[Testing] Sapphire X2900XT

[cadaveca]

Cantankerous..the smoothness would come from having a more consistent framerate, not neccesarily faster, but with less "peaks and lows".

[cantankerous]

Cantankerous..the smoothness would come from having a more consistent framerate, not neccesarily faster, but with less "peaks and lows".

thanks cadaveca. Makes sense. So would one say that ATI cards have a more consistent framerate without so many peaks and lows like Nvidia cards or is all this just an old wives tale?

[Losphoron]

This is a very good question when asking about the smoothness. This is not the first time I have heard someone say that gameplay with an ATI card is smoother than that of an nvidia card. I don't understand how a card as powerful as the 8800GTX can't play a game smoothly where as an ATI branded card can? Is there any truth to this? Are Nvidia cards really not as smooth and fluent when playing games when compared to an ATI card? What makes an ATI card smoother if the case may be so?

I had a similar problem with some games... When you put tones of eyecandy in a game at a specific resolution the game feels heavy and loaded with detail and you feel somekind of lag when you move in the game even if you have more than 30fps constant..Usually nvidia wasn't so good when activating eyecandy.They just aren't so eficient..The same thing is happening now with 2900xt vs 8800GTS..When you increase the resolution the 8800 can't keep up..This translates in higher value for the lowest fps in specific scene..I speak from my experience over time with gaming..

[DilTech]

I had a similar problem with some games... When you put tones of eyecandy in a game at a specific resolution the game feels heavy and loaded with detail and you feel somekind of lag when you move in the game even if you have more than 30fps constant..Usually nvidia wasn't so good when activating eyecandy.They just aren't so eficient..The same thing is happening now with 2900xt vs 8800GTS..When you increase the resolution the 8800 can't keep up..This translates in higher value for the lowest fps in specific scene..I speak from my experience over time with gaming..

The 8800GTS beats out the HD 2900xt in gaming, even at high rez. Unless you're talking about the 320mb, which wasn't intended for high-rez in the first place.

Wow! good thread guys. I'm glad to see some familiar xs members test these cards. Thats awesome that the mcw50 and 60 will fit because I still have one from my old 9800 pro. Does the HD2900XT feel any smoother in games than the 8800?
I feel like ati always has a "smoother" feel in games or is that just my imagination? Anyway I thank you guys for testing the cards but I still don't see a reason to get one if you already have a high clocking 8800gts.

There's no reason to get a HD 2900xt if you own a 8800GTS currently, as for right now the 8800gts wins in gaming anyway. I'd sit on what you have, and perhaps the refresh(R650) or the G92(4th Q this year) will give you a solid reason to upgrade.

[LowRun]

The 8800GTS beats out the HD 2900xt in gaming, even at high rez. Unless you're talking about the 320mb, which wasn't intended for high-rez in the first place.

Actually no.

[DilTech]

Actually no.

Reviews tell me otherwise?

[deathman20]

Reviews tell me otherwise?

And most reviews you have to take with a grain of salt.

I do agree though from my personal experience at least seeing nVidia cards in systems, the ATI seems to have a smoother feel to it. While sure the FPS might be lower but its lows and highs aren't as bad.

The 2900XT can take on the 8800GTS in games very nicely. You can't go by all reviews because everyone seems to use different drivers and they are all over the board with them. OC's on cards, references clocks, etc.

[DilTech]

Again, reviews tell a different tale.

Of course, things may all change when the new driver finally releases. That's why I said in my earlier post currently. Because in all honestly, currently the 8800gts can't be touched by the HD 2900xt.

[cadaveca]

Sry Diltech, but my own testing show far differnt numbers than alot of reviews have. In fact, I find most reviews to be completely erroneous except for clockspeeds.

[DilTech]

Sry Diltech, but my own testing show far differnt numbers than alot of reviews have. In fact, I find most reviews to be completely erroneous except for clockspeeds.

Odd...

http://forums.extremeoverclocking.co...0&postcount=22

Application Controlled but I activate adaptive AA. Well I will stop testing games because as of right now, the HD2900XT cannot touch the 8800GTS 640MB. Did the Lost Coast test and needless to say the 8800GTS 640MB did like 45fps over. In the meantime I've moved the HD2900XT to watercooled.

He has both cards, and has quit even trying games on the HD 2900XT due to it getting stomped by his 8800GTS. His results with the games he tested were said to say the least.

Again, this just adds to the reason I say currently the hd 2900xt can't compare to the 8800gts.

[mascaras]

it is not necessary to cool the mem, but the ic's have to be cooled, so i cool those things with the stock cooler (without the copper block for the GPU, which i replaced with the mc60).

albert can you post some pics plz ??

i bought one MC60 today and i like to see how you use the stokcooler + MC60


regards

[cadaveca]

Problem lies in nVidia card being there first, and not a fresh install of Windows, such as I have. This leads to a totally different sytem environment, which, in my case, leads to better numbers.

You'll also notice that some of the benches I have posted are a fair bit higher than others have been able to get at the same clockspeeds...

I chaulk it up to reviewers using drivers that the public currently does not have access too, or using incomplete drivers, as I hsve similar performance using CAT7.4, but the update offered on ATI's website leads to far different numbers than what review websites have.

And of course you can't currently compare...you don't have the cards. I do tho, and it paints are far differnt picture.


Anyway, I will add this, which you will like...most reviews did not mention 2d crashes, 2d rendering errors, and numerous other niggles that I have been dealing with since getting my cards. These problems are evident on all of the 6 cards I have currently, and this points quite obviously to driver issues.

I mean really, Diltech...498gflops in r600, 347gflops in G80...it's IMPOSSIBLE for G80 to be faster unless there are driver issues.

[nicepun]

Let me grab some popcorn because DilTech can argue forever....LOL

[cadaveca]

lol i know..he and I have "gone at it" numerous times.

[cantankerous]

And most reviews you have to take with a grain of salt.

I do agree though from my personal experience at least seeing nVidia cards in systems, the ATI seems to have a smoother feel to it. While sure the FPS might be lower but its lows and highs aren't as bad.

The 2900XT can take on the 8800GTS in games very nicely. You can't go by all reviews because everyone seems to use different drivers and they are all over the board with them. OC's on cards, references clocks, etc.

Oh for the love of God almighty himself why does it never end! What I am being long about is the decision which is impossible for me to make in regards to getting the 2900XT or the GTX. I have had this constant battle in my mind for weeks over this and I just can't come to a decision. I had my mind all set on the 8800GTX due to its massive performance and supposedly better IQ but then I come across statements like above and it puts me into a panic to spend the money on the card just in case. I would hate to spend the money on the GTX and have choppy, erratic, non smooth gameplay especially since I have been nothing but a pro ATI guy since the 9800 pro. This really is killing me. If the GTX being as powerful as it is isn't at least as smooth as my current X1900XTX I would just cry with buyers remorse. Someone kill me.

[deathman20]

Odd...

http://forums.extremeoverclocking.co...0&postcount=22



He has both cards, and has quit even trying games on the HD 2900XT due to it getting stomped by his 8800GTS. His results with the games he tested were said to say the least.

Again, this just adds to the reason I say currently the hd 2900xt can't compare to the 8800gts.

Well thats one test. HL2 has issues with FPS currently especially depending what is enabled. Fog issues some AA issues its a hit on performance thats for sure and thats what hes seeing because I notice it. Did he even try other games?

This is my results so far what I've tested sorry about speradic tests. One thing is for sure though which I didn't test all the way is CPU speed is a factor and from 3.0Ghz to 3.6Ghz there are very nice increases in games. I didn't record the results since thats going to take even more time and I just want to spend that amount. If anything people can look at the setup that I did and its will get better with higher CPU speeds. As well also keep in mind this is VISTA. Results will differ from XP tests in some cases.
http://www.ocforums.com/showthread.php?t=511746

albert can you post some pics plz ??

i bought one MC60 today and i like to see how you use the stokcooler + MC60


regards

Thats easy... Attach the MC60 and just remove the plastic cover and the heatsink. You must have a fan blowing on the card though otherwise it will get extremely warm. If you want you can even remove the fan off the heatsink. Theres 3 screws if not mistaken holding that in.

Even this solution cools it better than the stock while only thing your doing is blowing a fan over the plates.

[DilTech]

Problem lies in nVidia card being there first, and not a fresh install of Windows, such as I have. This leads to a totally different sytem environment, which, in my case, leads to better numbers.

You'll also notice that some of the benches I have posted are a fair bit higher than others have been able to get at the same clockspeeds...

I chaulk it up to reviewers using drivers that the public currently does not have access too, or using incomplete drivers, as I hsve similar performance using CAT7.4, but the update offered on ATI's website leads to far different numbers than what review websites have.

And of course you can't currently compare...you don't have the cards. I do tho, and it paints are far differnt picture.


Anyway, I will add this, which you will like...most reviews did not mention 2d crashes, 2d rendering errors, and numerous other niggles that I have been dealing with since getting my cards. These problems are evident on all of the 6 cards I have currently, and this points quite obviously to driver issues.

I mean really, Diltech...498gflops in r600, 347gflops in G80...it's IMPOSSIBLE for G80 to be faster unless there are driver issues.

Since when has GFlops always meant real-world performance, just out of curiousity? It's very possible for a card much faster in GFlops to be slower, so I have no idea where this "it's impossible" comes from, because that's far from the case.....

Going by your logic, the R580 would be faster than the G80. That isn't the case in reality though, now is it?

Also, you make it out that I want this card to suck, and that's not the case. Infact, I want the competition between NVidia and Daamit to be steeper, so as to call for bigger price cuts, which is better for all consumers. As it stands currently though, the competition just isn't there.

[cadaveca]

Oh for the love of God almighty himself why does it never end! What I am being long about is the decision which is impossible for me to make in regards to getting the 2900XT or the GTX. I have had this constant battle in my mind for weeks over this and I just can't come to a decision. I had my mind all set on the 8800GTX due to its massive performance and supposedly better IQ but then I come across statements like above and it puts me into a panic to spend the money on the card just in case. I would hate to spend the money on the GTX and have choppy, erratic, non smooth gameplay especially since I have been nothing but a pro ATI guy since the 9800 pro. This really is killing me. If the GTX being as powerful as it is isn't at least as smooth as my current X1900XTX I would just cry with buyers remorse. Someone kill me.

I hate to say it, but even after buyingthese cards, I am unsure myself whther this purchase was the right one or not.

I want to support the underdog, namely AMD. they need the help ATM...and the industry needs competition. The 2900XT is attractively priced, but I have a hard time saying it's a good buy.

I also jsutified my purchase due ot the "valve coupons" inside, as I would have bought the games irregardless, and now i don't have to.

So, keep your money in your pocket for a couple more weeks yet.

[DilTech]

Oh for the love of God almighty himself why does it never end! What I am being long about is the decision which is impossible for me to make in regards to getting the 2900XT or the GTX. I have had this constant battle in my mind for weeks over this and I just can't come to a decision. I had my mind all set on the 8800GTX due to its massive performance and supposedly better IQ but then I come across statements like above and it puts me into a panic to spend the money on the card just in case. I would hate to spend the money on the GTX and have choppy, erratic, non smooth gameplay especially since I have been nothing but a pro ATI guy since the 9800 pro. This really is killing me. If the GTX being as powerful as it is isn't at least as smooth as my current X1900XTX I would just cry with buyers remorse. Someone kill me.

Being a 8800GTX owner I can tell you right now gaming is smooth as can be on it, and that's with a "lowly" 939 X2 3800+.

[cadaveca]

Since when has GFlops always meant real-world performance, just out of curiousity?

DX10, my man. "STREAM" processing...many different applications.

Now, in r580 vs g80, there is specific functionality missing that prevents a decent comparison, but there are palces where the r580 outshines the g80...and it's definately not in gaming...

But there's the rub. if you have truly homohgeneous code, even then the compare is not accurate...

But the fact of the matter is that once workloads designed for these gpus are put on these gpus, perspectives may change. It's far to early to make any assesment as to who is truly faster...right now nV is...but not becuase thier HARDWARE is faster..it's better drivers.

I'll tell ya right now that those reviews aren't using release drivers for the G80. Those drivers sucked as bad...no...worse..than ATI's current "rel;ease" drivers. If they DID use those release drivers, the performance numbers would be very different.

PLease read my last post, and notice I hazard against a purchase in either direction...I'm not playing favorites, merely stating facts.


Btw...perfect example of r580 beating g80...Folding @ home. That's reality, man.

[Sentential]

Since when has GFlops always meant real-world performance, just out of curiousity? It's very possible for a card much faster in GFlops to be slower, so I have no idea where this "it's impossible" comes from, because that's far from the case.....

And there in lies the real issue. The fact is the R600 is supposed to be faster and on paper it is. The drivers are the issue, but more so the design of the card itself is also playing a significant role in this.

The deal with the G80 is that the way it uses its shaders causes all of its 96 / 128 shaders units to be active at all times. The R600 since it is super-scaler allows it to multi-task on its shading units which is both good and bad. The good news is that its alot more flexible and faster... the *BAD* news is that it is damn impossible to keep all shading units active at all times efficientally.

The drivers play a critical role in this and since as we all know ATi cant program drivers for , they will most likely never reach peak Gflops. If and I do mean IF it does, then yes it would be alot faster. Will it? No

Being a 8800GTX owner I can tell you right now gaming is smooth as can be on it, and that's with a "lowly" 939 X2 3800+.

Ive been in the nVidia camp for quite some time only buying 2 ATi cards ever, and I personally feel that nVidia cards are far smoother than ATi. In addition to that I find ATi's coloring methods make the picture too washed out and grey for my tastes.

[cantankerous]

I hate to say it, but even after buyingthese cards, I am unsure myself whther this purchase was the right one or not.

I want to support the underdog, namely AMD. they need the help ATM...and the industry needs competition. The 2900XT is attractively priced, but I have a hard time saying it's a good buy.

I also jsutified my purchase due ot the "valve coupons" inside, as I would have bought the games irregardless, and now i don't have to.

So, keep your money in your pocket for a couple more weeks yet.

I do greatly appreciate your honesty cadaveca. I am sorry to hear you are unhappy with your 2900XT purchase. That is the exact feeling I hate to have especially when dropping money I don't necessarily have. Any card I do get is dipping into my car fund which I am saving for for this fall. The choice has to be the right one for me to be happy about it and not regret saving that money instead of spending it.

[cantankerous]

Being a 8800GTX owner I can tell you right now gaming is smooth as can be on it, and that's with a "lowly" 939 X2 3800+.

DilTech, have you always been an Nvidia guy, ATI guy or both? I find people make the comment about the smoothness when it comes to being pro ATI and being used to the way ATI cards run, then all of a sudden jumping to the other side where things 'just don't quite feel right' you know?

How long have you had your 8800GTX? What is it clocked at and what is your 3800+ clocked at? You are saying it is greased lightning smooth and never hitches, pauses, stutters, lags or feels not as consistent in fps as an ATI card?

[DilTech]

DX10, my man. "STREAM" processing...many different applications.

Now, in r580 vs g80, there is specific functionality missing that prevents a decent comparison, but there are palces where the r580 outshines the g80...and it's definately not in gaming...

But there's the rub. if you have truly homohgeneous code, even then the compare is not accurate...

But the fact of the matter is that once workloads designed for these gpus are put on these gpus, perspectives may change. It's far to early to make any assesment as to who is truly faster...right now nV is...but not becuase thier HARDWARE is faster..it's better drivers.

I'll tell ya right now that those reviews aren't using release drivers for the G80. Those drivers sucked as bad...no...worse..than ATI's current "rel;ease" drivers. If they DID use those release drivers, the performance numbers would be very different.

PLease read my last post, and notice I hazard against a purchase in either direction...I'm not playing favorites, merely stating facts.


Btw...perfect example of r580 beating g80...Folding @ home. That's reality, man.

I know all about the release drivers for the 8800GTX.. I bought one at launch, and still had numbers higher than what the R600 showed, but the R600 wasn't slated against the GTX so that really doesn't count obviously. Seeing as how I didn't own a GTS I cannot speak upon it's performance at that junction. Also, why should reviewers go back in time just because ATi's drivers aren't currently up to snuff? You compare a product in it's current form to a product in it's current form. Not a new product now against the condition of it's competitor 7-8 months ago.

Also, I don't care about folding on my gpu, I don't care about side software on my gpu, I don't care which gpu can find a prime number fastest, and I certainly don't care about if my gpu can break down pi faster than the opposition...

All I care about is games, that's it, plain and simple. I don't buy a gaming video card to do other things with but play games, and I'm sure most people would agree with that statement. I wouldn't be surprised to see ati faster at GPGPU functions, but in the end that's not what people buy these cards for.

So, when it comes to video cards, frames per second with AA and AF maxed out is the only reality that matters to an overwhelming majority of card owners. Theres those who see things differently, and to them gpgpu makes a difference, but for the most part it's in game numbers that decide what graphics card they buy.

So yes, I'm still standing by my prior statement that it's very much possible for the G80 to be faster than the R600, even though the R600 can perform more GFlops.

[DilTech]

DilTech, have you always been an Nvidia guy, ATI guy or both? I find people make the comment about the smoothness when it comes to being pro ATI and being used to the way ATI cards run, then all of a sudden jumping to the other side where things 'just don't quite feel right' you know?

How long have you had your 8800GTX? What is it clocked at and what is your 3800+ clocked at? You are saying it is greased lightning smooth and never hitches, pauses, stutters, lags or feels not as consistent in fps as an ATI card?

I buy whatever card is the best at the time I'm ready to purchase a card, plain and simple.

I've had both ATi and NVidia cards, infact my old 9800pro was one of my favorite videocards that I've ever owned. For it's time, that card never gave me any complaints, and it still lives today.

I've had my 8800GTX since around launch, it was about a week to a week and a half after launch because we were hit with a huge snow storm that delayed packages a few days, over a foot of snow plus a few inches of ice under it.

My x2 is currently only clocked at 2.5 ghz because I really haven't had a chance to test it at higher frequencies due to work(I'm working right now as I type this), but at these small frequencies I've had no problems playing any games I play maxed out. This includes 16xCSAA.

What I'd advise for you though is this... You have a x1900, you're fine until the new drivers for the R600 come out. Wait for that, and for well trusted review sites to put out numbers for these drivers. VR-Zone promised a re-run of their review when ATi release cat 8.38, so wait for those numbers.

Reason being, even though it's pretty obvious the HD 2900XT isn't meant to compete with the 8800GTX, you'll atleast be able to decide if it's worth the extra money for whatever the performance percentage difference is.

I don't think there's anyone who will disagree with my logic in that advice, right guys?