ATI support poor or just slow?

[mugginz]

"one brand to rule them all."

kano: nvidia is the one brand rule them all because of Nvidia only PhysX and Nvidia only CUDA and Nvidia only 1 triangle per pixel tessellation.

Give him an example of a use case where ATI makes sense and he might surprise you.

you try to make jokes on me but i'm not the one "one brand to rule them all." Guy..

Firstly that post is no joke. I meant what it said.

I prefer OpenCL because of VIA-s3,Intel,amd and not Nvidia only.

I also prefer the thought of an open GPGPU standard over a propriety one. Sometimes people have to use what works best in a given situation with some immoveable constraints. Sometimes it's just that those situations don't provide for the kind of flexibility that would allow for the use of the most open choice.

So, in a perfect world, for me the open solution is the best solution.
Remember though, this isn't a perfect world.

[mugginz]

this is only true if you use low-resolution-textures gtx480 Texture Fillrate 42 (GT/s)VS 5870 68 (GT/s)

IF you use 2560x1600+high_resolution_Textures nvidia lose!

So this is the thing.

Many different parts of that PC you are using right now have various theoretical maximum throughputs. When you add all the numbers up you don't always get the same as the real world end results.

So sometimes it's prudent to supply actual real world repeatable benchmarks. These supersede any theoretical numbers in my view. The theoretical max may show what's achievable in an optimal for that hardware situation but that's not what we always get.

The benchmark results I've seen say that the 480 will beat the 5870 in Unigine with tessellation at 2560x1600. They also show the 470 will beat it too at that res.

Perhaps you can link me to the results you have that say otherwise.

[mugginz]

"So sometimes it's prudent to supply actual real world repeatable benchmarks."

unigine isn't a real-world benchmark unigine is just payed by nvidia!

You supply theoretical throughput as a basis for your argument and then have the indecency to tell me Unigine doesn't suffice!

Also aren't games going to be based on the Unigine engine? I certainly hope so because I've already paid for one purporting to use it. It's called Primal Carnage. Maybe you've heard of it?

ArmA2 for example is not payed by nvidia and o wow all settings on max+200% means longer-range-lod and 4xSSAA and nvidia 'lose' nvidia just lose on max Quality over 20-30%!!!!!

http://www.3dcenter.org/artikel/zota...rmed-assault-2

nvidia loses on real-world benchmarks!

So is the reason that a GTX-480 beats a 5870 at 2560x1600 while running Metro 2033 because nVidia pay the developers to hobble it?

Just to throw another gaming related question your way, how does Crossfire compare to SLI these days in relation to scaling?

[Kano]

That's normal that some games are optimized for ati and others for nvidia. The chips do not work the same way they just understand the same commands. And some ways are better for one vendor than the other, but of course that must be a conspiracy, right?

[mugginz]

if you wana play any real-world games on the max Quality means at minimum 4xSSAA(or down-sampling DSSAA)+16AF+High-resolution-texture-mods on ArmA2,Alien vs Predator 2010(dx11),Crysis,Gothic3 superhighres-texturmod,

thats because nvidia fails on SSAA/DSSAA and nvidia fails on high-res-texture-mods.

Well obviously the figures you're looking at as far as 2560x1600 4xAA goes are different to mine.

Firstly that post is no joke. I meant what it said.

then you just fail.

Are you kidding?

yes it isn't a perfect world just because of nvidia and people like kano!

Are you into conspiracy theories at all?

[mugginz]

DX11 tesselation benchmark 5870 win 55fps to 50fps to the gtx480:

http://www.geeks3d.com/20100528/test...70-vs-hd-5770/

You left out a link from that page.

http://www.geeks3d.com/20100407/gefo...on-comparison/

Hair DX 11 (Realistic Hair) - FPS
62 - GTX 480
24 - HD 5870

Island Demo
52 - GTX 480
14 - 5870

SubD11
468 - GTX 480
354 - HD 5870


Detail Tessellation
1394 - GTX 480
732 - HD 5870


PN Triangles
3043 - GTX 480
1064 - HD 5870

For this test, to quote directly from the article:

This sample, contributed by AMD, presents a technique for achieving smooth surfaces from the position and normal data of a low-density mesh. It utilizes Direct3D 11 APIs and hardware to make use of the new tessellation stages of the pipeline.

[mugginz]

LOL yes! you knowledge of unigine and tessellation isn't based on an real game just a benchmark!

no i do not heard of it.

but if the game uses the unigine+high-resolution-textures more than a gtx480 42 GT/s can handle than a 62 GT/s hd5870 win!

if not your game just has low-resolution-textures!

means if nvidia wins on 42GT/s its just a ugly game!

I surprised you've not heard of it. Let me link you then

http://www.primalcarnage.com/website/

metro 2033 is payed by nvidia because metro used physX and nvidia payed for it!

Isn't Alien vs Preditor an AMD sponsored title?

the 55fps AVP tesselation benchmark do have a crossfire part to!
2 pices of 5770 do have also 55fps and beats a gtx480!
for just 200?!

Bit of a wonky comparison you have there.

Try looking at 2 x GTX-480 in SLI vs 2 x 5870 in Crossfire.

[mugginz]

how cares about fake 2xAA == = ????

only 4xSSAA or 4xDSSAA matters!

If you say so but there are other forms of AA that work well.

not really ;-)

Well that's sad then.

at all? nvidia just doom the hole pc market!

I guess it's not too bad for a conspiracy theory on short notice.

[mugginz]

i have seen the video before because of chatting with a friend.
but yes my name memory is a black hole.


do you understand this part?

"but if the game uses the unigine+high-resolution-textures more than a gtx480 42 GT/s can handle than a 62 GT/s hd5870 win!
if not your game just has low-resolution-textures!
means if nvidia wins on 42GT/s its just a ugly game!"

yes? i think you don't understand this part!!

But the benchmarks I'm reading suggest that there's to and fro with regards to who is fastest at 2560x1600 SSAA and high res textures. I'm not seeing an ATI always wins this setgment theme at all from independant source but you are.

i don't know but arma2 isn't an amd sponsored title!

Maybe it should be because to get playable frame rates in those benchmarks you provided you have to disable SSAA on either nVidia or ATI cards. And that's at 1920x1200 not to mention 2560x1600.

? you can only compare per ? 2 pices of 100? 5770 beats a 450? gtx480!

200? vs 450? nvidia is just more than double high in the price!

and nvidia does only 50fps and the 200? cards do 55fps!

The two 5770's also beat your 5870 in that benchmark.
Using two weaker cards in place of a single better card caps your options a bit though. What happens when you want to run 2560x1600 SSAA?

2gtx 480 cost more than 2x5870

you can compare a 2x gtx480sli 900? to 4X hd 5770 400?

i think the 400? hd5770 win the benchmark!

or 4X 5830 for 800?

Firstly you'll need a more expensive motherboard to support the four 5770's or 5830's so that dampens the price advantage a bit.

And as SLI is reported to scale better than Crossfire does at the moment if you're talking multi-card I'd strongly recommend nVidia.

[mugginz]

http://www.pcgameshardware.de/aid,76...on-Spiel/News/

watch this

That demonstrates the type of graphics you're not going to get on a console for a while.

Almost tempted to buy Windows to give that game a try.