Details of the 9600 GT Green Edition Leaked

[alexp999]

Expreview, has got hold of information which details NVIDIA's new 9600GT Green Edition GPU. It appears to be nothing more than a BIOS tweak, but as such has resulted in a 37W (39%) drop in power consumption, allowing the card to function without the need of an external power connector. This has resulted in a new board revision number P562. The memory clocks remain the same at 900 MHz, but the core drops to 600 MHz, with 1500 MHz on the shaders. By doing this, NVIDIA have been able to drop the core voltage from 1.1V to 1.0V, most likely providing the biggest saving in power.



View at TechPowerUp Main Site

 

[evil_raptor]

is this a 55nm GPU?
9600GT is only 65nm AFAIK

 

[AnnCore]

Not the same as...?

Correct me if I'm wrong but if you underclock anything won't it consume less power?

External connector or not, less MHz should result in less power draw no?

 

[alexp999]

is this a 55nm GPU?
9600GT is only 65nm AFAIK

9600GT has been 55nm for a while:

http://forums.techpowerup.com/showthread.php?t=67004

 

[Deleted member 3]

Interesting, shame my board doesn't support SLI. Would be an interesting relatively low power solution while remaning quite a nice performance level. If AMD would come with such a card I'd gladly have two.

 

[alexp999]

Correct me if I'm wrong but if you underclock anything won't it consume less power?

External connector or not, less MHz should result in less power draw no?

Sure does, but like I said in the post, the biggest saving of power should be the drop in core voltage.

 

[Deleted member 3]

Correct me if I'm wrong but if you underclock anything won't it consume less power?

External connector or not, less MHz should result in less power draw no?

Sure, but not close to the claimed improvement. The connector isn't the cause, but an improvement, it's no longer required.

 

[AnnCore]

Sure does, but like I said in the post, the biggest saving of power should be the drop in core voltage.

I can understand that a drop in core voltage should result in power savings but the performance also dropped so this seems like an advertising gimmick to me.

If they had dropped the core voltage, got rid of the external power connector, and kept the same core speed as before then I'd be impressed.

Or am I still missing something?

 

[alexp999]

I can understand that a drop in core voltage should result in power savings but the performance also dropped so this seems like an advertising gimmick to me.

If they had dropped the core voltage, got rid of the external power connector, and kept the same core speed as before then I'd be impressed.

Or am I still missing something?

No your right, all they are doing in underclocking it, so make it use less power.
You could quite easily do the same thing on a current 55nm 9600GT

 

[evil_raptor]

this would be ideal for a dedicated physx card

 

[1c3d0g]

Correct me if I'm wrong, but moving from 96W TDP to only 59W TDP is a major, positive improvement, core voltage drop or not.

 

[AnnCore]

Correct me if I'm wrong, but moving from 96W TDP to only 59W TDP is a major, positive improvement, core voltage drop or not.

The only way to know for sure is to take an old 9600GT, down clock it to the 600MHz of this Eco version and see if it draws more than 59 TDP. If it does then obviously the volt drop via the bios flash is good for something and maybe not as gimmicky as I had said earlier.

 

[mdm-adph]

Interesting, shame my board doesn't support SLI. Would be an interesting relatively low power solution while remaning quite a nice performance level. If AMD would come with such a card I'd gladly have two.

I gladly present you with the HD 4670, in case you haven't already met:

Nice performance level with relatively low power draw. In CrossfireX, two are more powerful than a 4850.

 

[NeSeNVi]

I hate nowadays graphic cards because of power consumption, which for me is not acceptable, but this one interested me...

 

[Cheeseball]

Heh, so the new 9600GTs won't need the PCI-E 6-pin connector? Excellent! Hopefully the clock drop didn't affect performance too much, meaning this should be a HD4670 killer if given a good competitive price.

 

[timta2]

I would be willing to bet that if you used this card 24 hours a day you wouldn't notice a decrease in the power that your rig used. I mean even if you were paying 50 cents per kilowatt hour it would probably amount to 50 cents worth or less and handicapping my video card to save 50 cents a month doesn't seem logical to me.

 

[NeSeNVi]

I would be willing to bet that if you used this card 24 hours a day you wouldn't notice a decrease in the power that your rig used. I mean even if you were paying 50 cents per kilowatt hour it would probably amount to 50 cents worth or less and handicapping my video card to save 50 cents a month doesn't seem logical to me.

In Poland cost of energy is more than just 50 cents... and you can or not believe me, but I see the difference with use more or less power consumption devices on my bills.

 

[DaJMasta]

In Poland cost of energy is more than just 50 cents... and you can or not believe me, but I see the difference with use more or less power consumption devices on my bills.

50 cents per kWh?


It's neat that they can drop consumption that much with modest drops in speed and core voltage.... but I've gotta wonder whether this is really useful. If they make this card into a passive/low profile card then it gets much more appetizing, but if it's using the same old cooler and size it seems like you're getting a slightly slower card to get your green ePeen going. There's relatively few applications which can't afford the 40W difference between the cards but still need a midrange discrete card.

 

[Hayder_Master]

the better thing in 9600gt is the overclock , so with this i don't think so