AMD: R9xx Speculation

They include 6pin+8pin for 225W cards as an option when you draw 75W from each - if 8+8pin would be option when drawing only 75W from one and 150W from the other, it would surely be listed too.

 

Huh? I mean if it were to draw 150 from a 8-pin and 75 from a 6-pin connector placed in any of the two (8-pin) sockets.

This could still apply whether or not the second physical socket is actually 6-pin or not.

Edit: Bad car analogy time. Similar to making a track car designed for peak performance using slicks, but to be road legal too. No RSA would test the car using slicks, that's out of spec, but that doesn't stop the manufacturer from suggesting that the car will only perform to its fullest potential once you take it to the track and put slicks on it.

 

Ahh thanks! That isn't significant, I guess they make it up with volume?

 

Are you sure the Brazos revenue was broken out? I didn't try to look it up, but it's unusual for a company to give that level of specificity.

 

http://www.donanimhaber.com/ekran-k...ran-karti-Tum-detaylariyla-Radeon-HD-6990.htm

 

Draws 350 W during gaming , Powertune max : 375 W , OC : 450 W ..



I guess that means GTX 590 will consume up to 500 W !!

 

Yeesh, the thing does theoretically support up to 12x high resolution monitors (1920x1200) or 8x 2560x1200 displays. If that DL-DVI connector is independant that's potentially 9x 2560x1200 displays for a 3x3 wall of monitors. Although I doubt it'd be able to power a game at 7680x4800 very well.

Regards,
SB

 

I'm willing to bet that you worry about this way more than anyone at the PCI SIG.

How many people, do you think, who were planning to buy this card, are not going to do so if it turns out that there's no official PCIe logo on the box?

The PCI SIG doesn't approve anything. They publish specs. They organize plugfests for compatibility testing for those who wish to participate. They have a logo, the usage of which is probably restricted to some terms (don't know if power consumption is part of those.) That's about it.

 

I call it a fake. OCed will go up over 300W, stock card with default powertune will not.

 

It has 2x8pin power, I think breaking the 300W limit is a given.

 

I also expected it to keep the 300W due to powertune throttling it most of the time (maybe only hitting max frequencies with CF inactive) - and then maybe have something like a +50% powertune setting to set it free.

 

I would expect PowerTune to be set to 300W as a default value, but with a +33.3% setting available; hence the second 8-pin connector. If I were AMD, I would actually test and validate the card for high clocks at 400W, encouraging users to "overclock*" it to those levels, with a simple warning that it takes the card out of PCI-E specs.

*With the card specified for such clocks, it technically wouldn't be overclocking.

 

It seems my number was slightly off; the correct one is $32:

http://img12.imageshack.us/img12/2232/q4asp.gif

 

Here's a video of the 6990 from Cebit

 

Agreed.

Basically AMD is just releasing another 4870 X2, only this time they officially acknowledge its true power consumption. That's all there is to it, in my opinion.

 

But what's the reasoning behind such move?
Do they know nVidia is doing the same? I mean, why else would the break it when they could make a card which would wipe the floor with 5970 within 300W barrier anyway?

 

http://www.heise.de/newsticker/meldung/Radeon-HD-6990-Mit-zwei-GPUs-gegen-Nvidia-1200359.html

heise online says, that CCC says, it's clocked with 800 MHz.

 

AnandTech has some pictures of the card.

 

450W max powertune TDP for the OC version (which is a switch on top of the card) and 375W for the "normal" version.

Who needs to adhere to PCI-E SIG certification anyway? Certainly not AMD.

 

Yes they probably know that, or at least are prepared for it. This is as far as they can go and they went there. Now let's see what nVidia brings and whether they have improved their SLI scaling, otherwise 6990 is looking pretty good.