Microsoft XNA and DirectX 10 thread.

Most games don't scale anywhere near linear in performance with dual-core, let alone quad-core, but much work should be done in that area over the next 2-3 years. To keep 'applications' (cough, games) scaling in performance with the hardware.

I mean once every video card can do 2560 x 1920 at high enough detail with FSAA in games while keeping the minimum frame rate above 80fps where are we going to go next ?

I suspect the answer is more optimized code in game engines (The guys that make the Unreal engine must sleep on piles of money), and this is likely to include physics processing. Hopefully just using the extra CPU cores and offloading to highly advanced (96+ unit, each capable of crunching arrays of mathematical data) math co-processors to get far more 'real' effects in games. (vs them just looking pretty but only having basic interaction, and basic material/object dynamics).

Emphasis will, and about bloody time I might add, change to the MINIMUM FRAME RATE in benchmarks, and the ceiling will be limited to say 120fps so a long frame here and there will hurt the average, and weighted average, frame rate. The minimum frame rate is highly dependant on the CPU being able to process all the logic, physics, etc... and this is going to be a very good thing within 5 years.

The buzzwords for 2007 - 2012 will be based around:

- Minimum frame rate in a given test (Who cares if you get 100fps average, if it drops to 15fps sometimes ?, I know I do, thus I want my CPU performance and want to see www.clearspeed.com win out, as have PCI-X 64 bit slots ready for their dual-co-processor cards - See sig Razz)

- There will no doubt be other buzzwords of importantance, but they'll all relate directly to the above point. Cool

- Did I mention "It is really about time they shifted focus to minimum / weighted frame rate from average ?"

- XNA is rarely mentioned, in fact this is the 1st time I've mentioned it on a public forum. Considering it goes hand in hand with DirectX 10 I am shocked. See: http://www.microsoft.com/xna


Will XNA be the next big thing ?, Only time will tell.
2 answers Last reply
More about microsoft directx thread
  1. Quote:
    Most games don't scale anywhere near linear in performance with dual-core, let alone quad-core, but much work should be done in that area over the next 2-3 years. To keep 'applications' (cough, games) scaling in performance with the hardware.

    I mean once every video card can do 2560 x 1920 at high enough detail with FSAA in games while keeping the minimum frame rate above 80fps where are we going to go next ?

    I suspect the answer is more optimized code in game engines (The guys that make the Unreal engine must sleep on piles of money), and this is likely to include physics processing. Hopefully just using the extra CPU cores and offloading to highly advanced (96+ unit, each capable of crunching arrays of mathematical data) math co-processors to get far more 'real' effects in games. (vs them just looking pretty but only having basic interaction, and basic material/object dynamics).

    Emphasis will, and about bloody time I might add, change to the MINIMUM FRAME RATE in benchmarks, and the ceiling will be limited to say 120fps so a long frame here and there will hurt the average, and weighted average, frame rate. The minimum frame rate is highly dependant on the CPU being able to process all the logic, physics, etc... and this is going to be a very good thing within 5 years.

    The buzzwords for 2007 - 2012 will be based around:

    - Minimum frame rate in a given test (Who cares if you get 100fps average, if it drops to 15fps sometimes ?, I know I do, thus I want my CPU performance and want to see www.clearspeed.com win out, as have PCI-X 64 bit slots ready for their dual-co-processor cards - See sig Razz)

    - There will no doubt be other buzzwords of importantance, but they'll all relate directly to the above point. Cool

    - Did I mention "It is really about time they shifted focus to minimum / weighted frame rate from average ?"

    - XNA is rarely mentioned, in fact this is the 1st time I've mentioned it on a public forum. Considering it goes hand in hand with DirectX 10 I am shocked. See: http://www.microsoft.com/xna


    Will XNA be the next big thing ?, Only time will tell.


    Well said friend, but by the time that these things are met and 120 is the minimum framerate, the general gaming community will be shooting for something bigger and better. The day 120 frames per second is the lowest you can get is the day that 120 fps is not acceptable anymore and we will move on to bigger and better things. We are like big diseases that consume everything then move on. I forgot what i was talking about.
  2. Since consumers dictate demand and if their minimum, average, maximum, and weighted frame rates in a given test are all the same number (eg: 80 fps) I don't see consumers wanting more.

    Sure detail levels and viewdistance (often overlooked) will skyrocket but eventually a point is reached where people don't see the need to spend more to gain 'nothing they can perceive'.

    At the end of the day there is no point researching and mass manufacturering something slightly better (be it hardware / GPU, or software / games) if the cost will be so significant that it will not be made up for with revenue / profits.

    Add to that, that the world population is stablising, however by 2015 the number of Internet users will be 3-4x what it is now (potential market is there, for now) but once it gets to 100% world population with fast Internet, and the population 'flatlines', which is expected to happen around the same time 'perceived technical gain for users' starts to flatine, what will happen ?

    In 10 years processing power will be between 32x (worst case) and 1024x (best case, hardware offloaded techs like GPUs, etc, higher end machines) times what it is now (depending on the task), but will average around 100x as powerful.

    Screen resolution only rises by about 16x the total number of pixels per screen every 10 years... however it is starting to flatline now. (2560 x 1920, Can more pixels be crammed into a smaller space ?, Will monitors get larger and use more resources to produce ?, etc).

    Can you imagine a 10,240 x 7,680 resolution screen that is only 36" wide ?, Is it even possible to cram that much detail into an 'affordable' screen ?

    Game Level of Detail rises more than 'affordable' monitor resolution and size, so we can expect to pass a point where 'more detail' in a given square of pixels adds nothing to the user experience. Even extreme viewdistance is only so good as after a certain point you need to represent objects in sub-pixel space, and 'blend' such objects together, bearing in mind you'd need a 48+ bit Z buffer to do half the stuff I am thinking about. :P

    Bearing in mind that at apx 280 pixels per inch (or 78,400 per square inch) sitting a typical distance (if not further back, as was talking 36" wide) most people will not see any difference.

    10,240 x 7,680 on a 36" inch wide screen will look almost identical at 'safe' sitting distances, even when compared to 20,480 x 15,360 on another 36" inch wide screen (4 times the pixels).

    Bear in mind display devices won't reach that resolution quickly, or easily, while CPUs, GPUs, PPUs, etc will scale to have 100x the performance over a given 10 year period (at the same price point, and similar manufacturing costs, including when R&D + overheads are factored in).

    There are already systems out there that can run FEAR at 1,000 fps (bearing in mind that is a spike at lower detail, the average is less than half that.... it will not be long, few years or so, until it becomes the 'average' at maximum detail).

    So you go back to the original point, they'll just scale detail in the game engines, sure, etc... but they are working with a (comparatively) more & more finite number of pixels. The display resolutions are not infinite, nothing in the digital world really is, we'd need to go back to analog technologies to approach 'psuedo-infinite' resolutions - frankly that appears unlikely (see above example on 36" inch screen).

    So are they going to average a 64x64 sample into just 1 pixel (and that one pixel will be 1/16th the size they are today) ?, Would it look much different if they only averaged 16x16 sample into just 1 pixel ?

    As an average, or weighted average, the final rendered scene is going to 'look identical'. Just like an high res bitmap (eg: 8192 x 6144) image looks when compared to a JPEG (same res) with only 1-3% lossy compression.

    So when (not if) this happens where are they going to branch into next ?

    Bear in mind that as more tasks that are offloaded from the CPU performance can scale in an 'above linear' manner (when compared to now anyway, depends on graph style used). Just look at the sort of scaling the first 5 years of GPUs brought us, to accomplish the same with 'software rendering' via the CPUs in the same time frame, for the same cost, would've been near impossible.

    There are only 5 senses, and thought / logic, to games, so it is viable that no more than 6 'tasks' can be offloaded from the CPU to dedicated hardware (providing a large leap in performance vs using software each time, not that CPUs will need it in 10 years though :P) and the CPU cores will just run small isolated loops that control said accelerated functions.

    Do we need 5.1, 7.1, 12.2, 100.20, etc speaker systems ?, How much surround sound do people need ?, Audio was the first real sensory task to be acclerated and it is already reaching a point where 'humans can't perceive any more detail'. At 192 KHz sampling per second, using 32 bit samples, with 12 channels, and 2 bass channels we are pretty much immersed as much as we can be..... audio wise at least.

    Graphics will be next to 'hit this ceiling', within 10 years at my estimates. We'll have 2D vectors textures that scale 'seamlessly', highly advanced shading and bump mapping, far more advanced than now, elasticity, friction & similar effects simulated between GPU (Graphics) and PPU (Physics), and be able to run around 'small worlds' at 'life like detail levels'.

    So where to next ?, Physics, AI, Logic, etc...

    Once we have 1024 player servers, with an additional 8192 AI running around, in First Person, at the above detail levels audio and visuals wise, on islands (or should I say worlds ?) that take 48 hours to drive across, with a good balance of anti-air and High speed 'drop-in' air units, even 'space' units, where do we go next ?

    Within our life spans it is highly likely we'll observe this, first hand, happening. The trends have been in motion for years already.

    Are we (humankind) going to spawn a new sensory organ in the next 25 years ? - Almost certainly not.

    Humankinds quest for, better and better, will lead us to a points where a given product is 97.3% 'perfect' (as in close enough for entertainment and simulation purposes within said consumers budgets) and there is no justified point going that extra 2.7%.

    So then what will we aim to improve / simulate ?

    We all pack up our computers and walk outside, to find we can't see the Sun, we can't breath the air, we can't drink the water, or eat plant or animal matter, or grow anything ?, (Possible, but unlikely within 25 years, as this would cause economies, and nations, to collapse).

    We've still got another X years (say 60 on average) of life left..... So then what will we aim to improve / simulate ?

    Maybe we'll have genetic engineering downpat, and be able to make a few custom sensory organs for 'beyond life' realism levels. But we run out of places to attach / install them..... So it repeats, what will be improved next ?

    Maybe space travel, where such 'life like' energy efficient simulations, entertainment, training for colonisation of 'distance worlds' would become very important, considering it would take 100's of years to get to them, eve the speed of light speed.

    So we break the speed of light (maybe ?, unlikely within my life time IMHO).... and we can teleport people, goods, etc lightyears apart in fractions of a second, and setup a huge network of colonies (eg: planets become like countries are now)..... The allure of existance then implants in our thoughts..... "Well, what next ?"

    Eventually a quantum limitation is reached, where a certain thing simply can not be made better without making it substantially worse in another aspect, and thus said 'thing' might be branched of into two 'optimized' things, when thought of combined capable of the original task even better, and when seperate able to perform more specialist tasks even better, combined with the cons of neither and the pros of both..... and so they get redesigned in 2+ other 'things' each, and so on.

    There are not enough resources in the vastness of space, time and the universe to continue doing this. We are 'similar to' a virus, eating all in our wake (but enjoying it while it lasts).

    So then what ? (no I am not stoned, or on drugs at this point :P), Do we just see a giant list of credits for the universe ?, a huge 'GAME OVER - Thank You for Playing' sign ?

    Thankfully over that time period we'll learn to recycle materials a little better (:P) on an atomic level perhaps (:?:), but where are we going ?

    What crazy adventures will humankind entangle themselves with in the distant future ?, Do we have 'an ultimate goal' ?, Why are we constantly reverse engineering and trying to improve on what the universe offers ?

    Is our whole existance doomed to be driven by marketing, profits & the like ?, Will people more to even further away 'off worlds' to 'get away from it all' ?, Where does the cycle end ?.

    So we find a wall at the edge of the universe, made of nothing, pure nothing the strongest non-material known to us..... so we find a way to punch through the edge of the universe.

    Dare I ask..... So then what ?

    :? (Maybe I should read over some 'cDc' - Cult of the Dead Cow).


    But you see what I mean surely, even if a 'little' carried away :P, What is next in the world of first person simulations ?
Ask a new question

Read More

Performance Games Video Games Product