Radeon HD 2900 (R600) reviews thread [Archive]

Geeforcer

10-May-2007, 17:10

Beyond 3D R600 Architecture and GPU Analysis (http://www.beyond3d.com/content/reviews/16/)

It-review.com (P)review? (http://it-review.net/index.php?option=com_content&task=view&id=1314&Itemid=1)
VR-Zone (http://www.vr-zone.com/?i=4946&s=1) / Faster-loading Cached Copy (http://homepage.ntlworld.com/gerald.marley/index.htm) courtesy of Gerald at Rage3D
TweakTown (http://www.tweaktown.com/articles/1100/9)
OCWorkwench (http://www.ocworkbench.com/2007/asus/EAH2900XT/g1.htm)
T-break (http://www.tbreak.com/reviews/article.php?cat=grfx&id=511&pagenumber=1)
The Inq (http://www.theinq.net/default.aspx?article=39580)
HardOCP (http://enthusiast.hardocp.com/article.html?art=MTM0MSwxLCxoZW50aHVzaWFzdA==)
Guru3D (http://www.guru3d.com/article/Videocards/431/)
TechPowerUp (http://www.techpowerup.com/reviews/ATI/HD_2900_XT)
HardwareSecrets (http://www.hardwaresecrets.com/article/449)
Hexus (http://www.hexus.net/content/item.php?item=8687&page=1)
Driverheaven (http://www.driverheaven.net/reviews/r600reviewz/index.php)
Hothardware (http://www.hothardware.com/articles/ATI_Radeon_HD_2900_XT__R600_Has_Arrived/?page=1)
TechReport (http://www.techreport.com/onearticle.x/12458)
Presence-PC (http://translate.google.com/translate?u=http%3A%2F%2Fwww.presence-pc.com%2Ftests%2FAMD-Radeon-2900-XT-22674%2F23%2F&langpair=fr%7Cen&hl=en&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools) (translated from French)
Bjorn3D (http://www.bjorn3d.com/read.php?cID=1099)
ExtremeTech (http://www.extremetech.com/article2/0,1697,2128942,00.asp)
FiringSquad (http://www.firingsquad.com/hardware/amd_ati_radeon_hd_2900_xt_performance_preview/)
iXBT (http://www.digit-life.com/articles2/video/r600-part3.html)
TomsHardware (http://www.tomshardware.com/2007/05/14/r600_finally_dx10_hardware_from_ati/)
AnandTech (http://www.anandtech.com/video/showdoc.aspx?i=2988)
3DCenter (http://www.3dcenter.de/artikel/radeon_hd_2900_xt/)
NextGPU (http://www.nextgpu.com/web/index.php?option=com_content&task=view&id=119&Itemid=31) (Spanish)
Hardware.fr (http://www.hardware.fr/articles/671-4/ati-radeon-hd-2900-xt.html) (French)
Hardware.info: Dutch (http://www.hardware.info/nl-NL/articles/amdnZGpqZGia/ATI_Radeon_HD_2900_XT_test/1)/English (http://www.hardware.info/en-UK/articles/amdnZGpsZGGa/ATI_Radeon_HD_2900_XT_test/)
XbitLabs (http://www.xbitlabs.com/articles/video/display/r600-architecture.html) Archrev / Performance review (http://www.xbitlabs.com/articles/video/display/radeon-hd-2900-games_21.html)
Bit-Tech (http://www.bit-tech.net/hardware/2007/05/16/r600_ati_radeon_hd_2900_xt/1)

Thanks to everyone for the links!

IbaneZ

10-May-2007, 17:19

HD should be in front of 2900. :oops:

SugarCoat

10-May-2007, 19:50

at the risk of stating the obvious, whats the point of this thread if you dont want anything until the launch? makes the purpose of this thread even existing days before launch seem a little egocentric.

fallguy

10-May-2007, 20:19

Admin edit: Please, TPU's article wasn't supposed to be leaked, so please don't link to the copied assets and wait until NDA expires to read it properly. There's more to W1zzard's work than the numbers.

Gracias, and sorry fallguy!

fallguy

10-May-2007, 20:26

No problems. The numbers didnt add up to me anyways. X1900XT beating the HD 2900XT more times than not. Perhaps poor drivers. Guess we cant talk about it either, so Ill hush. :)

BRiT

11-May-2007, 04:42

Can we only post legit reviews and not shody (p)reviews that are still in doubt over their authenticity?

Sobek

11-May-2007, 10:07

Can we only post legit reviews and not shody (p)reviews that are still in doubt over their authenticity?

Roughly translated;

This thread is presently useless.

And I couldn't agree more... We have the Rumors and Speculation thread, which is where all the fogey and bogus 'p/reviews' are, there's nothing official yet, and we probably wont see anything until a short while after launch (a day is common). Seems very premature, we're going to see a few pages of tripe before anything even remotely interesting comes along. Either a mod sticks around to keep the thread clear of anything review-unrelated, or at least a temporary lock would be nice... (for the sake of a clean, focused thread when the time comes). :razz:

Maintank

14-May-2007, 00:39

BRiT

14-May-2007, 00:45

Perhaps they knew exactly the same sort of things as when they made their ill fated decision to create the FX 4x4 platform to go head-to-head with Core 2 Duo and Core 2 Quad? It is platform no one uses that's over 6 months late, appears blah, hot, eats power, and delivers sub-par performance? It seems like they're following suit with Barcelona too. :cry:

Skrying

14-May-2007, 00:46

Performance appears blah, hot, eats power, and IQ is on par with the 8800.
6 months late and this is all they can come up with? Makes me wonder how much AMD knew about these issues before they made their ill fated purchase for a chipset nobody uses and graphics cards that are late.

Chipset no one uses? I suggest you take a look at the chipset market share again. What enthusiasts use is not very reflective of what everyone else uses.

I also not sure why people state the "6 months late and this is all they can come up with" line. Its six months late for a reason, they had issues. That doesn't make up for the fact that the chip would have probably had the same performance then. But you can't make magic in six months, especially since that was likely a delay and you have to fix those problems before adding "teh magics" to it.

I'm highly disappointed mainly from the stand point that the product is not as good as Nvidia's. Its that simple really. G80 is more efficient, equal image quality, and what appears to be similar performance. I really wish there was an in depth view on why the last two ATi chips have been so much less efficient. R520, you could maybe say the image quality (but so much?), but this time around there appears to be little reason. Nvidia is predicting the markets needs much better it appears and has a much more clear image on its future.

trinibwoy

14-May-2007, 01:00

Skrying

14-May-2007, 01:10

t-break takes a quick look - http://www.tbreak.com/reviews/article.php?cat=grfx&id=511&pagenumber=1

It seems AMD's pricing is doing the trick - all reviews so far seem to be focusing on the GTS/XT match up where the XT does well.

Hmm, that certainly put the HD 2900XT in a rather positive light. Which is interesting, for the most part it did decently better than the GTS and if price is there then I'd certainly pick it over the GTS. Curious about those Serious Sam 2 scores as the 2900XT did worse than the X1950XT in that case.

I think we might see a Nvidia move here, you know drivers that come out later that magically improve performance. Hard to say. Waiting out further...

INKster

14-May-2007, 01:39

weeds

14-May-2007, 02:04

Hmm, that certainly put the HD 2900XT in a rather positive light. Which is interesting, for the most part it did decently better than the GTS and if price is there then I'd certainly pick it over the GTS. Curious about those Serious Sam 2 scores as the 2900XT did worse than the X1950XT in that case.

I think we might see a Nvidia move here, you know drivers that come out later that magically improve performance. Hard to say. Waiting out further...

No mention of drivers used for ATI or NV, But if you go back to the launch review of the 8 series, Nov 8th the results are exact for NV.
http://www.tbreak.com/reviews/article.php?cat=grfx&id=474&pagenumber=5
So looks like a copy and past job from the launch review, Forceware 96.xx..

MulciberXP

14-May-2007, 02:11

reviewer bias always shines in the conclusions. unashamedly comparing the price of the 8800ultra to the 2900xt.

SugarCoat

14-May-2007, 02:12

Not sure how old this is but ZZF seems to be leading the large etailer X2900 launch.

HIS/Diamond/Sapphire all instock @ $399-$425

http://www.zipzoomfly.com/jsp/ProductList.jsp?ThirdCategoryCode=111704&SortBy=D

I have an itchy tringer finger and just want to try one but i cant afford, nor does it make sense, having both that and a GTX :cry:. Been looking forward to the R600 for over a year though so it feels very underwhelming.

Plus kinda worried about an R580 scenerio where they'll launch a refresh in less than 3 months.

Skrying

14-May-2007, 02:23

No mention of drivers used for ATI or NV, But if you go back to the launch review of the 8 series, Nov 8th the results are exact for NV.
http://www.tbreak.com/reviews/article.php?cat=grfx&id=474&pagenumber=5
So looks like a copy and past job from the launch review, Forceware 96.xx..

Interesting, didn't notice. Welp, another crap review...

Geeforcer

14-May-2007, 02:25

reviewer bias always shines in the conclusions. unashamedly comparing the price of the 8800ultra to the 2900xt.

You know, I wonder. That's the second review I've seen that uses 7-month old launch drivers for 8800. This is borderline bizarre. Another thing that struck me in some reviews is some rather strange game-to-game variations with no rime or reason. One game is tested at one resolution, but not the other. One game is tested with 4xAA/16xAF while the other uses 8xAA/8xAF. And, what in my recollection is a first, card prices are included in every test result alongside the name. Trying a little too hard to make a point, eh? I hope those "ideas" didn't originate in a review guide.

trinibwoy

14-May-2007, 04:28

Nothing sinister there I'd imagine. I've seen that kind of arbitrary benchmarking from those sites before. Move on folks, no conspiracy here :)

trinibwoy

14-May-2007, 05:30

Guru3D and HardOCP chime in. Sorry, too tired and lazy to find links :)

Geeforcer

14-May-2007, 05:34

Hard OCP review is up as well (liked from the OP)

Damn you trini!

Skrying

14-May-2007, 05:58

SugarCoat

14-May-2007, 06:22

This does not sound good at all. All that time they had to engineer the card..

There's no other way of saying this ... the card is very noisy. With the fan in idle it's really okay. But every now and then the RPM of the fan will quickly spin faster, and that makes a lot of noise. Even in desktop mode this happens. With Vista; Aero will definitely stress the graphics card some more, so I predict you could get irritated by this. AMD has acknowledged this and stated that they will fix it soon.

In Idle you can expect a dBa level of roughly 48. This already is moderate. But as mentioned, the card's fan will spin up even when you are doing nothing, and then we hit a 53 dBa volume level coming from the PC. This is very audible.

We did some stress testing though, and the RPM of the fan will remain at roughly 30-40%. With the help of Rivatuner programmer Unwinder we got a little deeper into this matter. If the GPU core hits 100 Degrees C a safety feature will kick in and the fan will rotate 100%, and man I tell you; its like a helicopter landing in your room. We did a small test. We forced the fan to 30% (desktop mode) at 48 DBa, then ran 3DMark 06. Believe it or not but after a minute the core had reached 100 Degrees C and the 100% fan RPM kicked in. So in theory in warm circumstances (a really hot summer), this might happen to you.

nVidia learned with their blower vac, and have been doing excellent cooling solutions on their high end cards for the last few years, ATi on the other hand once lead in this area but since the X850 havent been doing much improvement. To me this sounds like they slapped on the same unit the R520 had, and when the chip is that much bigger, its gonna cause issues.

BRiT

14-May-2007, 06:23

Baring performance improvements over the next two weeks, it seems that the upgrade to my main system will be a complete switch from AMD + ATI to Intel + Nvidia. :runaway:

RobertR1

14-May-2007, 06:23

I'm still with my arms up in the air about this card. Sometimes its just simply golden, performance beating out the 8800GTX. Other times its crap, performance barely above the X1950XTX. Really, I would love to know WHY that is and if there is any chance it can be fixed through drivers. Otherwise this will be a major hurt for ATi/AMD. You can't have your card performing everywhere across the board, people like to play more than just one game generally...

So much for drivers being ready, eh!?

By the time Ati sort out the driver situation and show some stability across the board, you'll see the refresh come out. Coupled with lower usable IQ than the G80, it'd be almost silly to pick up a x2900xt. It's a damn shame too. I was really looking forward to dumping my x1900xtx and giving stalker a try.

Dave Baumann

14-May-2007, 06:30

To me this sounds like they slapped on the same unit the R520 had, and when the chip is that much bigger, its gonna cause issues.
Ummm, have you seen X1800 and HD 2900's fans?

SugarCoat

14-May-2007, 06:37

Ummm, have you seen X1800 and HD 2900's fans?

i actually ment CFM ratings of the two HSFs not the physical features of the design. Obviously its not moving enough air at a reasonable noise level or the GPU itself isnt shutting off things it doesnt need which again rests on the drivers. But theres no excuse for the card to get near 100C so the HSF has to kick on afterburners.

edit-or its having bad leakage issues.

poopypoo

14-May-2007, 06:41

I'm not so worried by the card -- it seems like a good competitor (albeit horrfically late) for the money. I am not even a $300-card gamer, but a $150 or so -card gamer, so as long as the cards are competitive that means good news for me (price drops! :D). However, the strategy here worries me. nV is pushing the envelope right now, which doesn't benefit me (nor most gamers) directly -- but it does benefit the market in general, expanding it, improving it, giving us more options -- and most importantly for nV, grabbing lots of attention. I think that AMD will get clobbered in "mindshare" this generation if they don't have a massive advertising campaign in mind. They're going to be fighting a sentiment that (correct or not) says "that new radeon looks good... but is it the best?" Then that customer will take their three brain cells down to best buy and accidentally buy a turbocache part from last generation or something. ;) But the point is, it'll be a lost sale. Equally worrisome is the fear that ATi may not have the technology (or front office balls?) to compete at the high-end next generation, but I think it's too early to assume that now.

As a (sane and fairly educated) consumer, I think this card presses a lot of good buttons for me. It fights high-end inflation, it performs admirably, and it exceeds the GTS's performance in almost all higher IQ situations. I just hope it doesn't bomb... :???: Better GPGPU function is also a greater marketing advantage than I would have guessed a year ago. High power draw and mixed press are large negatives.

Geeforcer

14-May-2007, 06:43

I am curious by what PART of the card they are talking about. There are hot spots in power regulator areas of almost every single PCB. The card would shut down before letting the GPU or memory come anywhere close to 100C.

So, "A part of a card gets to 100C" != "GPU gets to 100C"

SugarCoat

14-May-2007, 06:44

bdmosky

14-May-2007, 07:10

The Guru3D review annoyingly calls the new AA modes CSAA instead of CFAA repeatedly throughout the entire article.

ChrisRay

14-May-2007, 07:13

The Guru3D review annoyingly calls the new AA modes CSAA instead of CFAA repeatedly throughout the entire article.

I saw that too. Gave me a big what the heck?

Geeforcer

14-May-2007, 07:19

Still waiting on the reviews by most majors.

Twinkie

14-May-2007, 07:31

The sampling of images (http://www.hothardware.com/articles/ATI_Radeon_HD_2900_XT__R600_Has_Arrived/?page=12#) above represent almost all of the standard anti-aliasing modes available with the new Radeon HD 2900 XT. We say "almost" because these images don't take into account the temporal anti-aliasing modes available or the adaptive modes that also help smooth the jaggies visible in partially transparent textures.

We present these images to you to illustrate the effects of the different modes on a nearly identical frame. If you flip through the images, you'll see that ATI's new modes definitely do a better job as the number of overall samples is increased. Focus your attention on the antennas on top of the building at the upper left of the scene and you'll see what we mean.
But if you also focus your attention on the cobblestones in the street, particularly in the lower left portion of the scene, you'll also see that ATI's new custom filtered AA modes also blur the textures a bit, even if 16x anisotropic filtering is enabled as we have here. Image quality (http://www.hothardware.com/articles/ATI_Radeon_HD_2900_XT__R600_Has_Arrived/?page=12#) Nazis or anisotropic filtering aficionados may be put off by the slightly blurred textures in some of these images, but for the most part the affect is minimal.

Hothardware also does some performance hit on using the new CFAA modes. I wonder how this will be accepted by the general public. You get less jaggies in return for some textures being blurred out a small bit.

Pete

14-May-2007, 07:43

t-break takes a quick look - http://www.tbreak.com/reviews/article.php?cat=grfx&id=511&pagenumber=1

I like that they set the reviewed card as the baseline and offer the % difference for other cards. It makes it very clear as to how AA+AF affects it with certain games. It takes a dive with CoD2 and FEAR, but it closes the gap with Q4 and SS2.

We may not be able to draw conclusions given the apparently old G80 drivers, but it's still a better presentation than most.

Thumbs up to Brent's HOCP review. It covered all the bases thoroughly, including IQ. My eagle eye did spy R600 filtering a tiny bit further out than G80 in their AF sshots (http://enthusiast.hardocp.com/image.html?image=MTE3ODk0MTA0NU1IN2xkMnFWY1dfNl8xX 2wucG5n), but it's really minor and may not apply to all games (or even angles).

Very curious how the performance gap between R600 and G80 differs from one review to the next, but I can't be arsed (that's so much better than "bothered") to ferret out the demo and system details. The Call of Juarez DX10 demo stood out for me, though; IIRC, the Chilean site showed R600 doing much better than (IIRC) Tweaktown.

dizietsma

14-May-2007, 07:46

I can't read deeply B3D's technical piece quite yet as I am working but from skimming through the pages it looks very good indeed .. nice work Rys. Does the bacon bap mean he is no longer a veggie ? Made my mouth water here at work as I have not had breakfast yet.

I also skimmed through some game reviews on other sites. Forgetting the GTX for the moment and whether AMD are in pain not being able to charge more than $399 for it, for that price it more than does itself justice I think.

Other noticable thing is the the 8800GTS 320MB is not too shabby at all at resolutions that the becoming popular 22inch LCD monitor will kick out ( 1600 x 1050 ) and so it makes me wonder if AMD can do likewise with the HD 2900 XT and bring out a smaller memory version to compete with that ? My assumption is 256MB, but is that just slightly too small ?

Twinkie

14-May-2007, 07:52

Although other people may say otherwise, i really liked hardocp's review. Brent and Kyle did a superb job. They did their usual best gameplay bench including apple to apple benchmarks, loads of IQ comparisons and wasn't afraid of voicing out their opinions based on several aspects to R600, including CFAA.

edit - my post count took a nose dive :D

dizietsma

14-May-2007, 08:05

silent_guy

14-May-2007, 08:14

im pretty sure the thermal threashold for a shut down is ~120C. Its certainly not 100.
FWIW: silicion processes are characterized for a transistor junction temperature of 125C. Obviously, that's inside the chip itself, so the maximum allowed external temperature will be lower, depending on the quality of the cooler.

tEd

14-May-2007, 08:20

Just had a quick look at the hothardware review.

http://www.hothardware.com/articles/ATI_Radeon_HD_2900_XT__R600_Has_Arrived/?page=12

You could spend forever trying to work out which AA mode to use now, especially if you have to balance it with a slight degredation of AF.

24x Edge detect certainly looks nice but cost's too much performance. They would rather implemented a real 12xMSAA mode. Also while the tent filter do a good job on edges an thin lines they also eat texture detail and are also costly in higher modes.

_xxx_

14-May-2007, 08:40

Although other people may say otherwise, i really liked hardocp's review. Brent and Kyle did a superb job.

That, and they're the only ones who had balls to call R600 a flop (which it is, twist or turn it as you wish). Till today, I hoped for some sort of surprise from ATI, but duh. I hope the midrange parts won't suck, 'cause that's what I'll be buying this gen.

Geeforcer

14-May-2007, 08:45

That, and they're the only ones who had balls to call R600 a flop (which it is, twist or turn it as you wish).

I don't know, while the didn't use the word "flop" DriverHeaven was rather sour on the card as well, and even mentioned NV30, which to me is synonym to "flop" anyway.

n business it is always a good idea to learn from your mistakes and it also makes sense to learn from the mistakes of your competitors. It seems like AMD have not done so and throughout our experience with the card it has reminded us on more than a few occasions of the Geforce FX. Why?
First off the 2900XT is hugely late to market and as a result has trouble competing with an already well established part. In fact overall the XT is slower than the GTS when it matters. Next up we have the noise levels, don’t get us wrong, the 2900 isn’t anywhere near as loud as the FX was (thank god) however it is much louder than the competition. Whether watching movies or playing games the fan regularly speeds up and the variation in noise levels makes it incredibly difficult to mentally tune out. In this day and age there is no excuse for a noisy cooler on a high end card, Nvidia have very competently demonstrated how to create excellent cooling systems. The impact this has on the enjoyment of any task which involves audio is substantial.
Clearly AMD must have known that the XT was not going to beat the GTS and with the delays in launch they had time to rethink the product and add some unique selling points. One such point could have been the inclusion of full high definition video decoding such as that which will be featured in lower spec 2000 series products. Instead we have a card which when combined with any CPU other than the fastest available (high end dual core conroe) struggles with 1080p content. Even when trying to implement unique features things didn’t go well. The HDMI out with audio and video is a good idea in theory and it does work however we can’t recommend putting this card anywhere near the main TV in any house due to the noise issues mentioned above. It is just too intrusive.
We also can’t finish this review without mentioning that we had quite a nightmare finding AMD motherboards with onboard graphics which accepted the card without issue. AMD have investigated this issue however at this time it is unresolved (and unrepeatable within AMD) so we will reserve judgement on the issues which ranged from the Standard VGA driver not installing correctly (Minor) to the board failing to display the BIOS (major) and even on occasion not booting into windows at all.
All is not lost with the product though. Most of our main issues can be fixed by two changes; the first is a complete redesign of the cooling solution to create one which is as quiet, if not more so than it’s main competitors and the second is a drop of the retail price by at least £50 or $75 (from RRP $399) to compensate for the fact that overall the card is slower and has less memory than the nearest competition (The Geforce 8800 GTS 640Mb). If AMD or its partners can do both the Radeon HD becomes a much more attractive purchase, as it stands now however, we simply cannot recommend this hardware.

Galduta

14-May-2007, 09:47

Well , a good new in this game !! with AA the drop is normal :wink:

http://www.presence-pc.com/tests/AMD-Radeon-2900-XT-22674/19/

http://media.bestofmicro.com/3/Y/15694/original/056.gif

Pete

14-May-2007, 09:59

Yep, Presence-PC's review deserves a mention (http://translate.google.com/translate?u=http%3A%2F%2Fwww.presence-pc.com%2Ftests%2FAMD-Radeon-2900-XT-22674%2F23%2F&langpair=fr%7Cen&hl=en&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools) (<--Google-translated link). They've got some DX10 testing, too.

Geeforcer

14-May-2007, 09:59

Well, I am off to bed. Please keep posting links to the reviews as they come in, I will edit them in in the morning.

Geeforcer

14-May-2007, 10:28

Any ideas why 8xAF became so popular again all of the sudden while 8xAA suddenly vanished? For example, during their 8800 GTX review, ExtremeTech had tests with both, but none in the 2900 review.

Galduta

14-May-2007, 10:54

Any ideas why 8xAF became so popular again all of the sudden while 8xAA suddenly vanished? For example, during their 8800 GTX review, ExtremeTech had tests with both, but none in the 2900 review.

The deficiencies of the 2900 ? ;)

tEd

14-May-2007, 10:54

Neb

14-May-2007, 11:10

Read a few reviews of the HD2900XT and the perfomance is on par with a 8800GTS, sometimes a bit better and sometimes a bit slower. Even in DX10 (Call of Juarez) it's still competing against the 8800GTS. Although in some benches (a couple of them) it was almost nect to neck with the 8800GTX. Enabling AA and anisotropic filtering seems to make the HD2900XT perfomance dip quite a bit on several tests.
Neverthless let's see what future drivers make in favor of perfomance as long as they dont sacrifice visual IQ for rendering speed. :???:

tEd

14-May-2007, 11:36

vertex_shader

14-May-2007, 11:54

nyr

14-May-2007, 12:00

I heard COJ dx10 demo not runing on any g8x, how they test it? :smile:

It's a bug in the anti-aliasing, so disabling that allows it to run ok.

The game’s developer, Techland has found that this was caused by an application issue when MSAA is enabled. The benchmark can be run on both ATI Radeon and Nvidia hardware with MSAA disabled.

vertex_shader

14-May-2007, 12:05

It's a bug in the anti-aliasing, so disabling that allows it to run ok.

Thx :wink:

AnarchX

14-May-2007, 12:27

Lost Planet DX10 Benchmark
http://www.pcgameshardware.de/?article_id=601352
translation: http://translate.google.com/translate?u=http%3A%2F%2Fwww.pcgameshardware.de%2F %3Farticle_id%3D601352&langpair=de%7Cen&hl=de&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools

ouch...

vertex_shader

14-May-2007, 13:06

IbaneZ

14-May-2007, 13:56

http://www.firingsquad.com/hardware/amd_ati_radeon_hd_2900_xt_performance_preview/default.asp

Galduta

14-May-2007, 14:16

Of course there is some serious flaw with AA performance right now. Often slower than r580 and also drops to 1/3 performance 4x/noAA sometimes htat can't be how it supposed to be. I take it future driver will improve perfromance there.

Maybe is the AF ...

http://www.hothardware.com/articleimages/Item966/aa.png

EDIT / Review nonpublished

HL2: EPISODIO ONE & HardwareOC Tool v1.0 without AF and AA
2900XT/
1600x1200
med:161/
1920x1200
med:130/

HL2: EPISODIO ONE & HardwareOC Tool v1.0 AA@4x y FA@16x,
2900XT/
1600x1200
med:78/
1920x1200
med:68/

IbaneZ

14-May-2007, 15:32

http://www.tomshardware.com/2007/05/14/r600_finally_dx10_hardware_from_ati/

Only xbit and anand left now? :smile:

bigtabs

14-May-2007, 15:47

Wonder what's with DriverHeaven's Oblivion results (http://www.driverheaven.net/reviews/r600reviewz/oblivion.php)?

RobertR1

14-May-2007, 16:25

http://www.firingsquad.com/hardware/amd_ati_radeon_hd_2900_xt_performance_preview/default.asp

I've always enjoyed firing squads reviews.

It's sad when a person with a x1900xtx (mine oc'd to x1950xtx speeds) would find it hard to upgrade to the x2900xt. There's simply isn't enough to persuade me to spend the money on it. DX9 or DX10, it doesn't matter. The R600 can't cut it against the competition and is a marginal upgrade in a lot of cases over it's own previous gen.

INKster

14-May-2007, 16:48

Gelanin

14-May-2007, 17:02

It's especially frustrating when they even bothered to include results with both newer drivers in single-card mode and a dual-card setup, in the hopes to quash driver issues, etc.
Even with all their efforts the HD 2900 XT still trails the 8800 GTS 640MB at the high quality settings of a shader heavy game... :???:

It certainly doesnt trail the GTS in this review: http://www.digit-life.com/articles2/video/r600-part3.html

Bjorn

14-May-2007, 17:20

http://www.tomshardware.com/2007/05/14/r600_finally_dx10_hardware_from_ati/

Only xbit and anand left now? :smile:

Tom's came to a pretty funny conclusion:

Overall, HD2900XT is more forward looking than GeForce 8800 and it should be; it took an extra six months getting it to market. If you are leaning towards longevity of a card, R600 looks more attractive with a dedicated tessellator, programmable filters, high-clock speeds and crazy amounts of bandwidth.

Imo, the tesselator is interesting but will most likely not be used that often, especially since it's not a DX10 feature. The programmable filters makes things blurry and can afaik easily be done by the competition.

And since when is high clockspeeds automatically more forward looking ? And who cares about crazy amounts of bandwidth if that doesn't show in the benchmarks ?

INKster

14-May-2007, 17:22

It certainly doesnt trail the GTS in this review: http://www.digit-life.com/articles2/video/r600-part3.html

See the Firingsquad link above in the same game (using two different driver versions).

Geeforcer

14-May-2007, 17:28

Added Anands to the list.

RobertR1

14-May-2007, 17:42

Tom's came to a pretty funny conclusion:

Imo, the tesselator is interesting but will most likely not be used that often, especially since it's not a DX10 feature. The programmable filters makes things blurry and can afaik easily be done by the competition.

The program with Ati bringing "forward thinking" ideas is that they fail to court the developers and they never get used. It's no secret that nvidia's TWIMTBP program is highly successful in comparison to Ati's GITG (does it even exist anymore!?). So when nvidia is going out there and establishing relations all over, you can bet that developers will be using their hardware to develop, test and tune on. And really, why shouldn't they? Until Ati learns that dev support is as important as bringing new tech to the market, their efforts will mainly stay dormant until if/when nvidia decides to push those features forward. I'm sure you can find single instances where this isn't the case but in general, it is.

And since when is high clockspeeds automatically more forward looking ? And who cares about crazy amounts of bandwidth if that doesn't show in the benchmarks ?

Yeah, I don't get that either. And seeing OC'd results doesn't really make you feel all warm and fuzzy either. Highly clocked cards aren't scaling well to performance so it's not like we can say "sit back for a 65nm version with 850mhz core clock to really see what this card can do!" Back to the drawing board it is.

Arty

14-May-2007, 17:45

I'm half way through the review list (Rys' peice is definitely the best) and cant help but get the feeling that by the time devs begin to use the tesselator, Nvidia will have it too. Maybe AMD should now concentrate on performance first, features later.

Love_In_Rio

14-May-2007, 17:58

Imo, the tesselator is interesting but will most likely not be used that often, especially since it's not a DX10 feature. The programmable filters makes things blurry and can afaik easily be done by the competition.

And since when is high clockspeeds automatically more forward looking ? And who cares about crazy amounts of bandwidth if that doesn't show in the benchmarks ?

Anand points out in the first page of his review that programmable filters could be used as a solution due to ROPs being broken in the actual silicon revision. Broken hardware ? if so, i hope R650 to be far better.

Bjorn

14-May-2007, 18:02

Anand points out in the first page of his review that programmable filters could be used as a solution due to ROPs being broken in the actual silicon revision. Broken hardware ? if so, i hope R650 to be far better.

I read that too. Then there's some indications that TSMC's 80nm process is prone to leakage. Then there's the overclockability of the chip.

So things might actually look a lot better when they get the 65 nm R650 , or whatever it might be called, out of the door. Hopefully that'll happen sooner rather then later.

Love_In_Rio

14-May-2007, 18:10

I read that too. Then there's some indications that TSMC's 80nm process is prone to leakage. Then there's the overclockability of the chip.

So things might actually look a lot better when they get the 65 nm R650 , or whatever it might be called, out of the door. Hopefully that'll happen sooner rather then later.

I think it is very possible as i can´t imagine x2400 and x2600 leaving its AA power to the reduced number of ALUs.

swaaye

14-May-2007, 19:35

Honestly, the "tesselator" feels an awful lot like TruForm to me. I think it'll end up being wasted silicon. In 360, the devs will probably pay attention to something like that, but on PC things are much more about standardized features.

Star_Hunter

14-May-2007, 19:56

Tim Murray

14-May-2007, 20:13

I don't see how the tesselator is useless if its going to be required in a future version of DX (10.1?) based on what one of the reviewers said. However, I will say that even if that is the case R600 will be old before it would start seeing use and therefore it might not be as useful as Ati would like you to think.
Keep in mind that D3D is no longer caps based. There's no way that R600 is going to get support for a future D3D version. As it stands, yeah, the tesselator isn't going to be seeing a whole lot of use.

Dave Baumann

14-May-2007, 20:20

Keep in mind that D3D is no longer caps based. There's no way that R600 is going to get support for a future D3D version. As it stands, yeah, the tesselator isn't going to be seeing a whole lot of use.
There are methods where it can be accessed without necessarily needing and explicit cap in order for it to be utilised. Lets not forget OpenGL as well.

Tim Murray

14-May-2007, 20:31

There are methods where it can be accessed without necessarily needing and explicit cap in order for it to be utilised. Lets not forget OpenGL as well.
Is it exposed in the driver yet? I hate to bring it up, but I am vaguely reminded of the 9800's F-buffer...

Dave Baumann

14-May-2007, 20:43

Not in the shipping driver, but everyone that attended our press events has already seen it working, in operation under DX on Radeon HD 2900 - and pushing peaks of 1 Billion polys/sec no less. :cool:

RobertR1

14-May-2007, 21:03

Dave,

Is the R600 performance inline with AMD/ATI's internal expectations?
Many reviewers are placing blame on driver maturity for the lack of performance. Is this true?

Any other information you'd like to include on why the R600 has not lived upto public perception and why the performance is all over the charts would be greatly appreciated.

Thanks,
Robert.

Jack_Tripper

14-May-2007, 21:29

I've always enjoyed firing squads reviews.

It's sad when a person with a x1900xtx (mine oc'd to x1950xtx speeds) would find it hard to upgrade to the x2900xt. There's simply isn't enough to persuade me to spend the money on it. DX9 or DX10, it doesn't matter. The R600 can't cut it against the competition and is a marginal upgrade in a lot of cases over it's own previous gen.

Agreed, but you'd also have to agree that someone who upgrade in this generation soley to get DX10 features is a complete idiot too!

Unless ATI is planning on releasing a "high-end" variant, there's very little reason for an X1950 gen owner to consider upgrading (unless they wanted the incremental performance increases at insane resolutions like 20*15 and above) -

I guess I'm more disappointed in games (and the lack of reasoning to upgrade my hardware) than in the hardware itself - either generation is running current games at insane resolutions at "what most people would call" decent levels of performance - give me a piece of software, worth playing, that really pushes the envelope of "playable" and I'll upgrade

Jack

IbaneZ

14-May-2007, 21:50

Not in the shipping driver, but everyone that attended our press events has already seen it working, in operation under DX on Radeon HD 2900 - and pushing peaks of 1 Billion polys/sec no less. :cool:

That's more like it.

I don't understand why you and komb aren't more active today. Daamit needs some good PR daamit! :wink:

XMAN26

14-May-2007, 22:14

That's more like it.

I don't understand why you and komb aren't more active today. Daamit needs some good PR daamit! :wink:

Dave yes, Komb, no. Komb has been talking up the R600 for the last 2 weeks and now you can't find him. Even at rage3d where he was constantly dropping hits at stellar performance from this card. I have a feeling he is in hiding right now.

IbaneZ

14-May-2007, 22:29

Dave yes, Komb, no. Komb has been talking up the R600 for the last 2 weeks and now you can't find him.

Well, perhaps it's his day off? :lol:

http://www.rage3d.com/board/showthread.php?t=33890984

XMAN26

14-May-2007, 22:43

Well, perhaps it's his day off? :lol:

http://www.rage3d.com/board/showthread.php?t=33890984

Or maybe he found a very deep hole to hide in. Rage3d Nazies closing threads again.

DemoCoder

15-May-2007, 01:41

There are methods where it can be accessed without necessarily needing and explicit cap in order for it to be utilised. Lets not forget OpenGL as well.

Uh oh. Sounds like another FOURCC hack.

Dave Baumann

15-May-2007, 04:12

I don't recall that DX10 has FOURCC formats.

Brent

15-May-2007, 05:13

Thumbs up to Brent's HOCP review. It covered all the bases thoroughly, including IQ. My eagle eye did spy R600 filtering a tiny bit further out than G80 in their AF sshots (http://enthusiast.hardocp.com/image.html?image=MTE3ODk0MTA0NU1IN2xkMnFWY1dfNl8xX 2wucG5n), but it's really minor and may not apply to all games (or even angles).

Thank you Pete, it is much appreciated! Thanks for reading.

RobertR1

15-May-2007, 05:39

Thank you Pete, it is much appreciated! Thanks for reading.

Brent,

Not sure when you guys started the trend but the apples to apples part is very much appreciated in addition to the your standard style of reviews.

Evildeus

15-May-2007, 07:04

HFR Test (http://www.hardware.fr/articles/671-1/ati-radeon-hd-2900-xt.html)
With some branching info:
http://www.hardware.fr/medias/photos_news/00/19/IMG0019968.gif

Galduta

15-May-2007, 12:54

One small preview in one fan site (http://www.nextgpu.com/web/index.php?option=com_content&task=view&id=119&Itemid=31&limit=1&limitstart=5) - in Spanish -

http://img105.imageshack.us/img105/5545/scellgg1.png

Stalker indoor

http://img519.imageshack.us/img519/782/stalker2mf3.png

AlexV

15-May-2007, 13:40

WTF?Drivers used?System specs?SC:DA is a shit port with a ton of issues so I don`t really care for it, but those Stalker numbers are quite impressive...and unnaligned with what has been shown till now-so, again, WTF?

Galduta

15-May-2007, 14:10

Settings (http://www.nextgpu.com/web/index.php?option=com_content&task=view&id=119&Itemid=31&limit=1&limitstart=5)

Splinter Cell DA is very buggy , and this bench is with one tweak in Splinter Cell ini

In Stalker framerate was very bad and erratic until the reviewer reinstall the game . And the 2900 have 512 mb of VRAM - In Stalker videomemory usage is about 530 - 550 at 1600 x 1200 -

Before

First attempt

STALKER 1600x1200 bench 1
min:9
media:21
max:88

Second attempt

BENCH1
2900XT/8800GTS:

1600x1200
min:1/13
med:18/35
max:76/88

BENCH2
2900XT/8800GTS:

1600x1200
min:4/15
med:26/52
max:313/100

1920x1200 http://www.nextgpu.com/forum/Smileys/smilies_smf/crazy2.gif
min:31/11
med:51/44
max:102/90

.------------------------------------------------------------------------------------

After the reinstall the game
(http://www.nextgpu.com/web/index.php?option=com_content&task=view&id=119&Itemid=31&limit=1&limitstart=11)

IbaneZ

15-May-2007, 14:32

Same as before, enable AA and the card chokes.

vertex_shader

15-May-2007, 14:46

Settings (http://www.nextgpu.com/web/index.php?option=com_content&task=view&id=119&Itemid=31&limit=1&limitstart=5)

Splinter Cell DA is very buggy , and this bench is with one tweak in Splinter Cell ini

In Stalker framerate was very bad and erratic until the reviewer reinstall the game . And the 2900 have 512 mb of VRAM - In Stalker videomemory usage is about 530 - 550 at 1600 x 1200 -

Before

First attempt

STALKER 1600x1200 bench 1
min:9
media:21
max:88

Second attempt

BENCH1
2900XT/8800GTS:

1600x1200
min:1/13
med:18/35
max:76/88

BENCH2
2900XT/8800GTS:

1600x1200
min:4/15
med:26/52
max:313/100

1920x1200 http://www.nextgpu.com/forum/Smileys/smilies_smf/crazy2.gif
min:31/11
med:51/44
max:102/90

.------------------------------------------------------------------------------------

After the reinstall (http://www.nextgpu.com/web/index.php?option=com_content&task=view&id=119&Itemid=31&limit=1&limitstart=11)

How can be the card almost twice as fast in 1920x1200 than in 1600x1200?

Galduta

15-May-2007, 15:07

How can be the card almost twice as fast in 1920x1200 than in 1600x1200?

Is one problem in the reviewer system .In another system of a friend with one 2900 the result was normal.

Allendalle, 2MB caché 7x466=3,26Ghz, Ram 1:1 933Mhz 5-5-5-15 2x1GB
a 1600x1200 y 743/743/1656Mhz.

1)STALKER v1003, AFMax y noAA.

-Timedemo1: Av: 37,55, Min: 19,79, Max: 123,84,
-Timedemo2: Av: 60,14, Min: 37,60, Max: 291,21

AlexV

15-May-2007, 15:14

I`m still in WTF mode. They seem to use the latest alpha(I`ve only skimmed the article) driver, but still...their numbers are really against the current flow.

marifire

15-May-2007, 16:11

I`m still in WTF mode. They seem to use the latest alpha(I`ve only skimmed the article) driver, but still...their numbers are really against the current flow.
Not really, i myself contributed to that review. We repeated each test 4 times, and the reviewer need to reinstall STALKER cause of very weird results showed in the first runs, but we finally got the good ones which are posted in the final review:
http://www.nextgpu.com/web/index.php?option=com_content&task=view&id=119&Itemid=31

The drivers are as follows:
8.37.4.0...almost equal performance as the ones on AMds website right now 8.37.4.3. The question is we run different games benchmarks for the task. HD 2900 XT is a good card, but it still losees a lot when AA is enabled. It still shines in some games too...Please, have a look also to the following review:
http://www.matbe.com/articles/lire/383/ati-r600-radeon-hd-2900-xt/page1.php

Geeforcer

15-May-2007, 17:47

Usually when scores go up with resolution it means that something that was working before got disabled.

Arty

15-May-2007, 18:29

In dutch but http://www.hardware.info/nl-NL/articles/amdnZGpqZGia/ATI_Radeon_HD_2900_XT_test/1

I cant seem to find which drivers they used but the XT is quite faster than the GTS (even with AA, AF).

swaaye

15-May-2007, 19:09

I'd have that reviewer make sure his card is running PCIe 16X. When you overclock these 775 chipsets, sometimes PCIe drops to 1X. And it may only do it on the occasional reboot.

It would be best to test on a stock clocked machine though. Overclocking adds variables.

NocturnDragon

15-May-2007, 19:26

I don't recall that DX10 has FOURCC formats.

When will the whitepapers of the R600 be released? Programming guides or something like that?

leoneazzurro

15-May-2007, 19:37

Hardware.fr speaking about R600

http://www.hardware.fr/articles/671-4/ati-radeon-hd-2900-xt.html

15-May-2007, 22:57

In dutch but http://www.hardware.info/nl-NL/articles/amdnZGpqZGia/ATI_Radeon_HD_2900_XT_test/1

I cant seem to find which drivers they used but the XT is quite faster than the GTS (even with AA, AF).

Here's the English version: http://www.hardware.info/en-UK/articles/amdnZGpsZGGa/ATI_Radeon_HD_2900_XT_test/ (http://www.hardware.info/en-UK/articles/amdnZGpsZGGa/ATI_Radeon_HD_2900_XT_test/1)

Galduta

16-May-2007, 00:33

http://www.xbitlabs.com/articles/video/display/r600-architecture_14.html

http://www.xbitlabs.com/images/video/radeonhd2000/test_results/2900xt_noise_1m.gif

http://www.xbitlabs.com/images/video/radeonhd2000/test_results/2900xt_power.gif
Even though the GeForce 8800 GTX is not a chip that loves to save energy, the Radeon HD 2900 XT consumes even more: 161W is a new record for our lab. In fact, ATI claims that the Radeon HD 2900 XT can consume 215W, which is even more than we got, therefore, a noisy cooling system is not a surprise…

marifire

16-May-2007, 00:45

trinibwoy

16-May-2007, 02:29

Hardware.fr speaking about R600

http://www.hardware.fr/articles/671-4/ati-radeon-hd-2900-xt.html

Wow, it does pretty good there.

Hmmmm newegg has some bundles with a free G5 if you buy an XT before the 21st. That's a damn good mouse.

Dave Baumann

16-May-2007, 04:17

When will the whitepapers of the R600 be released? Programming guides or something like that?
Well, my group is working on the whitepapers and they should be a long soon. There will be an SDK and inevitably documentation from the ISV group, although I do not know their schedule.

Dave Baumann

16-May-2007, 04:18

Hmmmm newegg has some bundles with a free G5 if you buy an XT before the 21st. That's a damn good mouse.
I think its the graphics card thats the bit thats worth 400 bucks though! ;)

2900guy

16-May-2007, 05:48

I'm seeing a pattern of non USA sites potentially having a different driver than their European counterparts. Take this review for example, the 2900 looks WONDERFULL. The hit with AA is still there but nowhere as severe as most USA sites report. Similar results to this cane be found in other non USA sites.

http://www.hardware.info/nl-NL/articles/amdnZGpqZGia/ATI_Radeon_HD_2900_XT_test/16

ChrisRay

16-May-2007, 07:29

http://www.xbitlabs.com/articles/video/display/r600-architecture_16.html

Seems to have the most detailed comparison of CFAA I have seen so far. Includes performance and quality of narrow/wide tents.

MulciberXP

16-May-2007, 07:59

I think its the graphics card thats the bit thats worth 400 bucks though! ;)

Actually I have yet to see one for less than $429

2900guy

16-May-2007, 09:34

http://www.hardware.info/

Removed the "leeching"

Example performance graph from www.hardware.info:

http://www.hardware.info/nl-NL/productdb/ZGNnlmZpyZ9l/bGRkZJibmJTK,bGNkbZiTmJfK,bGNkapiUmJTK,bGNkapiZmJn K,bGNkapiUmJPK,bGRkZJiXmJXK,bGRkZJiXmJTK,bGNkbZiYm JjK,bGNka5iUmJPK,bGNkapibmJbK/0000FF,FF0000,FF0000,FF0000,FF0000,00FF00,00FF00,0 0FF00,00FF00,00FF00/benchmark/

For the remainder please go to http://www.hardware.info/
_ (http://www.hardware.info/nl-NL/productdb/ZGNnlmZqyZll/bGRkZJibmJTK,bGNkbZiTmJfK,bGNkapiUmJTK,bGNkapiZmJn K,bGNkapiUmJPK,bGRkZJiXmJXK,bGRkZJiXmJTK,bGNkbZiYm JjK,bGNka5iUmJPK,bGNkapibmJbK/0000FF,FF0000,FF0000,FF0000,FF0000,00FF00,00FF00,0 0FF00,00FF00,00FF00/benchmark/)
_ (http://www.hardware.info/nl-NL/productdb/ZGNnlmZnyZ1l/bGRkZJibmJTK,bGNkbZiTmJfK,bGNkapiUmJTK,bGNkapiZmJn K,bGNkapiUmJPK,bGRkZJiXmJXK,bGRkZJiXmJTK,bGNkbZiYm JjK,bGNka5iUmJPK,bGNkapibmJbK/0000FF,FF0000,FF0000,FF0000,FF0000,00FF00,00FF00,0 0FF00,00FF00,00FF00/benchmark/)
_ (http://www.hardware.info/nl-NL/productdb/ZGNnn2ZqyZxl/bGRkZJibmJTK,bGRkZJiXmJXK,bGRkZJiXmJTK,bGNkbZiYmJj K,bGNkapibmJbK/0000FF,00FF00,00FF00,00FF00,00FF00/benchmark/)

_ (http://www.hardware.info/nl-NL/productdb/ZGNnlmZpyZ1l/bGRkZJibmJTK,bGNkbZiTmJfK,bGNkapiUmJTK,bGNkapiZmJn K,bGNkapiUmJPK,bGRkZJiXmJXK,bGRkZJiXmJTK,bGNkbZiYm JjK,bGNka5iUmJPK,bGNkapibmJbK/0000FF,FF0000,FF0000,FF0000,FF0000,00FF00,00FF00,0 0FF00,00FF00,00FF00/benchmark/)

_ (http://www.hardware.info/nl-NL/productdb/ZGNnlmZnyZll/bGRkZJibmJTK,bGNkbZiTmJfK,bGNkapiUmJTK,bGNkapiZmJn K,bGNkapiUmJPK,bGRkZJiXmJXK,bGRkZJiXmJTK,bGNkbZiYm JjK,bGNka5iUmJPK,bGNkapibmJbK/0000FF,FF0000,FF0000,FF0000,FF0000,00FF00,00FF00,0 0FF00,00FF00,00FF00/benchmark/)

Geeforcer

16-May-2007, 09:51

I must be the only person on this forum who "no leeching" rule applies to.

trinibwoy

16-May-2007, 11:08

I think its the graphics card thats the bit thats worth 400 bucks though! ;)

Still, the game (and now mouse bundle) make for a much more compelling argument given that most people still don't see the XT as a great value at $400. With current drivers at least.

ChrisRay

16-May-2007, 11:33

I dunno about you. But large graphs loading and stretching the forums are pretty annoying. Cant you just directly link to them?

IbaneZ

16-May-2007, 12:28

I dunno about you. But large graphs loading and stretching the forums are pretty annoying. Cant you just directly link to them?

Well, the whole idea about cherry picking is to slap the pics in the readers face. Who has time to click the links? :lol:

2900guy, or x800guy (his rage3d name) has a bad habit of doing this. :wink:

dizietsma

16-May-2007, 13:12

Still, the game (and now mouse bundle) make for a much more compelling argument given that most people still don't see the XT as a great value at $400. With current drivers at least.

I wonder what Dave Orton would have thought 12 months ago if you suggested that he sell his next generation 700m transistor gpu for $399 with a mouse thrown in for free ?

AlexV

16-May-2007, 13:46

Not really, i myself contributed to that review. We repeated each test 4 times, and the reviewer need to reinstall STALKER cause of very weird results showed in the first runs, but we finally got the good ones which are posted in the final review:
http://www.nextgpu.com/web/index.php?option=com_content&task=view&id=119&Itemid=31

The drivers are as follows:
8.37.4.0...almost equal performance as the ones on AMds website right now 8.37.4.3. The question is we run different games benchmarks for the task. HD 2900 XT is a good card, but it still losees a lot when AA is enabled. It still shines in some games too...Please, have a look also to the following review:
http://www.matbe.com/articles/lire/383/ati-r600-radeon-hd-2900-xt/page1.php
Anyway, they´re using 8.37.4.2 there...but we were using ATIs oficial.

Don`t get me wrong, I`m not claiming a hidden agenda was involved. It`s simply odd that the numbers were quite different from what others have shown. Props for your contribution to the review, BTW.

fallguy

16-May-2007, 15:44

Well, the whole idea about cherry picking is to slap the pics in the readers face. Who has time to click the links? :lol:

2900guy, or x800guy (his rage3d name) has a bad habit of doing this. :wink:

I dont mind, but then again Im not running a wimpy res. :o

wishiknew

16-May-2007, 16:44

The benchmarks 2900guy linked to, I thought the HD 2900xt was being blown out by the 1950xt in Far Cry.

2900guy

16-May-2007, 16:59

thats just it, nearly every site is different. i noticed european sites show the 2900 more favorably. maybe they received different drivers?

swaaye

16-May-2007, 19:49

I wonder what Dave Orton would have thought 12 months ago if you suggested that he sell his next generation 700m transistor gpu for $399 with a mouse thrown in for free ?

LOL. Yeah must be rather surreal for him to reflect upon that.

But, hey, even with this thing's wildly unpredictable performance, depressing AF/AA perf hit, etc, it does really compete quite well with GTS and even GTX occasionally. But I definitely think a GTS 640MB OC is the better deal. $100 less, typically.

dizietsma

16-May-2007, 20:44

LOL. Yeah must be rather surreal for him to reflect upon that.

But, hey, even with this thing's wildly unpredictable performance, depressing AF/AA perf hit, etc, it does really compete quite well with GTS and even GTX occasionally. But I definitely think a GTS 640MB OC is the better deal. $100 less, typically.

I agree, eventually it will all come out in the wash, with the XT looking whiter than white, but probably by that time most people will have already hung it out to dry.

neliz

16-May-2007, 21:17

But I definitely think a GTS 640MB OC is the better deal. $100 less, typically.

That's the thing, here in europe it's the other way around. You'll typically shell out between €360 and €380 for an XT which is the exact pricepoint of the 640 (€350 for the Club3D and €445 for the overclocked EVGA model.)

So mail in rebates and price cuts might benefit the US market, the XT is very comfortable at it's current pricepoint over here..

Twinkie

16-May-2007, 21:26

http://www.bit-tech.net/hardware/2007/05/16/r600_ati_radeon_hd_2900_xt/1

Bit-tech's review is up.

Arty

16-May-2007, 23:56

http://www.bit-tech.net/hardware/2007/05/16/r600_ati_radeon_hd_2900_xt/1

Bit-tech's review is up.

Good job Tim! http://forum.beyond3d.com/images/icons/icon14.gif

marifire

17-May-2007, 00:05

Don`t get me wrong, I`m not claiming a hidden agenda was involved. It`s simply odd that the numbers were quite different from what others have shown. Props for your contribution to the review, BTW.

Thank you ;)

I really think we will see some good numbers coming from HD 2900 XT very soon...but time will tell. There are too many different driver sets and none of them is a final one. Also, different benchmarks have been run on different reviews. Just seeing how much did our STALKER results change with a game reinstall tells you a lot about the current situation. Of course, this is nothing but sad for AMD-ATI, but has nothing to do with the performance of the card which i think will improve in just a few days...

Galduta

17-May-2007, 00:18

Yes is real !!! AA 4X Free in R600

http://www.matbe.com/images/biblio/art_ati-r600-radeon-hd2900xt/000000057007.png

http://www.matbe.com/images/biblio/art_ati-r600-radeon-hd2900xt/000000057006.png

:wink: Source (http://www.matbe.com/articles/lire/383/ati-r600-radeon-hd-2900-xt/page20.php)

Geeforcer

17-May-2007, 00:21

Most amazingly though, the mere existence of R600 retroactively made AA free on R580 too! Way to go AMD!

Brent

17-May-2007, 00:23

AA does not work in Vegas, I believe it uses deferred lighting.

AlexV

17-May-2007, 00:40

Thank you ;)

I really think we will see some good numbers coming from HD 2900 XT very soon...but time will tell. There are too many different driver sets and none of them is a final one. Also, different benchmarks have been run on different reviews. Just seeing how much did our STALKER results change with a game reinstall tells you a lot about the current situation. Of course, this is nothing but sad for AMD-ATI, but has nothing to do with the performance of the card which i think will improve in just a few days...

Dunno...what`s holding me back from jumping on the whole-drivers will make it GTX beating material is the fact that AMD priced it the way it did. If they were certain that they could eventually grab nV by the balls in terms of across the board performance they would've priced it higher. It`s not like they can rethink the whole stuff once they get better drivers and say:sorry, we've just released CAT7.6, with much added performance. Also, our prices will now be 499$(for example). And since I`m a prick and don`t think that AMD are sent by God to do only good and I assume they actually want to sell their stuff as expensive as possible(in an envelope allowing it to sell), and factoring that the chip is likely not cheap(which means lowish margins, if any, at current price-point), I think that it`ll not become a GTX conqueror ever. I`d love to be proven wront though:)

Star_Hunter

17-May-2007, 00:44

So you would prefer that AMD sets it at $499 and and gets flamed for the next 3-4 months about the price and by time the drivers are getting decent R650 is out?

MulciberXP

17-May-2007, 01:00

So you would prefer that AMD sets it at $499 and and gets flamed for the next 3-4 months about the price and by time the drivers are getting decent R650 is out?

I cant even remotely see how that was implied...

Galduta

17-May-2007, 01:19

AA does not work in Vegas, I believe it uses deferred lighting.

It work with the forceware 160.03 in XP, , but with some errors in the shadows . But not many , not is frequent . The perfomance drop is great

http://img515.imageshack.us/img515/9351/r6vegasgame200705160316an9.th.jpg (http://img515.imageshack.us/my.php?image=r6vegasgame200705160316an9.jpg)http://img515.imageshack.us/img515/8370/r6vegasgame200705160317bk2.th.jpg (http://img515.imageshack.us/my.php?image=r6vegasgame200705160317bk2.jpg)http://img515.imageshack.us/img515/9798/r6vegasgame200705160315zg0.th.jpg (http://img515.imageshack.us/my.php?image=r6vegasgame200705160315zg0.jpg)http://img515.imageshack.us/img515/1992/r6vegasgame200705160057vj2.th.jpg (http://img515.imageshack.us/my.php?image=r6vegasgame200705160057vj2.jpg)http://img172.imageshack.us/img172/2359/r6vegasgame200705160057ry2.th.jpg (http://img172.imageshack.us/my.php?image=r6vegasgame200705160057ry2.jpg)http://img329.imageshack.us/img329/460/r6vegasgame200705160056iz5.th.jpg (http://img329.imageshack.us/my.php?image=r6vegasgame200705160056iz5.jpg)

U3 engine dont employ deferred lighting , is the shadowing algorythm the problem

http://forum.beyond3d.com/showpost.php?p=976631&postcount=6

Most amazingly though, the mere existence of R600 retroactively made AA free on R580 too! Way to go AMD!

:mrgreen::mrgreen::mrgreen:

ninjaspa

17-May-2007, 05:28

nopp0x00

17-May-2007, 07:22

r600 is looking might nice according to matbe... is it just me or are overseas reviews a lot more favorable of 2900xt?
partisanship seems a lot more rampant from stateside websites.

still, overall i've seen about 12 of 15 sites recommend ati's new hardware. not bad at all

Why would any sane site recommend an XT over a GTS or GTX?

A GTS matches it (and exceeds it at times) in performance and it doesn't have any of the image quality issues listed on this site:

http://www.xbitlabs.com/articles/video/display/r600-architecture_16.html

Oh, and it's significantly cheaper as well and can be overclocked nicely too.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814130071

http://www.newegg.com/Product/Product.aspx?Item=N82E16814130080

R600 is a bust until the new revision comes out.

Moloch

17-May-2007, 08:06

http://www.bit-tech.net/hardware/2007/05/16/r600_ati_radeon_hd_2900_xt/1

Bit-tech's review is up.

Nice quake 4 performance :smile:
Perhaps becase it finally has decent stencil performance?

neliz

17-May-2007, 09:36

Why would any sane site recommend an XT over a GTS or GTX?

Because not everyone lives in the US and thus the GTS isn't automagically cheaper in the rest of the world?
Because it does deliver major improvements in areas where we didn't expect it?
Because crossfire is improved beyond SLI?
Because it offers great performance when running HDR?
Because it seems to do great with Unreal Engine 3 (or 2.5) based games?
- This seems to be true in more games it has been "profiled"for
Because some people want the highest 3dMarks?

Geeforcer

17-May-2007, 09:54

...
Because it does deliver major improvements in areas where we didn't expect it?
...
Because it offers great performance when running HDR?
Because it seems to do great with Unreal Engine 3 (or 2.5) based games?
- This seems to be true in more games it has been "profiled"for
...

Neliz, could you be more specific in regard to these three points?

marifire

17-May-2007, 10:06

Dunno...what`s holding me back from jumping on the whole-drivers will make it GTX beating material is the fact that AMD priced it the way it did. If they were certain that they could eventually grab nV by the balls in terms of across the board performance they would've priced it higher. It`s not like they can rethink the whole stuff once they get better drivers and say:sorry, we've just released CAT7.6, with much added performance. Also, our prices will now be 499$(for example). And since I`m a prick and don`t think that AMD are sent by God to do only good and I assume they actually want to sell their stuff as expensive as possible(in an envelope allowing it to sell), and factoring that the chip is likely not cheap(which means lowish margins, if any, at current price-point), I think that it`ll not become a GTX conqueror ever. I`d love to be proven wront though:)
Well, its $349-399 € price point here, so it shouldn´t go for GTX or Ultra, but just GTS 640MB...it could be very good at some situations and compete well with GTX and Ultra, but don´t really think that will happen in most games.

AlexV

17-May-2007, 11:23

So you would prefer that AMD sets it at $499 and and gets flamed for the next 3-4 months about the price and by time the drivers are getting decent R650 is out?

Read again. Attempt to understand. Stop petting the ATi logo for a moment and understand that AMD/ATi are in this for the money. If in a 3-4 months timeframe they would`ve expected it to hit over/equal GTX performance across the board, they would`ve priced it higher, because, you know, you can afford to get some flak if you add about 100 bucks to your margins per unit sold.

Marifire...I tend to agree. Is this as noisy as the X1900XT was?

neliz

17-May-2007, 11:35

Neliz, could you be more specific in regard to these three points?

Forgive my pink glasses ;)

Running R6/SC at higher resolutions (16x12 and up) with AF most reviews show the XT on par with the GTX or trailing in the region of a few frames, more indicative of the "expected" performance.
TechReports ShaderMark performance shows the only place where the 2900 will actually win is in the HDR tests.
http://www.techreport.com/reviews/2007q2/radeon-hd-2900xt/index.x?pg=3.

With profiled I mean there are a few cases in which the XT seems to work "normally"
Company of Heroes performance is normal for instance, enabling AF or AA doesn't make the performance jump around like it does in Oblivion for instance. This (to me) looks like the scheduler is doing it's job just right there. Take stalker, it performance on GTS level at low resolutions but climbs up when resolution and AA/AF are enabled. This erratic behavior really looks like driver issues.

neliz

17-May-2007, 11:48

If in a 3-4 months timeframe they would`ve expected it to hit over/equal GTX performance across the board, they would`ve priced it higher

It depends how quickly it will be replaced by an R650 part. but I certainly know I wouldn't have bought it if it was priced at €480.

Ramalth

17-May-2007, 11:57

Hello everyone. New around here.

Read again. Attempt to understand. Stop petting the ATi logo for a moment and understand that AMD/ATi are in this for the money. If in a 3-4 months timeframe they would`ve expected it to hit over/equal GTX performance across the board, they would`ve priced it higher, because, you know, you can afford to get some flak if you add about 100 bucks to your margins per unit sold.

Marifire...I tend to agree. Is this as noisy as the X1900XT was?

Yeah, but if they (ATI/AMD) added 100$ to the price with the promise of better performance in 4 or 5 months, NOBODY would buy them right now... So they set the price according to present perfomance, and work on improving it while still having sales. What would happen if they sort out what seems to be drivers problems and increase perfomance to near GTX level? Well, they would have much better sales numbers, thats for sure... And they would still be a company making money (more sales = more money), not a charity.

Skinner

17-May-2007, 12:44

Some remarkable COH numbers:

http://www.dinoxpc.com/Tests/articoli/articolo/video_jetway_hd2900xt_images/video_4.gif

Sound_Card

17-May-2007, 12:57

R600 seems to win some, then lose some. This may be beaten to death already, but I would think some stuff in R600 is broken. Some games, R600 is slower than R580, then in a few others, it's around GTX speed. The card seems to be on and off across the board. From what I'm seeing, I would think their is some decent amount of driver head room in the chip.

7.5's are due anyday now. So we will see what that brings.

Tridam

17-May-2007, 15:17

AA does not work in Vegas, I believe it uses deferred lighting.

AA is not supported by the game. However Nvidia implemented it directly in the drivers just like it happened with Oblivion. I told AMD about this and Chuck should be working on it ;)

AlexV

17-May-2007, 17:16

Hello everyone. New around here.

Yeah, but if they (ATI/AMD) added 100$ to the price with the promise of better performance in 4 or 5 months, NOBODY would buy them right now... So they set the price according to present perfomance, and work on improving it while still having sales. What would happen if they sort out what seems to be drivers problems and increase perfomance to near GTX level? Well, they would have much better sales numbers, thats for sure... And they would still be a company making money (more sales = more money), not a charity.

This is getting tiresome...how can I explain so that eventuall somebody gets it. Oh, I get it....NO. Considering the hype that surrounded it, the fact that generally you can get away with a lot of stuff when launching a new product when the market has a certain mindset towards you(like everybody seems to be vigorously stroking when the ATi logo appears, singing Hallelujah, hail thee mighty saviour, doer of no wrong), mindset of which you are aware, you make the best of it if you know you won`t be caught by the balls later(in this case getting caught by the balls meaning inability to produce comparative/better performance across the board, as I`ve stated previously). Considering everyone believes almost anything ATi says(family launch, hugeish improvements through drivers, we're not late, everything is spiffy etc.), they could`ve safely gotten away with a higher price on it if they had utmost certainty that they`d outgun the GTX. The only other explanation would be that they`re late enough that the R650 is so near it wouldn`t have made sense and it would`ve put them in an awkward position, but there`s no noise about R650 yet, whilst in the R520->R580 transition there was quite a bit of it.

It may seem like it`s not possible but, really, you can`t sell at any price in any condition-if you have crap margins(how many dice per wafer do you think come back in tippy top X2900 fit shape, considering it`s a 700 mln trannie chip with fairly highish complexity aimed at a high clockrate), or you`re selling at a loss, you`re not doing much good to yourself. Because you have to also recuperate your investment, these chips don`t come out of thin air you know, R&D is expensive, for example. More sales=more money...more real money or more imaginary money?Coz, you know, if you make 1$ per card and it costs you 1.1$ to make it, factoring in everything you`ve put in, you`re actually losing money. Just food for thought.

Star_Hunter

17-May-2007, 17:42

mito

17-May-2007, 17:55

The ATI Radeon HD 2900 XT however is more akin to NVIDIA’s GeForce FX 5800

from HardOp.................. :|

Geeforcer

17-May-2007, 17:58

Forgive my pink glasses ;)

Running R6/SC at higher resolutions (16x12 and up) with AF most reviews show the XT on par with the GTX or trailing in the region of a few frames, more indicative of the "expected" performance.

Isn't HDR bugged in R6 on 2900XT?

AlexV

17-May-2007, 18:13

It should be noted that ATI only pays for the chips that work, therefore bad yields only hurt supply. If you set it at $499 almost no one would buy, its that simple. You talk about margins yet no one really knows the margins on the R600. I doubt they are losing money at $399 per card (although it could be losing R&D money overall) but sometimes you have to take a loss in order to limit the customer base loss. If the drivers were easy to write for the R600 and they know they can reach the GTX then they may have gone for the $499 pricepoint, but since its hard to code for it they wanted to make sure they were covered if they couldn't do it.

Why would almost no one buy?Why did ppl buy the FX, which was arguably worse than this at release(yes, even in DX8 it pretty much sucked balls when AA came in, it was louder etc.)?Why give a huuuuge hint(lowish price) to your extremely loyal customer base that you can`t compete at the high-end?Because they`re nice guys?I see a billion ppl whining that they made a poor choice by waiting for it, and ignoring that basically it doesn`t suck as a card and it`s ok. I`d argue many would`ve swarmed over it at a higher price-point, because it`s ATi and they always improve, and the fact that it does ok in some benchies would`ve fed this line of thought. Joe Doe rich prick doesn`t know anything worth a damn about ALUs and such, but he pretty much equates performance with price. It`s a mindset thing. And I doubt that AMD marketers are retarded and ignore. They`ve priced it as such, IMHO, because they knew they wouldn`t get away with pricing it higher and thus competing at the high-end, not because they feared some customer-base erosion(which is unlikely to have occured anyhow) and not because they want....let me remember this one...umm, market penetration.IMHO.

Subtlesnake

17-May-2007, 19:51

Nobody bought the FX 5800. The 5900 did better, but the Steam survey showed that only a small minority of gamers purchased them (when compared with the 9800). Nvidia's past dominance in the industry mattered little - the FX series tanked, despite offering very competitive performance in a lot of cases, with the FX 5700 being in a dead heat with the 9600 Pro.

Had AMD priced the card higher there's a real risk that a similarly negative perception of the series could have been created.

AlexV

17-May-2007, 20:05

Nobody?Mild exaggeration there?And the Steam survey is a really really poor source of information. In terms of marketshare, even though it sucked quite badly, the entire FX line didn`t hurt nVidia in a sinister way...they lost some, yes, but financially I think they were doing well even then, with good margins etc.

Anyway, this is pointless. Anybody who likes ATi will see the pricetag as their ultimate gift to end-users, proof of their ongoing love for the people etc. Since I don`t have any inclination towards any IHV, I see it as something forced, indicating that they`re in a tigtish spot. Overall, it`s an irrelevant argument.

Geeforcer

17-May-2007, 20:12

SugarCoat

17-May-2007, 20:35

While I am sure that NV3x sales on whole were much worse than R3xx, for that particular generation Steam survey is not the best measure of relative sales simply due to the fact that HL2 was bundled with Radeons. I wonder if this effect will again surface in this generation.

The actual voucher only came with some laptops equipped with 9600s and retail boxed 9800/9600XTs (the latter of which were rare). Not every card. You're also implying that a significant number of people went out and spent hundreds of dollars simply to have a voucher for a game that wasnt due out for quite some time.

Furthermore the survey has been restarted three or four times since it was first started, the last time being in November of 06, so i think that debunks numbers skewing from vouchers pretty much entirely, not to mention the second most common card is infact the FX 5200 (also one of the last cards sold in PCI version which i dont think is a coincidence). Its problably the most accurate and varried survey for its size in existence and an excellent source of who is currently using what.

Geeforcer

17-May-2007, 21:08

What I am implying is that the percentage of R9800/R9600 owners who were likely to participated in the survey is much higher than the percentage of FX 5xxx owners. That's all.

neliz

17-May-2007, 21:22

Isn't HDR bugged in R6 on 2900XT?

I'll tell you monday

Geeforcer

17-May-2007, 23:07

I'll tell you monday

Ok, thanks. Looking forward to it.

PeterAce

17-May-2007, 23:43

Was really hoping for a IXBT/Digital-Life review of R600 that used Rightmark3D (compared to 8800 GTX/GTS, X1950 and 7900) :(

Arnold Beckenbauer

18-May-2007, 00:59

Was really hoping for a IXBT/Digital-Life review of R600 that used Rightmark3D (compared to 8800 GTX/GTS, X1950 and 7900) :(
You mean this one?
http://www.digit-life.com/articles2/video/r600-part2.html
http://www.digit-life.com/articles2/video/r600-part1.html

Tim Murray

18-May-2007, 02:18

Murray's Law: Every thread comes back to NV3x versus R3x0. Every single one.

Razor1

18-May-2007, 03:20

Some remarkable COH numbers:

PIC

Agian these are the drivers that replace the SSAAA mode with EATM.

(http://www.dinoxpc.com/Tests/articoli/articolo/index.asp?id=644&p=4#)DirectX 9 (http://www.dinoxpc.com/Tests/articoli/articolo/index.asp?id=644&p=4#).0c, Catalyst 7.4, Forceware 158.22, Catalyst 8.37.4.2
(http://www.dinoxpc.com/Tests/articoli/articolo/index.asp?id=644&p=4#)

BRiT

18-May-2007, 04:17

Murray's Law: Every thread comes back to NV3x versus R3x0. Every single one.

Even threads talking about 3dfx Rampage? :wink:

Tim Murray

18-May-2007, 04:45

Even threads talking about 3dfx Rampage? :wink:
Yes, because the question of "how much of NV30 was based on Rampage" is inevitably raised.

Geeforcer

18-May-2007, 04:49

No, the question whenever a new card comes out is: "Is this finally the card that could have come close to Rampage, had it been released? For example, it is possible that 4 OCed 8800 Ultras could matched the power of the Rampage.

Trawler

18-May-2007, 05:09

Yes, because the question of "how much of NV30 was based on Rampage" is inevitably raised.

Cmon, everyone knows it was nv40 that was based on Rampage! :wink:

silent_guy

18-May-2007, 09:03

It should be noted that ATI only pays for the chips that work, therefore bad yields only hurt supply.
It should also be noted that your statement is complete fiction.

Fabless companies pay per wafer, not per chip.

Star_Hunter

18-May-2007, 09:49

It should also be noted that your statement is complete fiction.

Fabless companies pay per wafer, not per chip.

Well maybe thats not the case anymore but I know I read something that said that once I guess it was wrong.

While I don't think this was where I got the idea from its the only one I could find in a few minutes

"In addition, half of the inventory bloat was due to ATI underestimating the yields on its wafer by a factor of roughly 40% (i.e., the wafers yielded 40% more die than expected), further compounding its inventory glut (under ATI’s die buy model, ATI purchases on a per-die as opposed to per-wafer basis, implying that it has secured the price per die based on theoretical yields from the fab, and has committed to purchase the entire wafer no matter what the yield)"

http://forum.beyond3d.com/showthread.php?p=553087#post553087

I guess I miss understood that comment.

AlexV

18-May-2007, 13:03

You did. TSMC and other foundries would be out of business otherwise.

Geo

18-May-2007, 14:54

I'm going to guess that TSMC employs an actuary or two that Lloyd's would be more than happy to have on staff, and will take either end of the die vs wafer proposition you'd like them to, and have the end results to TSMC's profits not differ materially.

18-May-2007, 15:55

I'll tell you monday

Why wait till monday, tell us now? :)

Ok so I guess only I thought that was funny huh?

/me hangs up his short lived career as a comedian :D

neliz

18-May-2007, 16:11

Why wait till monday, tell us now? :)

Ok so I guess only I thought that was funny huh?

/me hangs up his short lived career as a comedian :D

I have no problem telling you now though.. but it'll be based on BS. I first need to bring RB6:LV and my 2900 together.. my brother has my copy of rb6.. :(

silent_guy

18-May-2007, 16:11

I guess I miss understood that comment.
It's hard to misunderstand the comment in any other way than 'ATI pays per working die'. Maybe it wasn't as much fiction as I though it was.:wink:

So let me qualify: some companies, like LSI Logic and IBM, will happily produce your chips at a fixed cost per working die. But in that case, those companies have full control over the backend process: they are given a netlist and they are responsible for scan and BIST insertion, analog cells, memories, P&R, the works. In such a case, they have full control and understanding of the whole chip and that makes it much easier to avoid design related yield problems.

Customers with a COT (Customer Owned-Tooling) flow, do everything themselves and dump a GDS2 file at the fab. It's basically a black box. This makes it harder and riskier to estimate correct yields (as apparently evident in the case of R520?). It also makes it a bit harder to set price models when you're dealing with redundant logic and lower-functionality parts.

So I always worked in the fixed wafer cost model and thought that this was the mode of operations of TSMC/UMC/...

As Geo wrote, a decent actuarian should find a way to make the numbers work for both parties and apparently that's what happened for R520.
But since in that case the fab is taking on all the risk, the customer is going to pay a more stable but higher price for it.

Jawed

18-May-2007, 16:38

R520 may have been ATI's first PC GPU with fine-grained ALU redundancy.

Admittedly, in R520, this would have been 1 extra ALU for each 4, so not amazingly fine-grained (as it would seem to be in R600, 1 for each 16).

So, perhaps the redundancy worked better than expected? That seems pretty unlikely though, doesn't it?

Presumably fabs and customers, jointly, are used to fine-grained redundancy when memory is a major part of a die. Again, relatively speaking, R520 was prolly a step-change in register file size over previous GPUs.

So, perhaps the combination of a large increase in memory (with attendant redundancy) and fine-grained ALU redundancy caused the effect.

Jawed

Sxotty

18-May-2007, 16:38

The actual voucher only came with some laptops equipped with 9600s and retail boxed 9800/9600XTs (the latter of which were rare). Not every card. You're also implying that a significant number of people went out and spent hundreds of dollars simply to have a voucher for a game that wasnt due out for quite some time.

That is simply not true. I received two vouchers one without those purchases and one with a 9600 of some type.

18-May-2007, 18:55

I have no problem telling you now though.. but it'll be based on BS. I first need to bring RB6:LV and my 2900 together.. my brother has my copy of rb6.. :(

Well I was just making a play on words to try to be funny. :P

And I would rather have your BS than others, so we will be waiting to see what you have for us on Monday! :)

neliz

18-May-2007, 23:38

Well I was just making a play on words to try to be funny. :P

keep trying! ;)

And I would rather have your BS than others, so we will be waiting to see what you have for us on Monday! :)

Hope to get it finished by then.

Twinkie

19-May-2007, 00:03

This sounds insane... but 93 page review of the R600!

http://www.pcpop.com/doc/0/194/194233.shtml

Its in chinese however :wink:

Google translated version here.
http://translate.google.com/translate?u=http%3A%2F%2Fwww.pcpop.com%2Fdoc%2F0%2 F194%2F194233.shtml&langpair=zh%7Cen&hl=en&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools

neliz

19-May-2007, 00:37

Interesting in that review that in some cases the scores for both the radeon and the geforce are higher in vista than in xp when the resolution increases.

Amazing, that review also shows the XT handily beating the GTS in Far Cry!
http://translate.google.com/translate?u=http%3A%2F%2Fwww.pcpop.com%2Fdoc%2F0%2 F194%2F194233.shtml&langpair=zh%7Cen&hl=en&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools

Twinkie

19-May-2007, 00:45

Could it be because of the bandwidth advantage that the R600 has? Id say its a possibility since the res is at 1920x1200 and 2560x1600. Interesting results especially the DX10 results.

neliz

19-May-2007, 00:53

Could it be because of the bandwidth advantage that the R600 has? Id say its a possibility since the res is at 1920x1200 and 2560x1600. Interesting results especially the DX10 results.

The thing is that Far Cry so far has shown the XT to sometimes be outperformed by a 1950Pro, this new benchmark completely flips the tables.

The DX10 tests were interesting too, but I'm even more amazed about the performance of the Vista drivers in some situations, cases where it's 15% faster should be a testimony to what can be achieved with vista driver optimization.

trinibwoy

19-May-2007, 01:43

Hkepc apparently has some 8.38 numbers - http://www.hkepc.com/bbs/hwdb.php?tid=790821&tp=AMD-R600-HIS&rid=790821

Rangers

19-May-2007, 05:52

http://www.firingsquad.com/hardware/lost_planet_demo_directx_10_performance/

FS does some lost planet benches. They found LP runs slower in DX10 than DX9. Unfortunatly they only tested DX10 on 8800 cards. However, they did remark that they were pretty impressed with X2900, which in this case was able to consistently place between 8800GTS and GTX. Even though ATI informed them drivers for LP where not optimized. So it's good to see 2900 pull ahead of GTS in a more modern shader heavy game.

Twinkie

19-May-2007, 06:51

Hkepc apparently has some 8.38 numbers - http://www.hkepc.com/bbs/hwdb.php?tid=790821&tp=AMD-R600-HIS&rid=790821

I did notice one thing

NVIDIA Forceware Driver 158.48 :wink:

vertex_shader

19-May-2007, 11:32

After Valve canceled the black box (http://pc.ign.com/articles/790/790013p1.html) , hd2900xt valve black box voucher still remain as black box, or goes up to orange box? (orange box has 2 more games, but this 2 released already, hl2 and hl2 episode1).

neliz

19-May-2007, 12:08

your code no the voucher unlocks a certain range of games. When valve decides to unlcok hl2 en ep1 as well it will happen. so unless they tell you so, NO

trinibwoy

19-May-2007, 12:26

I did notice one thing

NVIDIA Forceware Driver 158.48 :wink:

Don't know if that implies anything. Could just be a typo.

Geo

20-May-2007, 00:03

I think there's a 158.48 running around, so maybe not.

neliz

20-May-2007, 00:04

I think there's a 158.48 runnig around, so maybe not.

THG reviewed with it. (vista and gtx only, xp was done on the 22)
http://www.tomshardware.com/2007/05/14/r600_finally_dx10_hardware_from_ati/page19.html
http://www.thg.ru/graphic/amd_radeon_hd_2900_xt/amd_radeon_hd_2900_xt-07.html

fallguy

20-May-2007, 01:57

.48's are available. Just because you dont have them... doesnt mean they dont exist. :o

Twinkie

20-May-2007, 09:45

.48's are available. Just because you dont have them... doesnt mean they dont exist. :o

Lies!

Ok i didn't realise, but what i find hilarious about this is that both camps are releasing drivers not every month.. not every week, but everyday it seem.

:lol:

PeterAce

20-May-2007, 14:22

You mean this one?
http://www.digit-life.com/articles2/video/r600-part2.html
http://www.digit-life.com/articles2/video/r600-part1.html

Thanks.

Ugh..... Nooooo... just noticed that benchmark only the 8800 GTS for all G80 results!

I wanted both the GTX and GTS to be in the review (both very useful data points!).

banksie

21-May-2007, 05:34

neliz

21-May-2007, 07:47

SugarCoat

21-May-2007, 09:09

...I must admit after reading all the reviews complaining bitterly about the noise under load much to my relief the card is actually quieter than the X1800XL it has replaced. Now I don't have a db meter handy but I had thought this was supposed to be the loudest thing thus far released, right?

Maybe there is variation in build quality going on.

no, actually the initial X1800XLs were some of the loudest released, Xbit even stated just that ;). So in your case, it is indeed a case of going from worse to better.

Even though it may not be as noisy as some other graphics cards, such as Radeon X1800 XL with the first version of the cooler, it is definitely much noisier than the main rival

ChrisRay

21-May-2007, 09:41

Whats a good discussion about GPU DBs without remembering the FX 5800U? :P

http://video.google.com/videoplay?docid=-6044284326929549145&q=FX+cooler+Nvidia

neliz

21-May-2007, 09:52

Whats a good discussion about GPU DBs without remembering the FX 5800U? :P

http://video.google.com/videoplay?docid=-6044284326929549145&q=FX+cooler+Nvidia

THG's video was better :0

http://video.google.com/videoplay?docid=-828507850880681606&q=FX5800

ChrisRay

21-May-2007, 10:01

Yes but Nvidia made the movie I posted. Anyway! Back on topic!

Twinkie

24-May-2007, 04:34

With these 8.38 drivers or what ever its called, i read somewhere that ATi replaced AAA with EATM which resulted in performance increases (with IQ being degraded of course).

Is this true? what is going on with all these review sites conflicting with one and another.

I saw one review where the XT suddenly beat the GTX across the board while consuming less power than the GTX :lol:. There was one with AA supposedly enabled on the 2900XT and it beat the GTX.

All im saying is that can anyone cough*beyond3d*cough draw the line on this matter?

tEd

24-May-2007, 06:15

Unknown Soldier

24-May-2007, 06:35

I'll tell you monday

So neliz, is the HD2900XT bugged in HDR?

US

Unknown Soldier

24-May-2007, 06:38

I did notice one thing

NVIDIA Forceware Driver 158.48 :wink:

Erm.. last time I looked the 158.48 were official Vista beta drivers.

NVIDIA - Forceware 158.48 (Vista - 8800GTX)
www.nvidia.com/object/winvista_x64_158.42.html
www.nvidia.com/object/winvista_x86_158.42.html

Release Date: May 14, 2007

US

neliz

24-May-2007, 07:56

So neliz, is the HD2900XT bugged in HDR?

US

Plugging it in my machine now, I've been away two days (yeah, work sucks)

8.361 drivers definitely have a bug, it's not noticeable at first but when looking into a light source, or close to it shadows will flicker.
it's noticeable in the first level. you duck behind a container to shoot one guy, everything is still okay. walk to the little town square with the small shootout, still okay, walk in the alley to the left to encircle the two soldiers, still okay, now if you look back to the town square from the point where you exit the alley you will see part of the car shadows flicker, look away from the square and the shadows return to normal.

going to test with 8.37.4.3 now (asus 8.38 doesn't accept the card.0

Unknown Soldier

24-May-2007, 09:33

Plugging it in my machine now, I've been away two days (yeah, work sucks)

8.361 drivers definitely have a bug, it's not noticeable at first but when looking into a light source, or close to it shadows will flicker.
it's noticeable in the first level. you duck behind a container to shoot one guy, everything is still okay. walk to the little town square with the small shootout, still okay, walk in the alley to the left to encircle the two soldiers, still okay, now if you look back to the town square from the point where you exit the alley you will see part of the car shadows flicker, look away from the square and the shadows return to normal.

going to test with 8.37.4.3 now (asus 8.38 doesn't accept the card.0

:(

Hope it's just a driver problem.

neliz

24-May-2007, 11:06

:(

Hope it's just a driver problem.

Definitely, tried it with 8.37.4.3 and the problems are almost gone. I do see the shadows inside the 'copter but I'm not sure if they should be like that. and have only seen the shadow bug show up once, and that was when a body dropped halfway through a barrel.

Geeforcer

24-May-2007, 16:13

So the problem Anand showed in his screenshots is (largly) gone now?

neliz

24-May-2007, 16:23

So the problem Anand showed in his screenshots is (largly) gone now?

well.. yeah.. there are probably some spots where it will show up but the big area's with 8.361 are gone.

swaaye

24-May-2007, 19:01

HD 2900 drivers remind me of Rage Pro drivers. Hopefully no "Turbo" edition arrives. :wink:

So has anyone figured out how long the drivers have been in development? Is there any way to tell? If they've been toiling for 6+ months on these, that's not a good sign....

trinibwoy

25-May-2007, 04:58

http://www.xbitlabs.com/articles/video/display/radeon-hd-2900-games.html

Karma

25-May-2007, 07:46

Yeah I like that last review. It really shows how good (and how bad) the R600 can be. Drivers are the key, man! I really think that the R600 has the ability to be better in almost all games than the GTX, but time will tell whether the driver effort is put into it.

neliz

25-May-2007, 08:51

http://www.xbitlabs.com/articles/video/display/radeon-hd-2900-games.html

This review has a far better 2900 than I've seen in the first batch.
Remember how Far Cry toasted the 2900 below 1900Pro level? The 2900 is suddenly almost three times faster than a 1950XTX! both in minimum and average frames. the pier level shows hardly any performance drop-off while out pacing the GTX as soon as resolutions are higher than 1024x768. The Research demo still shows a healthy lead.

Battlefield 2142 is suddenly equal

The rest of the reviews still show it getting a resounding beating, but improvements seem to be spectacular so far.

Karma

25-May-2007, 09:04

This review has a far better 2900 than I've seen in the first batch.
Remember how Far Cry toasted the 2900 below 1900Pro level? The 2900 is suddenly almost three times faster than a 1950XTX! both in minimum and average frames. the pier level shows hardly any performance drop-off while out pacing the GTX as soon as resolutions are higher than 1024x768. The Research demo still shows a healthy lead.

Battlefield 2142 is suddenly equal

The rest of the reviews still show it getting a resounding beating, but improvements seem to be spectacular so far.

Yeah. Check out the difference between HardOCP's numbers with a previous driver (although I can't find the version they used) and Xbit's numbers for B2142: the minimum frame rate at 1600x1200 4xAA & 16xAF apparently went from 29FPS to 66FPS!

http://www.hardocp.com/article.html?art=MTM0MSwxMCwsaGVudGh1c2lhc3Q=

SC: DA has the R600 right with the GTX, too. Amazing.

http://xbitlabs.com/articles/video/display/radeon-hd-2900-games_9.html#sect3

But seriously, they still need to polish up the drivers for other games: NWN2, anyone? ;-)

neliz

25-May-2007, 09:16

But seriously, they still need to polish up the drivers for other games: NWN2, anyone? ;-)

Supreme Commander still has it lagging behind the XTX too and some synthetic benches show this as well. after all . there is still hope.

Pete

25-May-2007, 09:31

Heh, Bioware's games have never been kind to ATI.

I don't mean to keep banging the R600 drum to the exclusion of G80 (which is obviously a superb piece of engineering and seemingly superb example of execution, at the very least relatively), but how can one not conclude drivers are holding R600 back when it's got higher clocks, way more bandwidth, and more of everything than R580+ yet only benches even with it on so many games? It's veird!

Karma

25-May-2007, 09:38

Well, it's only been out for a week and a half, so such large improvements in the drivers in such short a time are good to see. Hopefully by the time the official 7.6's are released it'll all be fixed up.

btw, in COH you can see that the average FPS is on par with the GTX, but the minimum is about half of the 1950XTX.

Eet is very vierd indeed!!

neliz

25-May-2007, 09:41

It's veird!

That's the opposite of Svelte?

As I said, Far Cry is obviously a good indication of where drivers where and where they are going. now we can't have 2.5 or 3 times the performance in every scenario but there should be some consistency on the higher "minimum" frames etc.

as Karma says, in Coh for instance, I've tested it too and indeed there is some incredible slow-down noticeable when the nozzle flashes of the AA guns are on screen, it really drops to teens-twenties and after that resuming it's normal high fps rate.

Unknown Soldier

25-May-2007, 09:50

You know, for a card that AMD boasts to have been around since January(some of you might argue but AMD delayed the release in March to release it with it's family, so they definitely had the card ready in January), it's shocking that they didn't have proper drivers available for launch. I'm gonna have to rap AMD over the knuckles for this one.

Other than that, it's starting to look good by that X-bit review.

Hopefully, they can sort out other problems.

Slower than the X1950XTX using HDR+16xAF in Stalker at all Res. (http://www.xbitlabs.com/articles/video/display/radeon-hd-2900-games_8.html) .. tsk tsk

US

Galduta

25-May-2007, 10:12

http://www.xbitlabs.com/articles/video/display/radeon-hd-2900-games_11.html#sect0

The Elder Scrolls IV: Oblivion

Like some of the games above, The Elder Scrolls IV: Oblivion doesn’t support FSAA and HDR simultaneously, so we tested it in HDR mode only translation : Like in all games , AMD 2900 doesn’t support FSAA without a one 50 % perfomance drop, so we tested it in HDR mode only :roll::roll:

Karma

25-May-2007, 10:17

btw, I know this question is slightly off topic, but does a person really need to enable AA when playing at 1920x1200?

**Austin Powers impression:**

I mean, honestly!

**end impression**

neliz

25-May-2007, 10:20

btw, I know this question is slightly off topic, but does a person really need to enable AA when playing at 1920x1200?

**Austin Powers impression:**

I mean, honestly!

**end impression**

I know I just enable 2x at 1600x1200 that's honestly enough in my opinion though situations vary with the games you play, you don't really need AA in something like Medieval II but it's nice to see it in BF2.

Skinner

25-May-2007, 12:08

The fact that it's performing very good in certain games, gives hope for better drivers, and the texturering power wouldn't be the limiting factor then? Because Far Cry and COJ have plenty of high res textures?

Not bad for €340.

trinibwoy

25-May-2007, 13:31

Performance is looking up but so is the price - it's been heading north since launch. Demand must be relatively strong because there don't seem to be any supply issues.

neliz

25-May-2007, 13:41

Performance is looking up but so is the price - it's been heading north since launch. Demand must be relatively strong because there don't seem to be any supply issues.

Yep, my Sapphire card's price rose €2 after launch but I've seen some brands rise €20 (Club3D) and availability not sufficient anywhere. The next batch of cards are supposed to arrive at reatailers on the 31th, next week.. maybe with some whql 7.5 drivers on the site..

Skinner

25-May-2007, 14:16

Yep, my Sapphire card's price rose €2 after launch but I've seen some brands rise €20 (Club3D) and availability not sufficient anywhere. The next batch of cards are supposed to arrive at reatailers on the 31th, next week.. maybe with some whql 7.5 drivers on the site..

Neliz, look here if you want:

http://gathering.tweakers.net/forum/list_messages/1221404/2

I have done some tests with the 8800GTX (600/1900 mhz)

Can you compare it with your X2900XT?

It's that I spend to much money on going out lately, otherwise I would have bought one X2900XT to play with it against my 8800GTX.

Skinner

25-May-2007, 14:25

nm.

neliz

25-May-2007, 14:30

will look into it tonight..

AlexV

25-May-2007, 16:06

I just wanted to say I hate you ATi, for fucking up something in your latest 8.38 RC candidate:the friggin 1440x900 resolution. Why in god`s name would you screw something like that up?It says out of range out of range out of range with the HD2900XT, but it was ok with the X1900s prior to them?So my LCD`s resolution suddenly becomes out of range?WTF?WHY must I stare at a scaled desktop because Chuck made a fuck up?

But nice card, BTW...definitely less noisy than the x1900s, at least during boot-up...I`m like a kid in a candy store with my new toy, if I like it, I`ll get a sister for her...and a new PSU, since my lowly 700W won`t cut it probably. But giddy, otherwise.

FIX MY RESOLUTION!:D

Dave Baumann

25-May-2007, 16:20

Id' suggest submitting a bug report.

AlexV

25-May-2007, 16:39

Actually, I was hoping you`d tell me I`m a retard and I did something horribly wrong, enabling me to fix the issue instantly, instead of through a bug report...oh well:(

AlexV

25-May-2007, 17:18

Yea, I was a moron, with the latest oficially available ones from ATIs site it's ok. So the groovy part stays:)

BTW, why doesn`t the Lost Planet DX10 demo work with the 2900?It gives some stupid error upon starting up...how did the benchmarkers use it?

vertex_shader

25-May-2007, 17:29

Asus EAH2900XT AquaTank (http://www.tweaktown.com/news/7574/index.html)

ERK

25-May-2007, 19:08

Haven't put my 2c in yet, but just wanted to say that based on the shots I've seen so far, the CFAA looks really really really good, especially the Oblivion one here (http://www.techreport.com/reviews/2007q2/radeon-hd-2900xt/index.x?pg=6).

ERK

neliz

25-May-2007, 20:55

So neliz, is the HD2900XT bugged in HDR?

US

8.38RC7 has no more shadow bugs, but also no AA.. at least.. not in this screenshot:
http://home.wanadoo.nl/deachterberg/R6Rappel.png

that's 1600x1200/16hqaf/2xaa+NT HDR+shadows high.

sonix666

25-May-2007, 23:54

http://www.xbitlabs.com/articles/video/display/radeon-hd-2900-games_11.html#sect0

translation : Like in all games , AMD 2900 doesn’t support FSAA whitout a one 50 % perfomance drop, so we tested it in HDR mode only :roll::roll:
Get your facts straight. Oblivion itself does NOT support HDR and FSAA at the same time. Believe me, I play the game. However, ATi has managed to get FSAA working with HDR for Oblivion by forcing FSAA in the driver panel and set Oblivion to HDR mode. But it seems, not everyone knows that yet.

SugarCoat

26-May-2007, 00:01

Asus EAH2900XT AquaTank (http://www.tweaktown.com/news/7574/index.html)

This is all they ever do, create some exotic model that never really gets much retail exposure or doesnt come out until months after a product launches, and does so with a minor clock change and thats it (see sapphire TOXIC/HIS IceQ). I really wish they'd wake up and start really bumping clocks, i think we could see some nice performance out of something 850/1800 or 850/2000 which is what nVidia's partners have been doing with great success for years now.

Get your facts straight. Oblivion itself does NOT support HDR and FSAA at the same time. Believe me, I play the game. However, ATi has managed to get FSAA working with HDR for Oblivion by forcing FSAA in the driver panel and set Oblivion to HDR mode. But it seems, not everyone knows that yet.

think its worth mentioning ATi isnt the only one who can do HDR+AA in Oblivion. That said i dont really understand the reviewers comments myself.

Sobek

26-May-2007, 00:18

However, ATi has managed to get FSAA working with HDR for Oblivion by forcing FSAA in the driver panel and set Oblivion to HDR mode. But it seems, not everyone knows that yet.

Yep, that's exactly what they did to get it working! :lol:

That said i dont really understand the reviewers comments myself.

Perhaps they just couldn't get it to work, so for lack of knowing any better assumed it was broken. There have been times where for no reason at all AA just refuses to apply in certain games unless I disable/enable random options in ATT until it just magically works.

Either that or they are just dense!

SugarCoat

26-May-2007, 00:43

Perhaps they just couldn't get it to work, so for lack of knowing any better assumed it was broken. There have been times where for no reason at all AA just refuses to apply in certain games unless I disable/enable random options in ATT until it just magically works.

Either that or they are just dense!

From the way they basically said "Oblivion doesnt support AA+HDR" let alone naming a driver problem specific to the X29000, im betting they just didnt know. Although its not really untrue either. Oblivion itself really doesnt do AA+HDR, maybe they didnt see any sense in doing the driver hack version since it might compromise the unbias of the test.

Pete

26-May-2007, 03:16

Xbit knows (http://www.xbitlabs.com/articles/video/display/gf8800-games_18.html#sect0) modern cards can push AA with HDR in Oblivion.

neliz, you Dutch hater, I was just summoning my inner Goldmember (http://www.imdb.com/title/tt0295178/quotes) (see quote #3). :razz:

wishiknew

26-May-2007, 04:00

The highs and lows of the r600 all in that earlier xbit review. Like to think the situation will look different by the time the r650s nda expires.

neliz

26-May-2007, 13:18

neliz, you Dutch hater, I was just summoning my inner Goldmember (http://www.imdb.com/title/tt0295178/quotes) (see quote #3). :razz:

;) :)

AlexV

26-May-2007, 14:36

Hmm, I like this card...mostly. I can`t use the 8.38 driver though, it fooks something with 1440x900, I`ve isolated it and contacted ATi. Sigh. I`m pondering wheter or not to get another one for Crossfire...I`d have to get a new PSU too, and a decent one with the required juice would cost me something quite near what the second 2900 would cost me...hmm, decisions, decisions.

Overall, coming from my CFed X1900s, it feels like a step forward. Yes, the cooler is rather annoying when it spins-up when in desktop mode and it`s also noisy when in 3D BUT, it doesn`t(or at least I haven`t noticed it as much) have the absolutely mind destroying up/down/up/down cadence of the x1900s coolers. And it`s far more quiet upon start-up. The performance also feels like a step forward...oddly enough, Stalker, the game that should suck in terms of performance feels spiffier...so does NWN2 with 8X AA activated and everything as high as is allowed, though I haven`t yet went to an outside area(which are the main performance hogs in that game). I`ve also toyed with Two Worlds(since someone was asking in another thread), and it seemed to perform better.

Still need time to have a full opinion. but until now I like it quite a bit. I`d also need a spiffier processor, but I`m waiting for the Core 2 Quads with 1333 FSB and lower prices:)

BRiT

26-May-2007, 18:53

AlexV

26-May-2007, 20:55

Which I quite obviously am not, sadly:) But yes, that Silverstone is interesting, I`ve taken notice of it, but the catch is it costs 352$ here:|

pakotlar

26-May-2007, 22:28

Actually, I was hoping you`d tell me I`m a retard and I did something horribly wrong, enabling me to fix the issue instantly, instead of through a bug report...oh well:(

use powerstrip and add the resolution. restart and ur all set.

AlexV

26-May-2007, 23:03

Powerstrip crashes for some odd reason on my Vista X64 install, tried that already...the new Cats should be released anyday now, so we`ll see either way. Since I was checking out PSUs i found this:http://www.seventeam.com.tw/Backend/UpLoadPic/ST-1000E-AD.pdf, which seems interesting and is actually affordable around here. The catch is that I`ve never heard about em`, save for the fact that JohnnyGuru has one that he`s been testing for...umm...a long long time:) Any oppinions?

swaaye

26-May-2007, 23:12

A feller just posted over on Tech Report that R600 taped out only about 6 months after C1 "Xenos". This was a comment about as to why the UVD accel was left out.

The reason they left it off is because R600 is a 2-year old GPU. ATi worked on it at the same time as the Xbox 360 GPU. At that time, they were probably going for pure performance win instead of a feature win.

I was in discussion with MS at that time about the manufacturability of ATi's design. I worked for both ATi and MS's mutual contract manufacturer. R600 was about 6 months behind the Xbox360 C1 gpu in terms of tape-out date.
Interesting to anyone? Telling of how closely they are related, perhaps.

AlexV

27-May-2007, 00:08

If it is reliable...the relationship is clearly there, as both Xenos and R600 are offspring of the mythical R400 that caused the mother of all shuffling at ATi. Dunno if what that guy says is true though.

As something to add to a review thread:the tent mode are horrid. At least IMHO. Perhaps I`m more sensitive to blurring, but I find the blur they bring unbearable-I hope they either tweak the way they do the sample weighing ATM, or go another route.We`ll have to see how the edge-detect AA turns out. 8X is fairly nice though. A very fun game that could provide interesting benchmarking is Marvel:Ultimate Alliance. It has niceish graphics and used to run like crap on x1900s-it`s great on the 2900.