besides my original point, and also incorrect. (Also, I never brought up hit registration, if you had played this game back in the alpha and beta, you'd realize just how far we've come.)
For example: Natural Selection itself had similar server-side requirements (very similar). There has been no upgrades in the quality of aiming/player/structure/voip etc information that would represent a tangible performance decrease like we see in ns2 (Also, the netcode isn't going to be fundamentally different mechanic from game to game, and if it is, that's by choice/design, not because of some inherent restrictions in the latest technology, especially not in Max's work which would normally be seen in a much larger studio.).
As far as BF3 is concerned, I think if we could actually look at the numbers, BF3 prob has a lot more processing, and data being pushed around than NS2, and even if it didn't, it would be in the same ballpark, and it does so, smoothly. (Consider the average dynamic objects, and amount of information a person requires at any given time in both games, on average, the bf3 player will be looking at more things, players, bullets (they aren't hitscan in BF3 like they are in ns2), vehicles (which are interactive and physics based keep in mind), more minimap objects, the slew of personal devices the classes have and use at all time. Just a UAV alone looking at a map of players is massive load on the server than NS2 gets to avoid. Also, the game doesn't slow down for some shadow reason like NS2 does. Heuristics, algorithms, great prioritizing, BF3 prob had more hands dip into the netcode process than UWE has on staff, and with years of experience to boot!
Thank you for recommending that I look into this game in more detail, I will indeed, though I'm inclined to believe the depths to which we see are not the same.
**Though I have heard the LUA layer, or maybe I'm using the terminology wrong, is a cause for concern, and I believe the team is actually writing a code-bridge between the engine and the LUA code to increase speed? That's a question for people who know what I'm referring to though.
well dunno why such gets addeded to this game - actually as long as performance issues are out with the existing build theres NO REASON for such renderstuff. the time should be used for something useful.
Sure mate - Spark uses 'Direct3D9' for part of its graphics system. DirectX is a broader standard, encompassing DirectSound, DirectInput, and other bits and bobs. These days, engines are often referred to as being 'DirectX9/10/11' even if it's just part of their rendering process. So while not strictly, technically correct, it would be colloquially acceptable to refer to Spark as 'DirectX9' engine. Similarly, while Direct3D11 will only make up part of CryEngine 3, people in the gaming work refer to CryEngine 3 as a DirectX11 engine.
Now, why doesn't Spark use Direct3D11? Wasn't Direct3D9 released way back before humans started using stone tools?
That's a very complex question, and one that honestly requires a blog post on unknownworlds.com to do it justice. But I'll use another engine analogy here, to give a rough idea.
Internal combustion engines are systems that have been around for a very long time. Most engines used in cars, bar some fun oddities (I'm looking at you, rotaries!) are four stoke piston engines that operate using the suck-squeeze-bang-blow sequence.
Ever since the four-stroke was invented, new ancillary systems have been popping up to improve it. Sometimes, these improvements are clear-cut wins, and not incorporating them into a vehicle would be foolish. For example, lead-free fuels.
Other ancillary systems are not such clear-cut wins. For example, turbochargers and V-cylinder-layouts. Adding a turbocharger to an engine can have benefits: Recovering the energy from exhaust gases can increase power output and reduce fuel consumption. But they also add weight, slow throttle response and dull exhaust notes. A V layout has packaging and vibration-balance advantages, but can also increase weight and complicate integration of other systems.
So, while V's and turbochargers are considered 'modern' and 'cool' features, flat-layout naturally aspirated engines can still stomp them when properly implemented. No one would doubt the powerful yet efficient delivery of the VW TwinCharger engines, or the world crushing power of the Veyron's W16 Quad-turbo. But equally the 'older' technology used in the AMG M156 or BMW S62
engines do not stop them being equally, if not more brilliant.
So Direct3D11 - There are differences, and they can make rendering better. But they are not sure wins, and they are not necessarily the most efficient wins UWE could strive for. In practice, the Direct3D9 pieces of Spark could be converted to Direct3D11. With work, that could bring certain benefits to rendering. But Direct3D11 is not lead-free fuel. So greater benefits could be achieved by perfecting natural aspiration, and that is the path we are going down.
Remember too, there is an element of marketing in the Direct3D line. Microsoft, hardware vendors and yes game developers have inventive to attach a new, shiny brand to their latest products. To the less well informed customer, this brand can make 'superceded' product look inferior.
A topic that we shall no doubt have to revisit!
Edited for spelling.
Except that computers are absolutely, 100% nothing like cars. Most games that use D3D11 retain the option to use D3D9 for compatibility reasons.
Except that computers are absolutely, 100% nothing like cars. Most games that use D3D11 retain the option to use D3D9 for compatibility reasons.
I think he mean't it as a reference...DX11 doesn't just make things better, it has its "weight" as he would like to call it. He also says that marketing makes DX11 seem like a light sabor next to a sword compared to dx9., when there really isn't much different. And after reading some articles and changing settings in games comparing the two...have to say I have to agree. There really isn't THAT big of a difference.
But you knew it was a reference....didnt you? Sure yeah did.
Except that computers are absolutely, 100% nothing like cars. Most games that use D3D11 retain the option to use D3D9 for compatibility reasons.
I think he mean't it as a reference...DX11 doesn't just make things better, it has its "weight" as he would like to call it. He also says that marketing makes DX11 seem like a light sabor next to a sword compared to dx9., when there really isn't much different. And after reading some articles and changing settings in games comparing the two...have to say I have to agree. There really isn't THAT big of a difference.
But you knew it was a reference....didnt you? Sure yeah did.
Except that computers are absolutely, 100% nothing like cars. Most games that use D3D11 retain the option to use D3D9 for compatibility reasons.
I think he mean't it as a reference...DX11 doesn't just make things better, it has its "weight" as he would like to call it. He also says that marketing makes DX11 seem like a light sabor next to a sword compared to dx9., when there really isn't much different. And after reading some articles and changing settings in games comparing the two...have to say I have to agree. There really isn't THAT big of a difference.
Depends. Every time I switched to a DX10+ renderer compared to the DX9 renderer in games, I rather got quite some performance improvements than new effects.
L.A. Noire used to run with ~25-30 fps on medium details for me. Then it got a patch that added DX10 support and as soon as I enabled that, I could crank up the details to maximum and still got fluent 45 fps out of it.
Except that computers are absolutely, 100% nothing like cars. Most games that use D3D11 retain the option to use D3D9 for compatibility reasons.
I think he mean't it as a reference...DX11 doesn't just make things better, it has its "weight" as he would like to call it. He also says that marketing makes DX11 seem like a light sabor next to a sword compared to dx9., when there really isn't much different. And after reading some articles and changing settings in games comparing the two...have to say I have to agree. There really isn't THAT big of a difference.
Depends. Every time I switched to a DX10+ renderer compared to the DX9 renderer in games, I rather got quite some performance improvements than new effects.
The thing is though, as we all know the FPS issues are mostly the result of the game code eating up CPU cycles. The renderer is quite fast, so I doubt switching the DX tech would change that...
Except that computers are absolutely, 100% nothing like cars. Most games that use D3D11 retain the option to use D3D9 for compatibility reasons/
In your infinite wisdom you appear to have developed a theory for why the use of analogy is irrelevant when explaining complex concepts. Seeing as humanity has developed quite a taste for that device, you owe it to us all to share this theory.
Of course, you don't have such a theory, and if you did it would probably hold less water than my CPU waterblock. So don't make disparaging, time-wasting comments on posts I've taken time to write to explain complex development concepts. Some people don't presume to know everything about everything, and want to engage in useful discussions about interesting concepts and issues.
for those of you narcissistic types out there, you can now stare at yourself in shiny objects!
Sorry, the current reflections are static - they only reflex (approximately) the static geometry. So you can't see yourself or approaching enemies in the reflections ...
OTOH, they don't make much of a difference FPS wise in-game (because the renderer thread is not usually the bottleneck for fps once you start playing anyhow).
Still better than bloody source cubemaps, god I hated those things. They look nice, interested to see how they look on alien structures.
I believe its open gl, or a custom renderer. spark seems uses a few features from dx (?) but given its inherent problems with transparency and complex surface shaders, I'd say its not a large portion of the engine. its what I've heard and I don't really understand it all myself, so you could be right. again any news on when decent transparency shader will be added to the engine?
I don't think you can just 'use a few features from dx'.
Either you use opengl, or you use directx, pretty sure you can't just tack half of one onto the other and have it work, and microsoft would murder you if you started pulling their directx to bits and sticking it in your opengl distribution.
Except that computers are absolutely, 100% nothing like cars. Most games that use D3D11 retain the option to use D3D9 for compatibility reasons/
In your infinite wisdom you appear to have developed a theory for why the use of analogy is irrelevant when explaining complex concepts. Seeing as humanity has developed quite a taste for that device, you owe it to us all to share this theory.
Of course, you don't have such a theory, and if you did it would probably hold less water than my CPU waterblock. So don't make disparaging, time-wasting comments on posts I've taken time to write to explain complex development concepts. Some people don't presume to know everything about everything, and want to engage in useful discussions about interesting concepts and issues.
You frequent these forums enough you should know his mode of operation.
Make obvious, cynical, snarky comment normally coupled with some type of ad hominem or anecdotal buffer that provides no real insight or value to the conversation. If these forums had a ignore posts by user feature, he would be the first on mine.
Except that computers are absolutely, 100% nothing like cars. Most games that use D3D11 retain the option to use D3D9 for compatibility reasons/
In your infinite wisdom you appear to have developed a theory for why the use of analogy is irrelevant when explaining complex concepts. Seeing as humanity has developed quite a taste for that device, you owe it to us all to share this theory.
Of course, you don't have such a theory, and if you did it would probably hold less water than my CPU waterblock. So don't make disparaging, time-wasting comments on posts I've taken time to write to explain complex development concepts. Some people don't presume to know everything about everything, and want to engage in useful discussions about interesting concepts and issues.
Analogies are only relevant when they fit. I have never in my life seen a car/computer analogy that fit. The two technologies are almost complete opposites in their evolution and implementation making it completely pointless to compare them. I'm sure you put a lot of effort into your post but there's really no need to throw a hissy fit when someone challenges what you've said. Why not instead reply to my remark of the users ability to pick between D3D9 and D3D11. Also, while D3D11 almost assuredly isn't the silver bullet for performance issues it adds some very nice features like MSAA support for deferred rendering. I'm not saying that UWE should prioritize DX11 support or even saying that you should implement it. I'm simply saying it's much different from picking parts for a car.
Depends. Every time I switched to a DX10+ renderer compared to the DX9 renderer in games, I rather got quite some performance improvements than new effects.
The thing is though, as we all know the FPS issues are mostly the result of the game code eating up CPU cycles. The renderer is quite fast, so I doubt switching the DX tech would change that...
It's not just about rendering. DX10+ can also work around the memory problem that 32 bit OS users have, which probably played quite an important part in my L.A. Noire example as well.
Under DX9, video memory has to be mirrored in the address space of system memory. So if you have a graphics card with 1 GB VRAM and all of that is used by the game, you end up with effectively 1 GB less RAM for actual game memory. Add the fact that applications are usually limited to 2 GB of system memory on 32 bit OS and you quickly get into the critical area.
The problem no longer persists under a 64 bit OS, not only because the memory limit per application is higher and there is generally more potential RAM available but simply because the VRAM no longer needs to be mirrored in normal RAM. DX10+ can achieve the same effect under a 32 bit OS already, though.
So it's probably a feature that would only benefit a small niche of players (32 bit OS users that are able to run DX10 or DX11, i.e. Win Vista or 7 in 32 bit plus a compatible graphics card), but it's still a feature.
In your infinite wisdom you appear to have developed a theory for why the use of analogy is irrelevant when explaining complex concepts. Seeing as humanity has developed quite a taste for that device, you owe it to us all to share this theory.
Analogies compare two things that usually have no physical or logical relation to each other. When used well, you paint a mental picture of something that isn't easy to visualize and it may even act as a mnemonic device to aid memory.
Now for the dark side of analogies. The reason DX9 was chosen is pretty simple; compatibility with older hardware and windows xp seemed to be important with a target release of 2008-ish; DX10 and 11 offer some features that are marginally useful and some better performance, but nothing huge. When you explain something that's not complex by means of an analogy, you add lots of words and make your post convoluted and harder to read.
Poor analogies also risk spreading confusion; two decisions could look similar even if the logic behind them is not the same. If you extrapolate from an analogy beyond its narrow scope where it applies you might end up with word sallad. This is particularly dangerous when discussion complex subjects, because it's not always easy to notice when the analogy has run off the rails.
Software has some very strange properties that do not apply to anything else. The big cost in software is design, code maintenance/QA and runtime resources. Nobody in their right mind would have a car with 2 or 3 different versions of the engine, not because it's costly to design the engines, but because they weigh too much; but for software it's fairly cheap and routinely done. Each additional unit of software costs almost nothing to make; each additional car engine is very expensive to manufacture. There's just too many ways to go wrong when comparing software to the physical world.
That's also the reason analogies are the favourite device of charlatans. If you see creationists in a debate, it is often nothing but a long string of analogies. It's very easy to bamboozle, confuse and mislead with analogies; that's why they like them.
@Kalabalana: I have no problem finding servers that can hold the tick rate of 30 even in end game. Given you don't play on servers that are > 18 players. Also the player-updates are decoupled from the server tick-rate. A higher tick rate does not improve hit reg, while a lower tick rate can cause problems. But again, this doesn't happen on servers with a reasonable player count.
Also keep in mind, that NS2 is not comparable to any kind of shooter. What does the server need to calculate in BF3? Player-Positions, -Directions and Status. Vehicles. Projectiles. CapturePoints. Thats it. NS2 has much more game logic than that. Unmanned vehicles don't create more load than dropped weapons. Manned vehicles don't create more load as the player already does. But in NS2 you got so many dynamic things, like buildings and logic connected to this buildings, that it needs more CPU-power to calculate it. On top of that comes the LUA that isn't the fastest.
If you look into this game in more detail, you will see, that it has simply more demand than the default shooter. It is heavier to calculate.
This is all besides my original point, and also incorrect. (Also, I never brought up hit registration, if you had played this game back in the alpha and beta, you'd realize just how far we've come.)
For example: Natural Selection itself had similar server-side requirements (very similar). There has been no upgrades in the quality of aiming/player/structure/voip etc information that would represent a tangible performance decrease like we see in ns2 (Also, the netcode isn't going to be fundamentally different mechanic from game to game, and if it is, that's by choice/design, not because of some inherent restrictions in the latest technology, especially not in Max's work which would normally be seen in a much larger studio.).
As far as BF3 is concerned, I think if we could actually look at the numbers, BF3 prob has a lot more processing, and data being pushed around than NS2, and even if it didn't, it would be in the same ballpark, and it does so, smoothly. (Consider the average dynamic objects, and amount of information a person requires at any given time in both games, on average, the bf3 player will be looking at more things, players, bullets (they aren't hitscan in BF3 like they are in ns2), vehicles (which are interactive and physics based keep in mind), more minimap objects, the slew of personal devices the classes have and use at all time. Just a UAV alone looking at a map of players is massive load on the server than NS2 gets to avoid. Also, the game doesn't slow down for some shadow reason like NS2 does. Heuristics, algorithms, great prioritizing, BF3 prob had more hands dip into the netcode process than UWE has on staff, and with years of experience to boot!
Thank you for recommending that I look into this game in more detail, I will indeed, though I'm inclined to believe the depths to which we see are not the same.
**Though I have heard the LUA layer, or maybe I'm using the terminology wrong, is a cause for concern, and I believe the team is actually writing a code-bridge between the engine and the LUA code to increase speed? That's a question for people who know what I'm referring to though.
First of all, I have played this game in the beta. And I know how far it has come.
Secondly, your main point was "The game has a 30 tick rate, which is a pretty big bottleneck" and this is simply wrong. You can easily search the forum for "tick rate" and find post from the devs, that explain why increasing the tick rate above 30 would not lead to a smoother experience for the players.
I may be wrong, that I thought you are pointing at hit-reg with your bottleneck theory. But that isn't important. Because as I wrote already, an normal server (<= 20 players) can hold the tick rate even in the end game.
I also don't know how deep your knowledge on network- or game engine-programming is. But projectiles in BF3 have not the same demand that complex objects like buildings have. You also forgot that NS2 has more such objects than NS1 had. Starting at cysts, power nodes, AI units and so on. The NS2 classic mod has shown that removing the cyst mechanic alone brings a huge performance boost.
Anyway. I'm on your side, that performance needs to get better and that the LUA code is one culprit. The news about a new and optimized LUA Virtual Machine are a bit old now (came up around release). I don't know if they are working on it.
Most people won't even be able to play with these kinds of settings/performance and you want people to record with them??.
Yes.
It's obvious why a 800 x 600 res low quality you tube video would not suffice for hard to see hit reg, so if you can only play the game like that you probably have large fps issues that are way more damaging to your game than the potential for slight hit reg issues.
However, If you can typically play in any reasonable resolution (graphic settings don't matter) above 30 fps, then you can record a demo which has zero performance impact. Then afterwards play the demo back while using some light weight video capture program.
Anecdotally, I have a four year old, entry level , first gen i7 and I can record with 1080 @ around 50 fps. But like I said this isn't the only method.
@ripsey: definitely multifaceted. But the scenario where you are aimed at an enemy model, fire, and do no damage (regardless of any circumstance) should not be occurring and is exactly what we can't seem to reproduce. That is precisely what hit reg refers to: being able to shoot at what you are pointing at..
Like I said there's suspicion.. but I personally believe it's due to things like the brain compensating for factors like lag compensation and input delay, feeling like something is slow/off and not as fluid as it should be. But actual hit reg issues *May* not be present anymore. (At least not common, after scrubbing through hours and hundreds of GB of hd footage frame by frame looking for what I swore up and down were hit reg issues while playing)
Out of curiosity, when this is implimented in the game will it ever mean that Marine placed structures and Kharaa placed infestation will be reflective? (Perhaps even units) Or will this likely just stay an environmental effect on the map's textures?
In your infinite wisdom you appear to have developed a theory for why the use of analogy is irrelevant when explaining complex concepts. Seeing as humanity has developed quite a taste for that device, you owe it to us all to share this theory.
Analogies compare two things that usually have no physical or logical relation to each other. When used well, you paint a mental picture of something that isn't easy to visualize and it may even act as a mnemonic device to aid memory.
Now for the dark side of analogies. The reason DX9 was chosen is pretty simple; compatibility with older hardware and windows xp seemed to be important with a target release of 2008-ish; DX10 and 11 offer some features that are marginally useful and some better performance, but nothing huge. When you explain something that's not complex by means of an analogy, you add lots of words and make your post convoluted and harder to read.
Poor analogies also risk spreading confusion; two decisions could look similar even if the logic behind them is not the same. If you extrapolate from an analogy beyond its narrow scope where it applies you might end up with word sallad. This is particularly dangerous when discussion complex subjects, because it's not always easy to notice when the analogy has run off the rails.
Software has some very strange properties that do not apply to anything else. The big cost in software is design, code maintenance/QA and runtime resources. Nobody in their right mind would have a car with 2 or 3 different versions of the engine, not because it's costly to design the engines, but because they weigh too much; but for software it's fairly cheap and routinely done. Each additional unit of software costs almost nothing to make; each additional car engine is very expensive to manufacture. There's just too many ways to go wrong when comparing software to the physical world.
That's also the reason analogies are the favourite device of charlatans. If you see creationists in a debate, it is often nothing but a long string of analogies. It's very easy to bamboozle, confuse and mislead with analogies; that's why they like them.
IronHorseDeveloper, QA Manager, Technical Support & contributorJoin Date: 2010-05-08Member: 71669Members, Super Administrators, Forum Admins, Forum Moderators, NS2 Developer, NS2 Playtester, Squad Five Blue, Subnautica Playtester, Subnautica PT Lead, Pistachionauts
edited February 2013
@soylent: seems like winxp is still largely used too, at 34% marketshare (compared to win7 at 39%) 0.0
I bet dx10 would at least bring some performance improvements.. but like someone else said, if the bottleneck is CPU you might not even see it / so it may not be worth the dev time.
Sorry to be a major killjoy but, as nice as it looks, I would have thought this would be waaay down on the list of priorities.
Your understanding of how game development works is severely lacking, and people reading your posts would do well to not confuse your assertiveness with wisdom.
Your understanding of what the community cares about appears to be severely lacking, especially considering your title at UWE as community organizer/PR/whatever else. Tweadle only expressed what the majority of this thread has already said. Considering this forum has been filled with performance and FPS related threads since beta, you had to know this viewpoint would be expressed, right?
I'm interested to hear more about how this update will make descent better, otherwise it seems just like another setting I will disable.
Your understanding of what the community cares about appears to be severely lacking, especially considering your title at UWE as community organizer/PR/whatever else. Tweadle only expressed what the majority of this thread has already said. Considering this forum has been filled with performance and FPS related threads since beta, you had to know this viewpoint would be expressed, right?
I'm interested to hear more about how this update will make descent better, otherwise it seems just like another setting I will disable.
Now you've done it you have spoken out against a dev prepare to be burned alive
Comments
This is all besides my original point, and also incorrect. (Also, I never brought up hit registration, if you had played this game back in the alpha and beta, you'd realize just how far we've come.)
For example: Natural Selection itself had similar server-side requirements (very similar). There has been no upgrades in the quality of aiming/player/structure/voip etc information that would represent a tangible performance decrease like we see in ns2 (Also, the netcode isn't going to be fundamentally different mechanic from game to game, and if it is, that's by choice/design, not because of some inherent restrictions in the latest technology, especially not in Max's work which would normally be seen in a much larger studio.).
As far as BF3 is concerned, I think if we could actually look at the numbers, BF3 prob has a lot more processing, and data being pushed around than NS2, and even if it didn't, it would be in the same ballpark, and it does so, smoothly. (Consider the average dynamic objects, and amount of information a person requires at any given time in both games, on average, the bf3 player will be looking at more things, players, bullets (they aren't hitscan in BF3 like they are in ns2), vehicles (which are interactive and physics based keep in mind), more minimap objects, the slew of personal devices the classes have and use at all time. Just a UAV alone looking at a map of players is massive load on the server than NS2 gets to avoid. Also, the game doesn't slow down for some shadow reason like NS2 does. Heuristics, algorithms, great prioritizing, BF3 prob had more hands dip into the netcode process than UWE has on staff, and with years of experience to boot!
Thank you for recommending that I look into this game in more detail, I will indeed, though I'm inclined to believe the depths to which we see are not the same.
**Though I have heard the LUA layer, or maybe I'm using the terminology wrong, is a cause for concern, and I believe the team is actually writing a code-bridge between the engine and the LUA code to increase speed? That's a question for people who know what I'm referring to though.
Except that computers are absolutely, 100% nothing like cars. Most games that use D3D11 retain the option to use D3D9 for compatibility reasons.
That was the exact tweet I shot back to them, LOL!
I think he mean't it as a reference...DX11 doesn't just make things better, it has its "weight" as he would like to call it. He also says that marketing makes DX11 seem like a light sabor next to a sword compared to dx9., when there really isn't much different. And after reading some articles and changing settings in games comparing the two...have to say I have to agree. There really isn't THAT big of a difference.
But you knew it was a reference....didnt you? Sure yeah did.
It's a pointless comparison.
How is it better or free-er? OpenGL comes with it's own unique set of problems.
Well, gj on improving engine!
Depends. Every time I switched to a DX10+ renderer compared to the DX9 renderer in games, I rather got quite some performance improvements than new effects.
L.A. Noire used to run with ~25-30 fps on medium details for me. Then it got a patch that added DX10 support and as soon as I enabled that, I could crank up the details to maximum and still got fluent 45 fps out of it.
In your infinite wisdom you appear to have developed a theory for why the use of analogy is irrelevant when explaining complex concepts. Seeing as humanity has developed quite a taste for that device, you owe it to us all to share this theory.
Of course, you don't have such a theory, and if you did it would probably hold less water than my CPU waterblock. So don't make disparaging, time-wasting comments on posts I've taken time to write to explain complex development concepts. Some people don't presume to know everything about everything, and want to engage in useful discussions about interesting concepts and issues.
Still better than bloody source cubemaps, god I hated those things. They look nice, interested to see how they look on alien structures.
I don't think you can just 'use a few features from dx'.
Either you use opengl, or you use directx, pretty sure you can't just tack half of one onto the other and have it work, and microsoft would murder you if you started pulling their directx to bits and sticking it in your opengl distribution.
A lot of work
You frequent these forums enough you should know his mode of operation.
Make obvious, cynical, snarky comment normally coupled with some type of ad hominem or anecdotal buffer that provides no real insight or value to the conversation. If these forums had a ignore posts by user feature, he would be the first on mine.
Analogies are only relevant when they fit. I have never in my life seen a car/computer analogy that fit. The two technologies are almost complete opposites in their evolution and implementation making it completely pointless to compare them. I'm sure you put a lot of effort into your post but there's really no need to throw a hissy fit when someone challenges what you've said. Why not instead reply to my remark of the users ability to pick between D3D9 and D3D11. Also, while D3D11 almost assuredly isn't the silver bullet for performance issues it adds some very nice features like MSAA support for deferred rendering. I'm not saying that UWE should prioritize DX11 support or even saying that you should implement it. I'm simply saying it's much different from picking parts for a car.
My understanding is the only change of any significance is the inclusion of hardware tessellation.
Supporting DX11 is a waste of time in 90% of gaming.
There were also changes from DX9 to DX10.
Nope
DX11 has many changes. They are just not always something which every game de can make use of.
But for defered rendering engines.. there area couple of things off the top of my head which are great from a dx9 to dx 11 transition.
It's not just about rendering. DX10+ can also work around the memory problem that 32 bit OS users have, which probably played quite an important part in my L.A. Noire example as well.
Under DX9, video memory has to be mirrored in the address space of system memory. So if you have a graphics card with 1 GB VRAM and all of that is used by the game, you end up with effectively 1 GB less RAM for actual game memory. Add the fact that applications are usually limited to 2 GB of system memory on 32 bit OS and you quickly get into the critical area.
The problem no longer persists under a 64 bit OS, not only because the memory limit per application is higher and there is generally more potential RAM available but simply because the VRAM no longer needs to be mirrored in normal RAM. DX10+ can achieve the same effect under a 32 bit OS already, though.
So it's probably a feature that would only benefit a small niche of players (32 bit OS users that are able to run DX10 or DX11, i.e. Win Vista or 7 in 32 bit plus a compatible graphics card), but it's still a feature.
Analogies compare two things that usually have no physical or logical relation to each other. When used well, you paint a mental picture of something that isn't easy to visualize and it may even act as a mnemonic device to aid memory.
Now for the dark side of analogies. The reason DX9 was chosen is pretty simple; compatibility with older hardware and windows xp seemed to be important with a target release of 2008-ish; DX10 and 11 offer some features that are marginally useful and some better performance, but nothing huge. When you explain something that's not complex by means of an analogy, you add lots of words and make your post convoluted and harder to read.
Poor analogies also risk spreading confusion; two decisions could look similar even if the logic behind them is not the same. If you extrapolate from an analogy beyond its narrow scope where it applies you might end up with word sallad. This is particularly dangerous when discussion complex subjects, because it's not always easy to notice when the analogy has run off the rails.
Software has some very strange properties that do not apply to anything else. The big cost in software is design, code maintenance/QA and runtime resources. Nobody in their right mind would have a car with 2 or 3 different versions of the engine, not because it's costly to design the engines, but because they weigh too much; but for software it's fairly cheap and routinely done. Each additional unit of software costs almost nothing to make; each additional car engine is very expensive to manufacture. There's just too many ways to go wrong when comparing software to the physical world.
That's also the reason analogies are the favourite device of charlatans. If you see creationists in a debate, it is often nothing but a long string of analogies. It's very easy to bamboozle, confuse and mislead with analogies; that's why they like them.
First of all, I have played this game in the beta. And I know how far it has come.
Secondly, your main point was "The game has a 30 tick rate, which is a pretty big bottleneck" and this is simply wrong. You can easily search the forum for "tick rate" and find post from the devs, that explain why increasing the tick rate above 30 would not lead to a smoother experience for the players.
I may be wrong, that I thought you are pointing at hit-reg with your bottleneck theory. But that isn't important. Because as I wrote already, an normal server (<= 20 players) can hold the tick rate even in the end game.
I also don't know how deep your knowledge on network- or game engine-programming is. But projectiles in BF3 have not the same demand that complex objects like buildings have. You also forgot that NS2 has more such objects than NS1 had. Starting at cysts, power nodes, AI units and so on. The NS2 classic mod has shown that removing the cyst mechanic alone brings a huge performance boost.
Anyway. I'm on your side, that performance needs to get better and that the LUA code is one culprit. The news about a new and optimized LUA Virtual Machine are a bit old now (came up around release). I don't know if they are working on it.
It's obvious why a 800 x 600 res low quality you tube video would not suffice for hard to see hit reg, so if you can only play the game like that you probably have large fps issues that are way more damaging to your game than the potential for slight hit reg issues.
However, If you can typically play in any reasonable resolution (graphic settings don't matter) above 30 fps, then you can record a demo which has zero performance impact. Then afterwards play the demo back while using some light weight video capture program.
Anecdotally, I have a four year old, entry level , first gen i7 and I can record with 1080 @ around 50 fps. But like I said this isn't the only method.
@ripsey: definitely multifaceted. But the scenario where you are aimed at an enemy model, fire, and do no damage (regardless of any circumstance) should not be occurring and is exactly what we can't seem to reproduce. That is precisely what hit reg refers to: being able to shoot at what you are pointing at..
Like I said there's suspicion.. but I personally believe it's due to things like the brain compensating for factors like lag compensation and input delay, feeling like something is slow/off and not as fluid as it should be. But actual hit reg issues *May* not be present anymore. (At least not common, after scrubbing through hours and hundreds of GB of hd footage frame by frame looking for what I swore up and down were hit reg issues while playing)
/derail
I bet dx10 would at least bring some performance improvements.. but like someone else said, if the bottleneck is CPU you might not even see it / so it may not be worth the dev time.
Your understanding of what the community cares about appears to be severely lacking, especially considering your title at UWE as community organizer/PR/whatever else. Tweadle only expressed what the majority of this thread has already said. Considering this forum has been filled with performance and FPS related threads since beta, you had to know this viewpoint would be expressed, right?
I'm interested to hear more about how this update will make descent better, otherwise it seems just like another setting I will disable.