Classic nVidia FUD over 3870x2

[ShaidarHaran]

Those slides are percent differences, not differences in frames per second.

I know perfectly well what they are.

How fast do you think a 3870 will go at 2560x1600 w/ 8xAA? 10fps? It is pretty silly and misleading to compare % differences when it's probably not playable in the first place.

Please follow the conversation. This isn't even about the 3870. I grabbed those numbers because they're comprehensive, and no one else seems to test with such high settings. The point of all this is that what may not be playable on a single 3870 might well become playable on a 3870 X2, and is very likely to be playable on a CrossfireX configuration.

 

[ShaidarHaran]

you should look at the x2 review the tests are more comparable to what we were talking about

COH
http://www.computerbase.de/artikel/hardware/grafikkarten/2008/test_ati_radeon_hd_3870_x2/20/

Hmm Dx10 much faster for the geforces with x8 aa

Crysis
http://www.computerbase.de/artikel/hardware/grafikkarten/2008/test_ati_radeon_hd_3870_x2/21/

Better for the geforces again

LP better for the geforces again

http://www.computerbase.de/artikel/hardware/grafikkarten/2008/test_ati_radeon_hd_3870_x2/22/

with AA, the radeon's just flat out die (driver problems) at the highest res

WIC
http://www.computerbase.de/artikel/hardware/grafikkarten/2008/test_ati_radeon_hd_3870_x2/23/

you see the % loss for the radeon going to x4 AA in this game its quite large, more then a 50% drop. for the geforces its a bit less at least for the ultra and gtx around 30, now for the gt (new one) yeah its probable hitting a bandwidth bottleneck, same for the the hd3870, but for the x2 it just takes a pounding, still playable on that card though barely...

but all these games these cards are unplayalbe at these settings on pretty much all the games in this suite, so what are we talking about? I pointed this out earlier, its a moot point when you have FPS around 10 don't you think? Or do you play games that are lagging? So what are you saying telling games to come out at these reses and x8AA when they are unplayable mean that the radeons have an advantage? I mean is it really like that? No one has an advantage can't run with em. Now in LP and COH and COJ, the AA hit varies from the other titles, that tells me driver improvement in one way or form can fix the performance for the AA hit to some degree (don't know where the bottlenecks lie, while kinda do, and that would be some instances where the drop takes place, but only in a couple of those games), but I don't see nV really looking into the fix right now (not top priority), cause most new games aren't playable at that res and AA level at least on these cards.

You know what's great about that? Everyone of those games is a TWIMTBP title for which ATi either doesn't have a working CF profile or only just got one working. Not only that but LP and WIC in particular have traditionally performed much better on NV hardware ever since their respective releases. If you want to play that game I can bust out some Call of Juarez and UE3-based title performance numbers.

 

[Razor1]

You know what's great about that? Everyone of those games is a TWIMTBP title for which ATi either doesn't have a working CF profile or only just got one working. Not only that but LP and WIC in particular have traditionally performed much better on NV hardware ever since their respective releases. If you want to play that game I can bust out some Call of Juarez and UE3-based title performance numbers.

really

COJ in the same review

http://www.computerbase.de/artikel/hardware/grafikkarten/2008/test_ati_radeon_hd_3870_x2/19/


and UE3 in the same review

http://www.computerbase.de/artikel/hardware/grafikkarten/2008/test_ati_radeon_hd_3870_x2/17/

Oh its not over yet, Bioshock from the same review

http://www.computerbase.de/artikel/hardware/grafikkarten/2008/test_ati_radeon_hd_3870_x2/18/

As res goes up and settings get higherfor COJ (Dx9, with AA which shows the geforces don't really get hit harder), the HD series fall faster. Hmm Unreal well can't really test out AA on that for ATi, but there should be a driver out there that will enable this soon.....

 

[ShaidarHaran]

really

COJ in the same review

http://www.computerbase.de/artikel/hardware/grafikkarten/2008/test_ati_radeon_hd_3870_x2/19/

That's slower than the regular 3870. Clearly a driver issue.

and UE3 in the same review

http://www.computerbase.de/artikel/hardware/grafikkarten/2008/test_ati_radeon_hd_3870_x2/17/

Still the fastest card tested for this game (except AA doesn't work yet).

Oh its not over yet, Bioshock from the same review

http://www.computerbase.de/artikel/hardware/grafikkarten/2008/test_ati_radeon_hd_3870_x2/18/

Same as with UT3.

So 2 out of 3 it's still the fastest card, and the other is obviously a driver issue since it's slower than a single 3870.

 

[Razor1]

That's slower than the regular 3870. Clearly a driver issue.



Still the fastest card tested for this game (except AA doesn't work yet).



Same as with UT3.

So 2 out of 3 it's still the fastest card, and the other is obviously a driver issue since it's slower than a single 3870.

Oh so you can say something is clearly a driver issue , look at the AA drop for any of the ATi cards. and the x2 doesn't performance worse its a little better around 35% faster then the single, only at lower res does it perform worse which thats normal (edit:ah there seems to be issues with it at other settings too), more CPU bottleneck any dual chip setup will show that.

Now just wait till ya see the AA performance in UE3 games this driver should be out soon.

 

[ShaidarHaran]

Well, since no one else has published X2 or CFX AA numbers...

FPSLABS
- FEAR AA numbers on X2 & X2 CFX.

Looks pretty good to me. I know it's just one game, but it just so happens to be one of the most GPU-limited games around.

 

[Razor1]

Well, since no one else has published X2 or CFX AA numbers...

FPSLABS
- FEAR AA numbers on X2 & X2 CFX.

Looks pretty good to me. I know it's just one game, but it just so happens to be one of the most GPU-limited games around.

FEAR doesn't use UE3 does it?

 

[ShaidarHaran]

FEAR doesn't use UE3 does it?

No, I didn't mean to imply that it does, I was just providing another reference point for AA performance in GPU-limited titles with the X2 and with CFX in particular.

 

[Razor1]

No, I didn't mean to imply that it does, I was just providing another reference point for AA performance in GPU-limited titles with the X2 and with CFX in particular.

ah ok, I agree scaling is generally good.

 

[trinibwoy]

That's pretty impressive. Nothing better to beat back FUD with than a product that works as expected.

 

[XMAN26]

That's slower than the regular 3870. Clearly a driver issue.



Still the fastest card tested for this game (except AA doesn't work yet).



Same as with UT3.

So 2 out of 3 it's still the fastest card, and the other is obviously a driver issue since it's slower than a single 3870.

I'm curious as to how they tested. built in benches and walkthrus without a single bad guy or action or actually played the games. I'm leaning towards the former given there numbers.

 

[ShaidarHaran]

I'm curious as to how they tested. built in benches and walkthrus without a single bad guy or action or actually played the games. I'm leaning towards the former given there numbers.

That's right, let's just assume things for which there is no evidence. Otherwise we'd have to contradict our biases and we can't have that now.

 

[XMAN26]

That's right, let's just assume things for which there is no evidence. Otherwise we'd have to contradict our biases and we can't have that now.

Keep believing canned benches are they best way to test video cards.

 

[AlexV]

The X2 are cursed CURSED I TELL YOU. So, putting together the QuadCF rig I'm(ve, depends) building has been a sinister string of incomprehensible events. So first I burn a Maximus(gah, decided I wanted to try an Abit IX38 anyway so at least I got a proper reason to do it), and afterwards, just before plunging deep into an extended Crysis benchmarking spree, the 1000W Enermax Galaxy decides it's time to give up the fight as well, because, WTF, that was exactly the right time for it to do that.

Thing is I don't burn PSUs or Mobos usually, in spite of putting them through hell...so the X2s are cursed I tell you, cursed.

 

[Kaotik]

The X2 are cursed CURSED I TELL YOU. So, putting together the QuadCF rig I'm(ve, depends) building has been a sinister string of incomprehensible events. So first I burn a Maximus(gah, decided I wanted to try an Abit IX38 anyway so at least I got a proper reason to do it), and afterwards, just before plunging deep into an extended Crysis benchmarking spree, the 1000W Enermax Galaxy decides it's time to give up the fight as well, because, WTF, that was exactly the right time for it to do that.

Thing is I don't burn PSUs or Mobos usually, in spite of putting them through hell...so the X2s are cursed I tell you, cursed.

Don't you need CFX compatible mobo (as in, 790FX, forthcoming Intel P45) to use QuadCF?

 

[AlexV]

Don't you need CFX compatible mobo (as in, 790FX, forthcoming Intel P45) to use QuadCF?

MMnope(or, if you'd rather look at it this way, the X38 is CFX compatible). And I lifted the curse, BTW...these kids are pleasant to play with. Drivers still rough around the edges though.