Question about graphic card & about directx
 

Hi,

My actual card does support dx 9 only ( Nvidia gtx260), are there improvements in the image quality with newer cards which support dx 11 when plaing CoD ?

As well, some technical questions, what is recommended ammount of ram for the graphic card ? what is benefit from high quantity for that ram ?

And finally, is the graphic card able to make the game playable with a high enough number of fps in high res high graphics with a processor as limited as Phonom nowadays that much more powerful proc do exist, I mean is the proc limiting the fps or is it only a matter of graphic card power (after all the Phenom is quite capable anyway) ?

Thanks devs, thanks mates.

Hi,

first of all, the GTX260 does support DX10, but only if the PC is running MS Vista or Windows 7. No, a DX11 GC won't improve the quality in COD, because COD doesn't support DX11 (at the moment).
I've noticed a big improvement in fps when I did move from WinXp to Windows 7 (+50%).

On my system, the GC is the limiting factor, none of the other components is running at the 100% limit.

For the VRAM I think the same can be said as for money: the more, the better :D - I've heard a lot of good things about the GTX 560 Ti with 2GB

Nvidia gtx260 supports DX10.0 - which is crappy. What is far far better and the devs mean when they say dx10 will boost performance is DX10.1, which in essence is closer to DX11 than DX10/DX9 and insures the card can actually handle dx10 features but still DX11 is the way to go. DX11 builds on DX10.1 by further optimizations and adding a host of new possible effects and tools, which will be used more and more in the next year and to come. I doubt any computer only games released in 2012 will support DX9, BF3 coming later this year is multiplatform and only DX10+ aswell, EVE is cutting shader model 2.0 support (dx9.0) - and its an MMORPG...

Now as far as newer cards go, of course there will be a big performance boost, even if it is not dx11. gtx260 was originally a mid range card at release, currently we are about 1.5-2 gens ahead of that, so performance will probably be atleast 3-4x more on a 560 purely hardware wise. A 580 is probably 15x more powerful on a modern 10.1-11 game than your card considering the hardware tech (brute power), softwaree optimization and architectual change.

Ram is memory, its how much information your pc can hold for quick access at the same time. Think of it as notes you took in class that are right infront of you, thats ram, the more of it you have the faster and better you can answer questions. Since ram is limieted (as is the size of a page) if its not in ram the program must request data from the hardrive (called paging) which is mechanical and where as ram is solid state. So imagine harddrive paging like going to your textbook to look up a page and number, then adding that to your notes and erasing other notes to make room. Also remember there is Vram (for your card) and system ram, vram is only used for holding graphical assets while system ram is for the whole game and can feed your vram faster then a HD would.

What is your processor version? You want to aim at a 4 core CPU, preferably a i7 which are the best atm (the 40nm range not the 32nm sandy bridge). If your graphics card is good, it can be bottlenecked by the CPU, which routes the info and results in the GPU sitting and waiting for necessary rescources before it can perform a task (leading to stutters and other nasty things). Personaly I wouldnt touch the phenom with a barge pole, its very old tech. Get an i7 or wait for AMD Bulldozer.

You will notice a big performance jump when moving from 200 range to 400 or 500 range which is DX version based and hardware vs normally just hardware (outlined above).

So are you saying we should change the config.ini entry from d3d10_0 to d3d10_1?

I think there needs to be some clarification made here ...

There's no real difference between DX10 and DX10.1 in terms of feature sets, and there is certainly no performance gain using DX10.1 over DX10 at all.

DX10.1 was an update in standards not features, DX10.1 made graphic card vendors make some optional components of the DX10 API mandatory in their hardware.

The main standards to be mandatory were:
- 32-bit floating point filtering
- 4xAA sampling and two specific sample patterns.
- Programmable shader output sample masks and multi-sample AA depth read backs.

In terms of DX11, it is just DX10.1 with three additional features made to the API:

DX11 = DX10.1 + Tessellation + Multithreaded rendering + Compute Shaders

People need to remember that DX11 also includes COM objects from the DX7 API such as DXInput calls and the like. What that means is each new "version" of Direct X is not a complete rework of the API, it's merely added features being implemented.

http://msdn.microsoft.com/en-us/libr...=vs.85%29.aspx
http://en.wikipedia.org/wiki/Microso...#Direct3D_10.1

Every bench test I've seen, notably Tom's Hardware lists the i5 2500 as the best performer ATM in regard to gaming and Sandy Bridge is by all accounts the recommended path for top shelf performance. What do you base your recommendation on?

I was wondering the same thing ...

http://www.anandtech.com/bench/Product/142?vs=288

In the anandtech bench, the 2500K (~$250AUS) is comparable to the i7 980X (~$900AUS) in almost all aspects.

The new i7 990X might be a different story but that's currently going for ~$1300 :rolleyes:

You are wrong, I am so tired of correction people on this forum its driving me up the wall. Also for a start I dont see why you repeating my statements about DX11 helps your argument because thats exactly what I said. But despite that the DX11 features are significantly more powerful then anything in dx10/10.1, I have gone over this is in great detail a million times already so I will just link you to the posts.
http://forum.1cpublishing.eu/showthr...795#post233795

http://en.wikipedia.org/wiki/High_Level_Shader_Language
http://en.wikipedia.org/wiki/DirectX#DirectX_10
http://en.wikipedia.org/wiki/Direct3D#Direct3D_10.1
Dx10.1 allows shader model 4.1 and parrallel cube mapping (basically it improves shader/lighting performance greatly) and established standards for hardware so you dont have crappy cards that cant actually run any dx10 features used, aka a dx10.1 card will be able to use dx10 features decently where as a dx10 card most likely cannot. For example look at Shogun 2 - DX10 cards will not be getting AA while dx10.1 cards will be.

DX11 features added are a massive performance improvement when used properly. But out of interest, what did was the purpose of your post exactly? Do you disagree that a 400 range card will easily beat out a 200 range? Or are you just saying dx10=dx11? Because there is a monstrous differance interms of upper end capacity, but in any case DX11 is 100% the way to go.

The fact that my hardware post was used by Creative assembly and pinned at the top of their forums for the guide on pc hardware? The fact that I compete in EVGA rig competitions. Also I have about 5-6 computers running atm of various ages for work, so I am familiar with a range of setups.
Sandybridge is not worth the extra $, you are paying for an integrated gpu on a cpu thats not needed. Above that sandybridge is the same as gulftown except the architecture was shrunk and they slaped on that integrated gpu then marked up the price. If you get a 40nm i7/i5 you can overclock it to 4.25+ghz on air stable, and 4.5-4.75 with a good water cooling system. Also buying an i5 is retarded if you want to keep the pc for any amount of time, how long are those 4 threads going to last you when BF3 comes out later this year and uses 8? For now its ok but it has no lasting ability as more companies come out with heavily threaded engines.

In addition the fact that you picked the 15 2500 is lol worthy. You do realise for like $20 more you could of got the K which has the better integrated GPU, and has a unlocked multiplier for overclocking? If you bought it, dumb decision you should of read more.

@Codex - no they are not really comparable, the 980 wins out, and those tests were favourable to the i5, so keep your thinking cap on ok? If a program only uses 4 threads then of course there will not be a large performance differance, some of those programs were from 2007... Now on the otherhand I have 12 threads and 6 cores, which means an immediate 33% boost due to the cores and up to (but unlikely) 3x more performance with heavily threaded programs (graphics design etc).

As for my CPU I needed it for rendering for Autodesk/Maya cgi, I also intend this pc to last for a bit so will be eventually adding another 12g of ram, and 2 more 580s with a nice raptor raid setup or even better SSDs. So to be bottlenecked by the CPU would suck, which wont happen when I overclock the 980x to 4ghz+...