Official HD 2900 Discussion Thread

[sladesurfer]

I got 20536 marks X2900XT default & E6600 @ 3600mhz

I got 17633 marks BFG 8800GTS OC (550/1600) & E6600 @ 3600mhz





--3DMARK 2005 -- X2900XT 512mb DEFAULT (cat 8.37)-- E6600 @ 3600Mhz ( 20536 Marks)




3Dmark2005 @ BFG 8800GTS OC (158.19) @ E6600 3600mhz 17633 Marks

Thats 3dmark05 my post was 3dmark06

[dnldnt]

on 3dmark you need to add -nosysteminfo to the shortcut and rivatuner and atitool do not work as of yet... use AMD GPU tool that was recently released. (http://www.techpowerup.com/downloads...Tool_v0.7.html)

Thanks for your quick reply, but " on 3dmark you need to add -nosysteminfo to the shortcut ..." can u be more specific? Sry for this and thank u!

[momomo6789]

if anyone wants to know fix for all 3dm is do delete direcpll.dll from system32 file and every thing sould work

[eXceeded]

dnldnt if you right click the 3dmark 06 shortcut and click properties and at the end of target (which is "C:\Program Files (x86)\Futuremark\3DMark06\3DMark06.exe" on my vista x64) add a space followed by -nosysteminfo ("C:\Program Files (x86)\Futuremark\3DMark06\3DMark06.exe" -nosysteminfo)

[dnldnt]

Hy, thanks for all your help guys, it's working now, standard clocks, first test 3DMark 06 - 10239 , and going up soon, many thanks and good luck to all, Danni. BTW, i have read on a review, they have used RivaTuner for o/cing, is not out yet that version?

[eXceeded]

try and find a beta version? Look on nzone, guru3d or ngohq...

[ahmad]

You understood wrong...

Problem with Shader AA is that you're taking shader power to do something that could be done by dedicated hardware.

So, now ATi(I refuse to say AMD had anything to do with this part) is going to attempt to offload physics onto the shaders AND AA onto the shaders? That's just not going to work.

Also, ATi don't have to force their driver, it already does it. That's why performance is so low with AA.

I don't think we can say that ATI cards do not have the capability of doing AA without passing it off to the shaders. Thats what the current drivers are showing now and it maybe that ATI is attempting to cover new grounds by going that path. That does not eliminate the possibility of it being fully capable of not depending on the shaders for AA... let us keep that in mind.

[xlink]

how are the vista drivers? about the same as the XP drivers?

[gunit]

Mmm, now that you mention CPU limiting. It's almost like developers are spending more time programming in a threaded model than optimizing. The later is still far more advantageous performance wise.


From recent news I heard the problem is exaggerated and underrated. The AF problem is exaggerated; the difference between G80/G84 and 2900 is supposedly very hard to find in games. The AA problem is far worse when in motion, BUT! it only occurs when wide & narrow tent are used, there is no problem when your using pure MSAA (2x, 4x, 8x).

exactly, everyone is making such a hubbub about bluring (which in motion looks kinda good), with those new AA modes, but regular old 4Xaa still kicks ass and everything is sharp.

[Sumanji]

Haha, HD2900... so much drama. I bet none of us thought it would be such a roller-coaster ride!

[poullos]

http://www.hkepc.com/bbs/hwdb.php?ti...HIS&rid=790821

New driver did the trick?!

To collect higher-order market, Radeon 2900XT HD cheaper prices than expected, Official pricing for $ 399, and competitors for a meeting of the GeForce 8800GTS.

According to this test, Radeon 2900XT HD performance is better than GeForce 8800GTS. About 5~30% the leading edge of the test results exceeded even more similar to the level of 8800GTX.

But AMD also frank and R600 graphics core driver has yet to be continued optimization, So in the future effectiveness still further improve the performance space.

[Ubermann]

Won in pretty much every test, but isnt that the one with bad quality ?

[gunit]

http://www.hkepc.com/bbs/hwdb.php?ti...HIS&rid=790821

New driver did the trick?!

wow, they are using 8.38. wish i could get my hands on those. those numbers are close to other reviews, i mean the 2900 was meant to compete with the GTS, and it does so very nicely.

[Noobie]

Does anyone know if the HD2900XT works on the ASRock 775-DUAL-VSTA mobo (which is limited to PCI-e 4x)? The 8800 did not work at first, because NVIDIA screwed up the drivers. This was fixed later on.

[jimmyz]

the second card on a 965 chipset works @x4 so my guess would be yes, but i don't know for sure.

[gunit]

here is what I would like to know. IF in fact there is something broken when it comes to AA, which is just speculation at this point, why is the AA hit not consistent in all games? Look at the link below. In X3 the 2900 takes a pretty big hit with AA. But in splinter cell, its very small and consistent with the other cards.


http://www.dinoxpc.com/Tests/articol...sp?id=644&p=10

[Noobie]

here is what I would like to know. IF in fact there is something broken when it comes to AA, which is just speculation at this point, why is the AA hit not consistent in all games? Look at the link below. In X3 the 2900 takes a pretty big hit with AA. But in splinter cell, its very small and consistent with the other cards.
http://www.dinoxpc.com/Tests/articol...sp?id=644&p=10

The 2900 performs AA resolve on the shaders. The back-end sends the data back up to the shaders over the ring-bus.

Not every game uses many shaders. This is best seen by looking at the 1900 and the 1800. The 1900 has many more pixel shaders. In some games the 1900 beats the 1800, such as FEAR and BattleField 2149 (or whatever year). In other games they are equal.

If a game does not use a lot of shaders, there are many shaders left that can be used. X3 isn't that shader intensive, but Splinter Cell Chaos Theory is (HDR + SMR 3.0).

X3
1280: 113.7 -> 78.4 -> 0.7
1600: 102.1 -> 63.9 -> 0.6
2048: 81.2 -> 45.5 -> 0.55

In X3 we clearly see that when the res goes up, the shaders & ring-bus have more work to do, and the performance hit from enabling AA increases.

Splinter Cell
1280: 165 -> 140 -> 0.85
1600: 125 -> 108 ->0.86
2048: 83 -> 75 -> 0.9

In Splinter cell we actually see that when the res goes up, the performance hit from enabling AA decreases. Why is this? It could be that at 1280 the CPU is bottlenecked (let's assume not).

Another theory is that the at higher resolutions the shaders can not perform their work in a single pass, thus a second pass is needed. With the second pass a whole lot of shaders are left doing nothing.

This would also explain why X3 takes a small hit from increasing resolution (there are enough shaders to take care of the resolution increase), whereas splinter cell takes a huge dive when increasing resolution (not enough shaders).

At this point we have to marvel at the ingenious ATI design. You can either increase the resolution without taking a huge performance hit (X3), or you can increase AA without taking a huge performance hit. The downside of this is that games tend to be indoor/outdoor a lot, which means different situations require different strategies. Maybe in time we will see auto-adjusting cards.

That is the first thing that comes into my mind, so I may be completely insane and very wrong.

[v_rr]

http://www.hkepc.com/bbs/hwdb.php?ti...HIS&rid=790821

New driver did the trick?!

New driver do the magic.
Careful GTX, because things are getting very closer with simple updates in a pre-release drivers

[gunit]

oh i see. can u explain why the XT performs so well in doom 3 and unreal 3 engine based games, beating the GTX even, but its very mixed bag in other engines. here is quake 4 for example, which I am happy to see since my next game purchase will be ETQW.

http://www.bit-tech.net/hardware/200..._hd_2900_xt/20

[ahmad]

here is what I would like to know. IF in fact there is something broken when it comes to AA, which is just speculation at this point, why is the AA hit not consistent in all games? Look at the link below. In X3 the 2900 takes a pretty big hit with AA. But in splinter cell, its very small and consistent with the other cards.


http://www.dinoxpc.com/Tests/articol...sp?id=644&p=10

Here is the thing: doing AA in the shaders will take a large performance hit if there are no engine specific optimizations in drivers. Meaning, for every game engine, ATI has to specifically add some code to make AA work at appropriate speeds for each game.

It does well in the doom engine because that is the first thing ATI started optimizing for, so I would expect it to. Having the freedom of doing AA in the shaders has its disadvantages, but it can also be a good thing if they fine tune performance.

Naturally, I wouldn't mind specific optimizations for each title as long as ATI can keep up with the games being released

[Eastcoasthandle]

http://www.hkepc.com/bbs/hwdb.php?ti...HIS&rid=790821

New driver did the trick?!

As the drivers mature and the games are better optimized to use the arch. of the HD 2900XT you will see better performance, you can count on it . I say wait until 8.39/8.40 is matured to see what the HD 2900XT can really do.

[DeathReborn]

oh i see. can u explain why the XT performs so well in doom 3 and unreal 3 engine based games, beating the GTX even, but its very mixed bag in other engines. here is quake 4 for example, which I am happy to see since my next game purchase will be ETQW.

http://www.bit-tech.net/hardware/200..._hd_2900_xt/20

I'd like to see how the megatexture they're using in ETQW is going to perform with the R600. According to the fud monkey R650 is going to fix the flaws in R600 including the poor texturing performance.

[ahmad]

Well yes. ATI keeps going with 16 TMUs, but I think its time to step out of the dark ages and see the light.

[mascaras]

http://www.hkepc.com/bbs/hwdb.php?ti...HIS&rid=790821

New driver did the trick?!

nice News , all tests with WINDOWS VISTA and in company of heroes even beat the 8800GTX , im waiting for those drivers


regards

[GLENBOY]

hiya mascaras yes if you find the 8.38 drivers before they are officially out please post on here link