HD 2900XTX Benches

[Elfear]

Originally posted by: swtethan

I like how in the graph I see that:

HDR+AA gaming that 2900XT cannot beat a single GTS, but Crossfire XT's can beat SLI GTS? How is that possible? (serious sam graph)

New iteration of Crossfire is more efficient than SLI?

http://www.tbreak.com/reviews/article.php?cat=grfx&id=474&pagenumber=7

on this graph it shows up to 80% increase in SLI mode, but on the bench you provided, only 50% improvement? Whats going on there? lol

Yes I know they are diffrent benches run on different timedemos BUT there SHOULD be a bigger improvement in SLI than 50%

When I ran SLI (7800GTs), I'd sometimes see performance degradation when running both cards at once. That's certainly not the norm but there are quite a few cases where performance increases are only in the 50% range.

http://www.tbreak.com/reviews/article.php?cat=grfx&id=474&pagenumber=9

85% improvement in SLI mode vs single in serious sam HDR vs the 39% on the same res and settings as the site you posted

I probably wouldn't trust the slides off the r800 blogsite either. Not because the numbers aren't possible, but because it's an unknown site with slides directly from AMD/ATI. Hopefully real reviews will come sooner than later.

 

[Fox5]

Maybe the sli tests were run under Vista? Aren't nvidia's sli drivers for vista supposed to suck?

 

[vadp]

Originally posted by: fern420
i was actually trying to find some useful info, not a cat fight but i dont think these forums are moderated anymore.

Just wait for a few days and you'll get the real info.
Right now it's all SPECULATION mixed with BS (including the unverifiable benchmarks).

 

[Ackmed]

Originally posted by: apoppin

Originally posted by: fern420

Originally posted by: mendocinosummit
Is there a way to block certain users in a forum? ^^^^^^

i thought there was a way to add someone to your ignore list and you wont be bombarded with their spam posts but i guess not. i was actually trying to find some useful info, not a cat fight but i dont think these forums are moderated anymore.

what "anymore" ?

you can ignore/block PMs

you can skip over someone's post

Skipping over and blocking are two different things. Especially when a single post is twice as big as my monitor, and its 24". Makes for lots of scrolling. Id like to block a few on here, but you cant do that. Nothing against you, but your posting style is very annoying. 5 words per line, half in italics, other ones with asterisk's, half a dozen faces.. sigh. Just post in a normal paragraph already...

 

[Zstream]

Originally posted by: swtethan

Originally posted by: apoppin
to what ?

undocumented benchmarks on an unknown site?

You failed to read which sadly is not unexpected. It is using Vista and Vista drivers, read the entire post before replying next time please. We all know SLI sucks in Vista.

 

[Munky]

Originally posted by: Nightmare225

Originally posted by: ADDAvenger
I don't really get why the XTX is so much worse than the GTX, you'd think the 320 shaders vs 96 shaders would count for something. Besides, wasn't AMD supposed to be using SIMD shaders so that they'd be more powerful per-clock? Also I don't understand the core clock vs shader clock thing in G80, how does all this fit together?

The GTX has 128. :disgust:

It's this slide which is scaring me: http://bp2.blogger.com/_BabjUDZIqPw/RjL...I/AAAAAAAAACs/omHwk7C2K8c/s400/004.jpg

If true, and if this represents general DX10 performance, somebody save NVIDIA!

I have a hard time believeing Nvidia cards do so badly in these benches. Seems more like someone is trying to make the g80 look bad on purpose.

 

[Extelleron]

Originally posted by: munky

Originally posted by: Nightmare225

Originally posted by: ADDAvenger
I don't really get why the XTX is so much worse than the GTX, you'd think the 320 shaders vs 96 shaders would count for something. Besides, wasn't AMD supposed to be using SIMD shaders so that they'd be more powerful per-clock? Also I don't understand the core clock vs shader clock thing in G80, how does all this fit together?

The GTX has 128. :disgust:

It's this slide which is scaring me: http://bp2.blogger.com/_BabjUDZIqPw/RjL...I/AAAAAAAAACs/omHwk7C2K8c/s400/004.jpg

If true, and if this represents general DX10 performance, somebody save NVIDIA!

I have a hard time believeing Nvidia cards do so badly in these benches. Seems more like someone is trying to make the g80 look bad on purpose.

Well, it could just be the G80 DX10 drivers are very, very bad right now. We have no way of knowing what DX10 performance is like on current drivers, so perhaps it really is that bad. I know performance in nVidia's cascades DX10 demo is pretty bad on my GTS, that's not really a game, but the closest thing we have.

 

[Creig]

Originally posted by: Nightmare225

Originally posted by: ADDAvenger
I don't really get why the XTX is so much worse than the GTX, you'd think the 320 shaders vs 96 shaders would count for something. Besides, wasn't AMD supposed to be using SIMD shaders so that they'd be more powerful per-clock? Also I don't understand the core clock vs shader clock thing in G80, how does all this fit together?

The GTX has 128. :disgust:

It's this slide which is scaring me: http://bp2.blogger.com/_BabjUDZIqPw/RjL...I/AAAAAAAAACs/omHwk7C2K8c/s400/004.jpg

If true, and if this represents general DX10 performance, somebody save NVIDIA!

I don't think Nvidia needs to be putting in any overtime on DX10 performance just yet. Until we get some games/apps that actually utilitize DX10, it's a checkbox feature at best.

 

[Gstanfor]

Originally posted by: Zstream

Originally posted by: swtethan

Originally posted by: apoppin
to what ?

undocumented benchmarks on an unknown site?

You failed to read which sadly is not unexpected. It is using Vista and Vista drivers, read the entire post before replying next time please. We all know SLI sucks in Vista.

Do we?!?

 

[Munky]

Originally posted by: Extelleron

Originally posted by: munky

Originally posted by: Nightmare225

Originally posted by: ADDAvenger
I don't really get why the XTX is so much worse than the GTX, you'd think the 320 shaders vs 96 shaders would count for something. Besides, wasn't AMD supposed to be using SIMD shaders so that they'd be more powerful per-clock? Also I don't understand the core clock vs shader clock thing in G80, how does all this fit together?

The GTX has 128. :disgust:

It's this slide which is scaring me: http://bp2.blogger.com/_BabjUDZIqPw/RjL...I/AAAAAAAAACs/omHwk7C2K8c/s400/004.jpg

If true, and if this represents general DX10 performance, somebody save NVIDIA!

I have a hard time believeing Nvidia cards do so badly in these benches. Seems more like someone is trying to make the g80 look bad on purpose.

Well, it could just be the G80 DX10 drivers are very, very bad right now. We have no way of knowing what DX10 performance is like on current drivers, so perhaps it really is that bad. I know performance in nVidia's cascades DX10 demo is pretty bad on my GTS, that's not really a game, but the closest thing we have.

Let's hope Nv get's its DX10 stuff in top condition soon, because if those benches are true, that will be the biggest upset in video card history.

 

[Matt2]

Originally posted by: Zstream

You know if you stopped using those silly things people would respect you more.

Here are somemore benchmarks More Benchmarks

They are using windows Vista with a few DX10 demo's. No idea about driver settings, I am still looking.

I have a GTX and GTS but I will not touch vista at all. Anyone willing to verify these scores with vista and a GTS/GTX using 1600*1200 8xAA and 16xAF?

Also if you can please download the global illumination demo and test and show us some SS of the test. I would like these verified if possible ASAP so I can say they are fake or possibly correct.

Haha.

We're supposed to believe "AMD ATI officiall test slide" from a site called r800.blogspot.com, but DT is full of crap?

Good one.

Although I do agree with you that someone should download the demo and verify the G80 scores. However, since it's an "AMD ATI officiall test slide", we have no way of knowing how badly the results (if any were even run) were manipulated.

I wonder where HurleyBird is to preach to us how the methodology of those benches should invalidate the results.

 

[apoppin]

Originally posted by: Ackmed

Originally posted by: apoppin

Originally posted by: fern420

Originally posted by: mendocinosummit
Is there a way to block certain users in a forum? ^^^^^^

i thought there was a way to add someone to your ignore list and you wont be bombarded with their spam posts but i guess not. i was actually trying to find some useful info, not a cat fight but i dont think these forums are moderated anymore.

what "anymore" ?

you can ignore/block PMs

you can skip over someone's post

Skipping over and blocking are two different things. Especially when a single post is twice as big as my monitor, and its 24". Makes for lots of scrolling. Id like to block a few on here, but you cant do that. Nothing against you, but your posting style is very annoying. 5 words per line, half in italics, other ones with asterisk's, half a dozen faces.. sigh. Just post in a normal paragraph already...

seriously ... unless i get addressed by you i *never* read your posts

somehow i already know exactly what you are gonna say and what "side" you are gonna take
:roll:

nothing against you but your posts also annoy the crap outta me
... but i didn't want to mention it

--first

 

[sisq0kidd]

I love you all.

 

[swtethan]

Originally posted by: sisq0kidd
I love you all.

:| <3

 

[CrystalBay]

New Drivers BS or not

 

[natto fire]

I didn't read this thread, and I never will.

The few posts I did read were silly in that they seemed to be happy that there was no competition in the GPU market. Heres hoping something will come of software developments, or that someone will become more competitive when it comes time to replace my 7900GS. In the meantime I hope all you people enjoy arguing with each other and wasting your money...

 

[HurleyBird]

Originally posted by: Matt2

Originally posted by: Zstream
I wonder where HurleyBird is to preach to us how the methodology of those benches should invalidate the results.

Funny. Of course these benchmarks can't be taken as authoritative either for obvious reasons. My position remains unchanged. Wait for the damn NDA to lift before passing judgment.

 

[apoppin]

we ARE waiting



you are just getting one *preview* from DT

Hopefully, AMD's r600 XTX won't be a total 'wash' ... new drivers may mitigate it somewhat ... DX10 performance will have to weighed ... and of course it's "features"
... but i guess XTX won't be out for quite awhile - if ever in quantity

So, i guess the treat is XT vs GTS and GTX

not too long now

 

[Gstanfor]

new drivers may mitigate it somewhat

Well at 350mb I'd say they almost *have* to be miracle drivers, and they had better be publicly available too. A set of drivers turning up only for review purposes and being so large will look mighty suspicious indeed.

 

[HurleyBird]

Originally posted by: apoppin
we ARE waiting

The smart people are waiting. The less smart have already made their decisions one way or the other.

 

[Ackmed]

Originally posted by: apoppin


seriously ... unless i get addressed by you i *never* read your posts

somehow i already know exactly what you are gonna say and what "side" you are gonna take
:roll:

nothing against you but your posts also annoy the crap outta me
... but i didn't want to mention it

--first

Doubtful. What side I took? Yeah, I took the 'wait and see side'. Unlike most people in this thread, which have already made an opinion on unfounded numbers, either way. Sure I prefer ATi, everyone prefers one side or the other, no matter if they admit it or not.

One day you're going to have to go out in the real world, and make a real paragraphs that are comprehendable...

Originally posted by: HurleyBird

Originally posted by: apoppin
we ARE waiting

The smart people are waiting. The less smart have already made their decisions one way or the other.

True. Sadly most are not smart enough to wait. None of these numbers prove anything to me. And they are hardly a real review. I want more that a one page article, with no mention of IQ, heat, noise, new features, etc. DT's numbers may very well be spot on, but Ill wait for real reviews to back them up. The others look even less likely.

edit, while DT didnt run these numbers themselves, they also took the bait by posting them. If you remember the whole PC Perspective debacle. Again, not saying DT is up to anything, but its just more reason for me to wait and see, from a real review.

 

[TanisHalfElven]

i don't get ppls response to this. 7900gtx was slower than x1900xtx and sometimes even x1900xt yet nvidia sold a ton of them. why will amd die for x2900xtx not beating 8800gtx when nvidia did not.

edit spelling and typos

 

[xSkyDrAx]

Originally posted by: Ackmed

Originally posted by: apoppin


seriously ... unless i get addressed by you i *never* read your posts

somehow i already know exactly what you are gonna say and what "side" you are gonna take
:roll:

nothing against you but your posts also annoy the crap outta me
... but i didn't want to mention it

--first

Doubtful. What side I took? Yeah, I took the 'wait and see side'. Unlike most people in this thread, which have already made an opinion on unfounded numbers, either way. Sure I prefer ATi, everyone prefers one side or the other, no matter if they admit it or not.

One day you're going to have to go out in the real world, and make a real paragraphs that are comprehendable...

Originally posted by: HurleyBird

Originally posted by: apoppin
we ARE waiting

The smart people are waiting. The less smart have already made their decisions one way or the other.

True. Sadly most are not smart enough to wait. None of these numbers prove anything to me. And they are hardly a real review. I want more that a one page article, with no mention of IQ, heat, noise, new features, etc. DT's numbers may very well be spot on, but Ill wait for real reviews to back them up. The others look even less likely.

edit, while DT didnt run these numbers themselves, they also took the bait by posting them. If you remember the whole PC Perspective debacle. Again, not saying DT is up to anything, but its just more reason for me to wait and see, from a real review.

The next best thing is always around the corner. Play the waiting game and that will be the only game you will be playing. Upgrade according to your present needs. All those people waiting on the R600 to say 'i told you so' instead of playing the games they wanted to sure are brilliant...i wonder what's their secret.

 

[Ackmed]

Originally posted by: xSkyDrAx
The next best thing is always around the corner. Play the waiting game and that will be the only game you will be playing. Upgrade according to your present needs. All those people waiting on the R600 to say 'i told you so' instead of playing the games they wanted to sure are brilliant...i wonder what's their secret.

I wasnt talking about waiting to see what reviews said, before buying. I was talking about waiting for real reviews before making the claim that ATi/AMD sucks, the R600 sucks, and its a crappy card. Many people are already making such claims.

I too dont agree with waiting to buy a new card. Unless you know for fact that one is coming out very soon. The 8800GTX has been the best card for about 6 months, and it might only get beaten by the 8800GTX Ultra. Which is probably going to cost a lot more, and not be all that much faster that current overclocked GTX's. The next best thing, is not always around the corner. I would just buy a 8800GTX right now, and not even consider waiting on a Ultra if I was in the market for a card.

 

[Extelleron]

Originally posted by: Matt2

Originally posted by: Zstream

You know if you stopped using those silly things people would respect you more.

Here are somemore benchmarks More Benchmarks

They are using windows Vista with a few DX10 demo's. No idea about driver settings, I am still looking.

I have a GTX and GTS but I will not touch vista at all. Anyone willing to verify these scores with vista and a GTS/GTX using 1600*1200 8xAA and 16xAF?

Also if you can please download the global illumination demo and test and show us some SS of the test. I would like these verified if possible ASAP so I can say they are fake or possibly correct.

Haha.

We're supposed to believe "AMD ATI officiall test slide" from a site called r800.blogspot.com, but DT is full of crap?

Good one.

Although I do agree with you that someone should download the demo and verify the G80 scores. However, since it's an "AMD ATI officiall test slide", we have no way of knowing how badly the results (if any were even run) were manipulated.

I wonder where HurleyBird is to preach to us how the methodology of those benches should invalidate the results.

Where can I download this global illumination demo?

BTW I think those of you who still believe these DT benches should look at what those who are still under NDA have had to say:

http://forums.amd.com/forum/messageview...STARTPAGE=20&FTVAR_FORUMVIEWTMP=Linear

I think it's interesting that DT would compare an underclocked XTX (750MHz core vs 800MHz, which is what the clock apparently will be) to a heavily overclocked GTX (650/2000 vs 575/1800).