Official HD 2900 Discussion Thread

[Ubermann]

The card was dealyed cuz they wanted to release all cards at the same time.
They were spitting out 65nm cards and there was no problem what so ever with silicon or software, this is what AMD said.

So we waited all this time and they only launched this card anyway..people were telling that it would be worth the waiting cuz we would get something extra that everyone wanted, wtf was that ? HL2 ?

This sux so much! Or is ít AMD that sux ? Im switching side..Im now an official Intel and Nvidia Fanboy idiot.

[Shintai]

The card was dealyed cuz they wanted to release all cards at the same time.
They were spitting out 65nm cards and there was no problem what so ever with silicon or software, this is what AMD said.

So they waited all this time and only launched this card anyway..
This sux so much!

You forgot the 100million GPUs this year with uber performance according to theinq. :P

[mascaras]

I received this today :








tests tomorow



regards

[Think]

No one in there right mind would upgrade until there is a value to having DX10, let alone a quad scenario anyway - wait until September.

[gunit]

no wait until 2008, no, fall 2009 or better yet 2010... Everyone needs to stop playing the waiting game, THERE IS ALWAYS SOMETHING BETTER A FEW MONTHS FROM NOW IN THIS HOBBY OF OURS!!!

[Shintai]

no wait until 2008, no, fall 2009 or better yet 2010... Everyone needs to stop playing the waiting game, THERE IS ALWAYS SOMETHING BETTER A FEW MONTHS FROM NOW IN THIS HOBBY OF OURS!!!

Currently there were already something better half a year ago

I think we have to wait for 2950/2650 before its "worth" it. Just like 8600 aint really worth it either right now.

[Noobie]

Dont expect anything but around 1950pro performance +DX10. 8600GTS didn´t show wonders either. And it will have the same 128bit bus as the 8600GTS, while the x1950Pro got 256.

If the x2600 uses a similar ring-bus as the x2900, and has 4 memory chips rather than 8, and still uses one ring-stop for 2 memory chips, it would be more efficient than the x2900, thus making up for the large drop in bus-width.

I wouldn't hold your breath on it tough.

There is another difference between 8600-8800 and the 2600-2900 relation. The 8800 has 6 shader clusters of 16 shaders each (96 shaders). The 8600 has 2 shader clusters of 16 shaders each (32 shaders). However the 2900 has 4 shader clusters of 16 shaders each (64 shaders), but the 2600 has a total of 24 shaders total. Apparently it has 2 shader clusters of 12 shaders each, or possible even 4 shader cluster of 6 each. The reduction of shaders in the shader cluster has an impact on the performance, positive or negative I do not know. It could also be so small it's unnoticeable....

[madcho]

no wait until 2008, no, fall 2009 or better yet 2010... Everyone needs to stop playing the waiting game, THERE IS ALWAYS SOMETHING BETTER A FEW MONTHS FROM NOW IN THIS HOBBY OF OURS!!!

wait is not ridiculus.

lot of guy's here have already big cards. Why should we take a new card ?

buy a g80 or R600 why ? Graphic power is low, and dx10 is far.

If you do alldays over 6k@3dmark06 like me, I don't think this so usefull.

[krampak]

I got my own numbers...

QX6700 @ 3.4Ghz
Asus P5K Deluxe
2x1GB Gskill PC8000HZ

Forceware 160.03 for NVIDIA card
6.87.4.2 for ATI Card

Asus EAH2900XT

3DMark 2001: 55524
3DMark 2003: 37779
3DMark 2005: 19830
3DMark 2006: 12041

AA16X = AA8X Filter Wide-Tent = 16 Samples

FEAR
1024x768 AA4X, AF 16X Tot max: 92 FPS
1024x768 AA16X, AF16X Tot max: 49 FPS

Company of Heroes
1680x1050 All max, no AA: 91 FPS
1680x1050 All max, AA16X: 55 FPS

Supreme Commander
1680x1050 high quality, AA4X AF16X: 45 FPS
1680x1050 high quality, AA16X AF16X: 26 FPS

Lost Coast
1680x1050, AA8X AF16X HDR FULL: 76 FPS
1680x1050, AA16X AF16X HDR FULL: 32 FPS

Episody One
1680x1050, AA4X AF16X HDR FULL: 128 FPS
1680x1050, AA16X AF16X HDR FULL: 39 FPS

Oblivion Indoor
1680x1050, AA4X AF16X HDR FULL IQ MAX: 54 FPS
1680x1050, AA16X AF16X HDR FULL IQ MAX: 39 FPS

Far Cry 1.4beta (Regulator)
1680x1050, AA4X AF16X IQ MAX: 77 FPS
1680x1050, AA16X AF16X IQ MAX: 28 FPS

GRAW
1680x1050, AA4X AF16X IQ MAX: 64 FPS
1680x1050, AA16X AF16X IQ MAX: 64 FPS (I guess AA16X wasn't applied correctly)

STALKER
1680x1050, AA MAX (ingame) AF16X IQ MAX HDR FULL: 28 FPS

Asus 8800 Ultra

3DMark 2001: 58734
3DMark 2003: 43949
3DMark 2005: 19369
3DMark 2006: 13884

FEAR
1024x768 AA4X, AF 16X Tot max: 156 FPS
1024x768 AA16X, AF16X Tot max: 145 FPS

Company of Heroes
1680x1050 All max, no AA: 113 FPS
1680x1050 All max, AA16X: 86 FPS

Supreme Commander
1680x1050 high quality, AA4X AF16X: 71 FPS
1680x1050 high quality, AA16X AF16X: 62 FPS

Lost Coast
1680x1050, AA8X AF16X HDR FULL: 127 FPS
1680x1050, AA16X AF16X HDR FULL: 115 FPS

Episody One
1680x1050, AA4X AF16X HDR FULL: 178 FPS
1680x1050, AA16X AF16X HDR FULL: 128 FPS

Oblivion Indoor
1680x1050, AA4X AF16X HDR FULL IQ MAX: 113 FPS
1680x1050, AA16X AF16X HDR FULL IQ MAX: 102 FPS

Far Cry 1.4beta (Regulator)
1680x1050, AA4X AF16X IQ MAX: 130 FPS
1680x1050, AA16X AF16X IQ MAX: 112 FPS

GRAW
1680x1050, AA4X AF16X IQ MAX: 102 FPS
1680x1050, AA16X AF16X IQ MAX: 95 FPS

STALKER
1680x1050, AA MAX (ingame) AF16X IQ MAX HDR FULL: 54 FPS

Power Consumption

GTS 320: 244W idle 437W full
GTX: 226W idle 447W full
Ultra: 259W idle 484W full
HD2900XT: 225W idle 502W full

[bloodbanger]

Why you put 2900XT against a Ultra?We already know it can NEVER compete against a Ultra...

[mascaras]

@KrampaK the new Wide Tent filter its same image quality of traditional AA x8 and you got less FPS with Wide tent filter , Try run same games with traditional AA x8 and see if the FPS go UP .....

BTW: Windows Vista or XP ???


regards

[CrimInalA]

Krampak where do you get these numbers from ?
is that own experience ?

[v_rr]

AMD Cautioned Reviewers On DX10 Lost Planet Benchmark

Tomorrow Nvidia is expected to host new DirectX 10 content on nZone.com in the form of a “Lost Planet” benchmark. Before you begin testing, there are a few points I want to convey about “Lost Planet”. “Lost Planet” is an Nvidia-sponsored title, and one that Nvidia has had a chance to look at and optimize their drivers for. The developer has not made us aware of this new benchmark, and as such the ATI Radeon driver team has not had the opportunity explore how the benchmark uses our hardware and optimize in a similar fashion. Over the next little while AMD will be looking at this, but in the meantime, please note that whatever performance you see will not be reflective of what gamers will experience in the final build of the game.

http://www.vr-zone.com/

[XeRo]

The card was dealyed cuz they wanted to release all cards at the same time.

I thought delay was so they could launch the entire family at once. BTW, what happen to we'll launch 10 DX10 GPUs?

[Noobie]

I got my own numbers...

QX6700 @ 3.4Ghz
Asus P5K Deluxe
2x1GB Gskill PC8000HZ

Forceware 160.03 for NVIDIA card
6.87.4.2 for ATI Card

That power consumption is insane. Can you stop OC'ing and test the power usage then?

GRAW uses the same (type of) engine as STALKER. If you're using the "Full Dynamic Lighting" option (which I can only assume you are), then any AA settings you have in your control panel have no effect. The in-game settings only provide marginal quality improvement, so you should also run both games with AA off.

[largon]

There's more sofware bottlenecks to R600 than the drivers ATi writes. The shaders on R600 are so much different that current DirectX can't properly utilize them.

I reckon we'll see noticeable gains as soon as Microsoft updates DirectX and more importantly, optimises the HLSL (high-level shader language) compiler in it for the shaders found in R600.

[DilTech]

Guys... ATi's R600 lacks the hardware for AA Resolve.... Sadly this means AA is likely to always be slow on the part, and no driver will fix that because it's using shader power to do this job.

Then we come to the ROP hardware, designed for high performance AA with high precision surface formats, at high resolution, with an increase in the basic MSAA ability to 8x. It's here that we see the lustre start to peel away slightly in terms of IQ and performance, with no fast hardware resolve for tiles that aren't fully compressed, and a first line of custom filters that can have a propensity to blur more than not. Edge detect is honestly sweet, but the CFAA package feels like something tacked on recently to paper over the cracks, rather than something forward-looking (we'll end up at the point of fully-programmable MSAA one day in all GPUs) to pair with speedy hardware resolve and the usual base filters. AMD didn't move the game on in terms of absolute image quality when texture filtering, either. They're no longer leaders in the field of IQ any more, overtaken by NVIDIA's GeForce 8-series hardware.

http://www.beyond3d.com/content/reviews/16/16

First, they refuse to call a spade a spade: this part was absolutely delayed, and it works better to admit this rather than making excuses. Forcing MSAA resolve to run on the shader hardware is less than desirable and degrades both pixel throughput and shader horsepower as opposed to implementing dedicated resolve hardware in the render back ends. Not being able to follow through with high end hardware will hurt in more than just in lost margins. The thirst for wattage that the R600 displays is not what we'd like to see from an architecture that is supposed to be about efficiency. Finally, attempting to extract a high instruction level parallelism using a VLIW design when something much simpler could exploit the huge amount of thread level parallelism inherent in graphics was not the right move.

http://www.anandtech.com/video/showdoc.aspx?i=2988&p=31

Due to this, it's unlikely the R600 will ever catch up with AA involved.

[DeathReborn]

AMD Cautioned Reviewers On DX10 Lost Planet Benchmark


http://www.vr-zone.com/

Funny how AMD are happy to give out an outdated version of CoJ that has bugs specific to nVidia Hardware then call fowl on Lost Planet. Pot, Kettle & Black anyone?

[Periander6]

I thought delay was so they could launch the entire family at once. BTW, what happen to we'll launch 10 DX10 GPUs?

Did people seriously believe this excuse? A more transparent piece of BS is hard for me to imagine.

[Caparroz]

There's more sofware bottlenecks to R600 than the drivers ATi writes. The shaders on R600 are so much different that current DirectX can't properly utilize them.

I reckon we'll see noticeable gains as soon as Microsoft updates DirectX and more importantly, optimises the HLSL (high-level shader language) compiler in it for the shaders found in R600.

Finally someone with some sense that actually read the architecture bit of the reviews. The erratic performance of 2900xt across several diferent game benchmarks is to be expected. It's simply the nature of the R600 architecture design.

Did people seriously believe this excuse? A more transparent piece of BS is hard for me to imagine.

People will believe whatever they're willing to believe and it takes very hard work to change that. As a advertising professional I thank for that.

[XeRo]

Did people seriously believe this excuse? A more transparent piece of BS is hard for me to imagine.

What can I say? I'm just very gullible

[deathman20]

There's more sofware bottlenecks to R600 than the drivers ATi writes. The shaders on R600 are so much different that current DirectX can't properly utilize them.

I reckon we'll see noticeable gains as soon as Microsoft updates DirectX and more importantly, optimises the HLSL (high-level shader language) compiler in it for the shaders found in R600.

Totally forgot about the HLSL. Good catch, I know I read some things on it but it sounded like it was already implemented, guess more or less they where talking about when its implemented.

[Caparroz]

Totally forgot about the HLSL. Good catch, I know I read some things on it but it sounded like it was already implemented, guess more or less they where talking about when its implemented.

HLSL is nothing new. It was introduced when DX9c was released, IIRC. The only diference is that all shaders must be writen in HLSL for DX10.

[eXceeded]

hmmm so all that bandwidth is wasted since the card is poor at doing AA anyway?! No way...

[nn_step]

hardware processing has always been and will always be the fastest however, it is the absolutely most limited. hence we have Software controlling our Processors instead of the computer being hard coded.