HD 2900XT & Driver 8.38

All you guys slamming "EastCoastHandle" I'd just like you to know that I know of him from another site and if he's the same guy I believe him to be he's not just making stuff up. I know I'm nobody here and anyone could copy a username from another site, but from what he's said here I think he's the same guy from the TPU forums(Great site,but the forums there are run by Kids and suck IMHO).
 
I really hope it does perform better than a GTS because it costs more.

On another note drivers can only do so much...it is confirmed that r600 has an inferior architecture compared to NVidia this round...the only thing that can save them is MASSIVELY high clocks.

IF this was 6 months ago I would say ATi did a so so job...but the fact that ATi delayed so many times (in attempt to get these massive clocks, which they DID NOT achieve) there is a right to be extremely disappointed as ATi could have very likely released this product 6 months ago (unless their original offering was even slower clock wise :eek: )
Your post above assumes a great deal of inside knowledge on the R600's R&D historical record. The only thing I agree is that driver improvements for any architecture do have their limits. That in and of itself does not make any specific architecture inferior. It's true for every product.

The objective of attaining the industry's highest clock speed is akin to a holy grail of sorts in IT, whether it makes an overall difference or not. IOW, it's a given that every manufacturer strives to achieve the highest clocks whether they're in the CPU market, graphics market, memory market, etc. Everyone strives for high clocks and it's only to be expected ATi would do the same. Considering the numerous reports of very good OC with the R600, clockspeed alone could not have hampered the release of the cards.
 
Let me post some facts for you:
THE DRIVERS THEY USED TO TEST FOR NVIDIA DO NOT, I REPEAT, DO NOT EXIST.

THE DRIVERS THEY USED TO TEST WITH ATI, DO NOT, AS OF RIGHT NOW, DO NOT EXIST.

Now, try to prove my "FACTS" wrong with your "FACTS".


You know I posted a thread here with a link to those drivers, Its only a few threads down from here. Many are using them and are happy with them
 
tom20cruisejpgsl4.jpg


ShiningHolden said:
 
Interesting numbers. Shame there aren't other 8.3x numbers for easy comparison.

In any case, it seems as if, on the surface anyway, that AMD is making good ground here, and so quickly as to almost have me concerned.

On an off-topic note, has anyone else noticed how cleanly and bitterly divided the community is since the advent of the 2900? I think this thread would make most children cry (and possibly some grown men as well).
 
phide, check out that PCPOP review as well. I hope the benchmarks get replicated again with newer drivers when they come out. I always find it disappointing how sometimes a rushed launch is crushed so badly by things that are not working properly (through driver problems), creating tremendously horrible PR for the company in question.
 
here is a translation(provided by google) of the conclusion.
ATI Radeon 2900XT though not HD efficiency, replace the GeForce 8800 Ultra graphics performance to become emperor, but the $ 399 price beautiful, GeForce 8800GTS not only meet, and efficiency were higher than 5 ~ 30%, and those for GTX. cost performance is very satisfactory.
I like the last line. "Cost performance is very satisfactory."
 
Looks like bullshit to me... Im sure the new drivers will help but come on lets be real thats too much improvement for drivers alone to make.
 
Looks like bullshit to me... Im sure the new drivers will help but come on lets be real thats too much improvement for drivers alone to make.

Given that ATI's architecture (superscalar) requires craploads of compiler tweaking, it isn't beyond the realms of possible.
 
I am inclined to believe HKEPC cause they are pretty trustworthy. Also it is a known fact that these drivers are floating around review sites and some of the fellas @ xtremesystems are already playing around with them. I think that if these drivers improve the performance by a good little bit I will be getting one of these for my lan machine.
 
Given that ATI's architecture (superscalar) requires craploads of compiler tweaking, it isn't beyond the realms of possible.
Agreed. It has happened before. Although the R600 might not beat the 8800GTX in all games with a new driver release, it certainly can approach it and best the performance of the GTS. AMD/ATI would be very happy if it can achieve that much.
 
Looks like bullshit to me... Im sure the new drivers will help but come on lets be real thats too much improvement for drivers alone to make.

How are you to make this judgment? I've worked in the software industry, and have seen performance gains of 100% or greater to specific runtime procedures when doing things in an optimized fashion versus not.

I'll give you an example of using OpenGL. If you use it to generate a terrain based on heightmaps using the GL_TRIANGLE_STRIP, without using display lists can give you ~1/3 the FPS of using display lists. See here:

http://www.lighthouse3d.com/opengl/displaylists/

As such, I think your claim of "thats too much improvement" is bullshit. Where does this claim come from or is it just "gut feeling"?
 
LOL, as I posted before, I don't care about nvidia drivers. As for the 8.38, it's not my problem to find them for you. :D As I stated before, wait until the new Cat 7.5 release and see then. No need to get so upset because I disagree with your point of view:
-you cannot find drivers therefore benchmarks are fake.

And, lets make this clear, this is not an issue that I posted the results using new drivers it's solely based on why I don't see these benchmarks as fake in absence of evidence to the contrary. Unless there is some solid proof (other then you cannot find the drivers) I reserve my opinion. :cool:

You know if they made up the results for Nvidia, they almost certainly made up the results for ATI. Duh?
 
How are you to make this judgment? I've worked in the software industry, and have seen performance gains of 100% or greater to specific runtime procedures when doing things in an optimized fashion versus not.

I'll give you an example of using OpenGL. If you use it to generate a terrain based on heightmaps using the GL_TRIANGLE_STRIP, without using display lists can give you ~1/3 the FPS of using display lists. See here:

http://www.lighthouse3d.com/opengl/displaylists/

As such, I think your claim of "thats too much improvement" is bullshit. Where does this claim come from or is it just "gut feeling"?

Very true optimized drivers can help alot but a huge increase all across on all the benches i say bullshit still. I have seen drivers improve performance in a game or 2 but you cant tell me in a week they optimized these drivers for all of the games out there.
 
Very true optimized drivers can help alot but a huge increase all across on all the benches i say bullshit still. I have seen drivers improve performance in a game or 2 but you cant tell me in a week they optimized these drivers for all of the games out there.

of course not , that takes time


regards
 
Very true optimized drivers can help alot but a huge increase all across on all the benches i say bullshit still. I have seen drivers improve performance in a game or 2 but you cant tell me in a week they optimized these drivers for all of the games out there.

I see your point. I cannot fully refute your claim that it's BS, so I will wait on more reviews with later drivers before I confirm this, however, my feeling is that HKEPC and PCPOP are not liars. If you believe they have lied on the review, then that is a MAJOR issue which deserves plenty of attention (and if indeed they did lie, they will be getting this major attention when people realize their cards aren't working as they should), so give it another week to tell. Also, I do not see what these review sites (that seem to have done quite a bit of work) would gain from lying, as once your reputation is down the pooper, it's a hard struggle back up.
 
Let's just, wait until the drivers are released?

If they are, the card will still draw so much power and cause so much heat.

But, if it makes it faster in some games here and there, and boosts some benchmarks, awesome.

If it makes CrossFire good too, awesome!

It isn't worth arguing over. :p

Drivers do have the ability to become optimized and increase FPS here and some scores there. :)

I honestly hope they do. That would put one good card on the market. If not, it is an option for anyone.
 
I hope everyone does not believe all this nonsense about heat and noise. My card gets 79'C at load, in fact my old x800 got to 82'C at load. also, its no louder than a 1800 or 1950. Everyone here who keeps say loud, heat whaaaaaaaa, doesn't even have to card.:rolleyes: I do. Its fan has a much deeper tone, its not bothersome at all. Also, during 2d its almost silent. another thing, yeah it draws the most watts, but this has to be the FIRST time ever people have whined over performance per watt.:confused: The 2900 is like a Ferrari maranello, the 8800 GTS like a corvette. Does the ferrari suck because its a fuel/power hog with its V12 engine, compared to the corvettes' V8? The corvette has about the same 0-60, but who would pick a corvette over a Ferrari?
 


How are they getting a 9144 in 3DMark default for a 640MB GTS with a Core 2 Extreme? I have an AMD 4000+ at 2822 and a GTS and I get 9206. On a bad day I get over 9000, and I don't think anyone wants to try to convince me that a Core 2 Extreme is slower than my AMD 4000+ @ 2.8Ghz. That site either A) has phony scores B) has no idea what they are doing C) is trying to sell video cards.

I'm thinking perhaps all of the above.

Here is my result from this evening using the 158.22's. If my little processor can do that with no "special preparation" or magical means, I'm thinking their results for the Geforce GTS/Core 2 Extreme are....off.
 
How are they getting a 9144 in 3DMark default for a 640MB GTS with a Core 2 Extreme? I have an AMD 4000+ at 2822 and a GTS and I get 9206. On a bad day I get over 9000, and I don't think anyone wants to try to convince me that a Core 2 Extreme is slower than my AMD 4000+ @ 2.8Ghz. That site either A) has phony scores B) has no idea what they are doing C) is trying to sell video cards.

I'm thinking perhaps all of the above.

What's your SM3 score for 8800 GTS 640? Is it around 3800? I'm just trying to take the CPU out of the equation here, since 3DMark06 has done a real SHIT job by incorporating the CPU score into the overall mark, so at least, please tell us if your non-overclocked 8800 GTS 640 is consistent in score with their non-overclocked 8800 GTS 640.
 
Here is a graphical representation. You can see how they got their results. They have the GTS underclocked to 500/1200! Real fair.

Anyway, here are mine vs. theirs:

resultsmethemwo4.jpg

moreresultslh2.jpg
 
And that, friends, is how to throw a benchmark. Underclock the card you are trying to beat, note it, but count on the fact that people aren't paying attention. I wasn't, but I knew the score was way too low for a Core 2 Extreme/GTS, because I bench my own box when I use a new set of drivers. It was only after I pulled up my own score from tonight and looked at it that I realized something was wrong - I had just happened to do so because I was going to play a HL2 mod and wanted to use new drivers so I benched my box.

I likely wouldn't have caught it had I not just gotten through running 3D06 myself. After I upped that to Imageshack, I realized they had underclocked the card because it's right there. It just looks like the GTS is performing badly because it is underclocked on both the core (15Mhz) and a whopping 400Mhz underclock on the memory.
 
u do realize there is no major discrepancy between your scores and theirs, right? Nothing that would suggest some sort of manipulation. Your overall scores are nearly IDENTICAL, and there is a small difference between SM 2.0 and 3.0. Whats the problem again? Also, the XT beats the GTS in 3dmark06 in EVERY site that did reviews.:p
 
Does that prove anything? No, because my processor is less powerful than the benched box. Usually Extremes get at least 2800, so factor that in.

What it does do, though, is show that their nVidia scores are fudged, and as Obi_Kwiet pointed out earlier, if they are playing around with the NVidia numbers, they are playing around with the ATI numbers too.

Why? To sell video cards. If it convinces even 1 person to go out and buy a video card, it served a purpose for them.

I wouldn't have even noticed had I not benched my box earlier.

Now I'm going to play some modded HL2. It is fast enough for me, and the graphics look great at 8XQ AA / 16XAF/ SS TA / 1680x1050. The best part is that I've had it for over 3 months, I paid around $365 for it even back then, and it still rules. :D
 
one more.

Beats GTS:

http://www.legitreviews.com/article/503/12/

I suppose all these sites are in a consipiracy against the 8800 GTS, LOL :rolleyes:.

As the site says after those graphs...

In 3DMark 2006 our 2900XT turns a great performance, the biggest part of that coming from the SM 3.0 score. This is just another example of why can't base a purchase decision completely on 3DMark.

It was rumored/known the 2900XT ran 3dmark with a high score before any of these reviews were posted. This is nothing new, and it is also nothing meaningful ;)
 
one more.

Beats GTS:

http://www.legitreviews.com/article/503/12/

I suppose all these sites are in a consipiracy against the 8800 GTS, LOL :rolleyes:.

To be honest, all the benchmarks, time demos and speed in the world doesn't matter a whole lot if it looks terrible. NVidia fans found that out the hard way with the FX series. I guess it is the ATI fans turn this time. Just remember - neither company gives a damn about you or your loyalty - they just want to sell video cards. I'm not affiliated with anyone, and I have nothing to prove or to sell you. Look in my signature and decide if I am "anti-ATI/AMD".

The 2900XT has to stand on its own merits, and fudged results against an underclocked card is not the way to do it. We will see in the coming months whether those who rejected it were all "stark raving nvidiots" or genuinely disappointed potential buyers that realized it was way too little way too late. Now I am really off to go play my game :D.
 
Caffeinated: Their clock speeds are Core/Shaders, not Core/Mem, so they have NOT underclocked their 8800 GTS. This was already mentioned -- please read through the whole thread. In case you do not believe me, you can read it here:
http://www.hardforum.com/showthread.php?t=1192693

This page also shows the clock speeds at which they tested:
http://www.pcpop.com/doc/0/194/194233_70.shtml

[FONT=宋体]
8800GTS 640MB(500/1200/1600MHz)
8800GTS 320MB(500/1200/1600MHz)
HD 2900XT 512MB(740/1650MHz)
X1950XTX 512MB(650/2000MHz)
​
[/FONT]
Edited again: I need some info on your 8800 GTS 3DMark06 SM2.0 and SM3.0 score, Caffeinated. You told us that their scores were not very good given the processor. Well, lets take the processor out of the equation and use the 3D scores only. They get approximately 3800 SM3.0. If yours is significantly higher or lower, then I agree, there is a problem. Please get back to me with this.

Edited again: It seems like I'm a bad reader, and I missed the part on Caffeinated's results -- my apologies. They look fine, however.
 
lol, all those sites did NOT underclock the GTS or "fudge" the results, moron. Nice of you to igonre all that, cuz you just got shown how wrong you are. And we were not talking about anything else except 3dmark 06, which you are hung up on.

You just got told. Accept it, and move on. Fact is, the XT owns the GTS in 3dmakr06, deal with it.



and @ Sr7, I already mentioned 3dmark06 doesn't mean anything. It does to caffeine however. Im making fun of him in case no one noticed.
 
yes, yes. Improvements. I like that. :D

But, all I want to know is:

Can I still power my mini-fridge or is that still a no-go if I overclock this beast? :p

I still say, a refresh/die shrink and this card will be worth it's wait in gold. I eagerly await this fall.
 
Caffeinated,

Your SM 2.0 score is 4123 and their score is 4017, I don't see that is a big difference when compaired to 4613, even 4123 is lower. Your SM 3.0 score is 4146 and their score is 3841, the difference is quite big but not as big as the difference between 5115 and 4146.

It seems to me that some 8800 owners especially the GTX owners are really pissed when the 2900XT beats their card even in just one benchmark, the 3DMark06, they are going to be more pissed if the new driver will improve the 2900XT's performance in games too :rolleyes: My advice to the 8800 owners: Don't worry, even if the driver improves the performance of the 2900XT significantly, you still have a good card. Being slightly slower is not the end of the world yet since you still have a much better card than the majority of computer user in this world.
 
Caffeinated,

Your SM 2.0 score is 4123 and their score is 4017, I don't see that is a big difference when compaired to 4613, even 4123 is lower. Your SM 3.0 score is 4146 and their score is 3841, the difference is quite big but not as big as the difference between 5115 and 4146.

It seems to me that some 8800 owners especially the GTX owners are really pissed when the 2900XT beats their card even in just one benchmark, the 3DMark06, they are going to be more pissed if the new driver will improve the 2900XT's performance in games too :rolleyes: My advice to the 8800 owners: Don't worry, even if the driver improves the performance of the 2900XT significantly, you still have a good card. Being slightly slower is not the end of the world yet since you still have a much better card than the majority of computer user in this world.


LMAO.. that's enough from this part of the forums for me today..

just..WOW.

You Fan-ATI-cs really don't like ppl pissing on your fire, do you?
 
Looks like bullshit to me... Im sure the new drivers will help but come on lets be real thats too much improvement for drivers alone to make.

Why ? It seems like aa is broken in many apps . If all these drivers are doing are fixing the broken fsaa modes then performance should go up . In some games it looses 60% of its performance by flipping on fsaa while the gts or gtxonly looses 20-30% .
 
Why ? It seems like aa is broken in many apps . If all these drivers are doing are fixing the broken fsaa modes then performance should go up . In some games it looses 60% of its performance by flipping on fsaa while the gts or gtxonly looses 20-30% .

the AA resovle is done by the ALU's it chould have more hit then the g80's easily.
 
lol, all those sites did NOT underclock the GTS or "fudge" the results, moron. Nice of you to igonre all that, cuz you just got shown how wrong you are. And we were not talking about anything else except 3dmark 06, which you are hung up on.

You just got told. Accept it, and move on. Fact is, the XT owns the GTS in 3dmakr06, deal with it.



and @ Sr7, I already mentioned 3dmark06 doesn't mean anything. It does to caffeine however. Im making fun of him in case no one noticed.

One more time, I am going to point out that the GTS in underclocked. I will also refrain from responding in kind, since you called me a moron, but I believe that perhaps you need to examine your own behavior before pointing at others.

Is the default core clock speed of an 8800 GTS 500?

No. It is 513. It is underclocked in those results.

Despite all of that, most everyone else also pointed out that 3DMark06 scores are irrelevant. This is true. It still doesn't change my original point, which is that those results were obtained using an underclocked GTS, which invalidates the conclusions they draw from it. Also, my own personal results using a significantly weaker processor (I have the link to my 3DMark scores, not to mention my system specs are sitting right in my signature for the world to see in case you missed those) are higher at default speeds of 513/1600. Thus my statement that they don't know what they are doing, are purposefully throwing the benchmarks, trying to sell video cards, or all three.

I did not address every other site in the world, just that particular one. Please read before making comments and making personal attacks.
 
One more time, I am going to point out that the GTS in underclocked. I will also refrain from responding in kind, since you called me a moron, but I believe that perhaps you need to examine your own behavior before pointing at others.

Is the default core clock speed of an 8800 GTS 500?

No. It is 513. It is underclocked in those results.

Despite all of that, most everyone else also pointed out that 3DMark06 scores are irrelevant. This is true. It still doesn't change my original point, which is that those results were obtained using an underclocked GTS, which invalidates the conclusions they draw from it. Also, my own personal results using a significantly weaker processor (I have the link to my 3DMark scores, not to mention my system specs are sitting right in my signature for the world to see in case you missed those) are higher at default speeds of 513/1600. Thus my statement that they don't know what they are doing, are purposefully throwing the benchmarks, trying to sell video cards, or all three.

I did not address every other site in the world, just that particular one. Please read before making comments and making personal attacks.

Caffeinated, I have done my best to address every comment that was reasonable, however you have simply pushed this too far.

500 is an implication, not a 100% correct clock speed. They did not break out their timer tools, loop a one-clock operation 500 million times, and time it with a highly accurate scientific tool. What they mean is "stock speed".

[sarcasm]
Well, guess what? ATi's card is supposed to be 748 MHz stock, but they only ran it at 740! What is this? Are they cheating us users of valid benchmarks or something?
[/sarcasm]

I am lucky I don't read too much into these things because just like the 8800 GTS, the implication is stock speed. I'm also amazed that you've managed to see this in the 8800 GTS, but not in the 2900 XT. Please tell me why that is. According to you, they would both be underclocked. The next thing you're going to tell me is...the 8800 GTS was underclocked 5 MHz more, right? and given the 500/513 difference vs. 740/748 difference, the relative difference is a larger, right? It's a conspiracy!

I'm waiting for these arguments from you, but I really really hope that you don't provide them.

Have fun with your speculation. You don't need to reply. All I need is for others to see this post so they can the incredulity of your own.
 
Your SM 2.0 score is 4123 and their score is 4017, I don't see that is a big difference when compaired to 4613, even 4123 is lower. Your SM 3.0 score is 4146 and their score is 3841, the difference is quite big but not as big as the difference between 5115 and 4146.
Well, let's look at the percentages in any case.

SM2.0: Down roughly 2.5%
SM3.0: Down roughly 7.5%

7.5% is actually a fairly substantial delta here (given that the core clock is down only 2.5%, and would have almost zero impact on raw shader performance), and Caff's questioning of the overall score is reasonable given the fairly critical CPU factor. The SM3.0 score should be in the ballpark of 4100 for a GTS, and PCPOP isn't even in the same area code.

I disagree that they're trying to "sell" 2900s, but I think it's good to question their numbers, as the performance delta seems to be statistically significant. Those here who aren't making the connection are doing so intentionally.

You Fan-ATI-cs really don't like ppl pissing on your fire, do you?
I quoted this merely because the dichotomy is amusing.

If 3dmark06 is SO important to you, what are you doing with slow a 8800 GTS?:p :cool: Are you gonna kill yourself now?:D :rolleyes: ;)
If 3DMark is unimportant to you, then why the need to defend a card's performance with it (five times)?

As a bit of advice, your post could have ranked a bit higher on the "childish factor" had you used more emoticons. Overall though, I'd say you've manage to rank somewhere in the 7-10-year-old range, so I offer you my congratulations. Believe me when I tell you that it is quite an accomplishment.
 
Back
Top