Xbitlabs 7600GT power consumption measurements

david25 · Post by **david25** » Wed Jun 14, 2006 9:52 am

http://www.xbitlabs.com/articles/video/ ... gto_6.html

=assassin= · Post by **=assassin=** » Fri Jun 16, 2006 5:08 am

I was reading that the other day - 35watts looks very low, which is good - strange how some other sites mention figures of 51watts and even upto 67watts. I've generally gone with XBitlabs figures though.

Beyonder · Post by **Beyonder** » Tue Jun 20, 2006 1:56 pm

I've wondered this exact thing for a while.

Both the 7600GT and 7600GS are based on the exact same core: the G73. Both have the same number of pixel and vertex pipelines, and both have the exact same 128 bit memory bus and GDDR ram. The only difference between the two is clock speed and memory bandwidth. Unless they've bumped the core voltage on the 7600GT to attain the higher clock speed, it'd make sense to me that it had power consumption similar to the GS versions.

...sure wish xbitlabs would come out with some 7600GS power figures...

JVM · Post by **JVM** » Tue Jun 20, 2006 3:20 pm

According to this spec sheet that appears to be from nVidia, the power consumption of the 7600 GT is 67 watts:

http://www.rojakpot.com/showarticle.aspx?artno=318

mattthemuppet · Post by **mattthemuppet** » Tue Jun 20, 2006 4:36 pm

all of the 7600GS cards I've seen have had GDDR2 memory as opposed to the GDDR3 chips on 7600GTs and rough estimates put them at around 30W under load. As frequency has a bigger effect on power consumption than no. of pipes etc, then I'd expect the 7600GT to have around 45-50W under load. Then again, the 7900GT comes in under 50W, so who knows?!

dragmor · Post by **dragmor** » Tue Jun 20, 2006 6:11 pm

JVM wrote:According to this spec sheet that appears to be from nVidia, the power consumption of the 7600 GT is 67 watts:

http://www.rojakpot.com/showarticle.aspx?artno=318

And the spec sheet of my 3500+ says 89w, but TCase says 36.4w. They are just setting themselves a limit. All samples will vary.

stupid · Post by **stupid** » Tue Jun 20, 2006 10:20 pm

dragmor wrote:
And the spec sheet of my 3500+ says 89w, but TCase says 36.4w. They are just setting themselves a limit. All samples will vary.

That's a TDP of 89w (Thermal Design Point) not the actual power consumption. Basically it is how hot the CPU can get (in terms of watts) and still function properly.

dragmor · Post by **dragmor** » Wed Jun 21, 2006 4:20 pm

stupid wrote:
dragmor wrote: And the spec sheet of my 3500+ says 89w, but TCase says 36.4w. They are just setting themselves a limit. All samples will vary.
That's a TDP of 89w (Thermal Design Point) not the actual power consumption. Basically it is how hot the CPU can get (in terms of watts) and still function properly.

Yeah and its the same reason that Nvidia lists 67w for the 7600GT but Xbit measured 35w.

QuietOC · Post by **QuietOC** » Tue Jun 27, 2006 8:09 am

mattthemuppet wrote:all of the 7600GS cards I've seen have had GDDR2 memory as opposed to the GDDR3 chips on 7600GTs and rough estimates put them at around 30W under load. As frequency has a bigger effect on power consumption than no. of pipes etc, then I'd expect the 7600GT to have around 45-50W under load. Then again, the 7900GT comes in under 50W, so who knows?!

The increased wattage of the 7600GT over the 7600GS is probably mainly due to increased voltage needed to maintain stability at the increased clockspeed, not due to the clockspeed increase itself.

The other thing to keep in mind is that voltage needs to be increased to maintained stability at a certain temperature. Thus a better heatsink will allow the voltage to be decreased at the same clockspeed. This combined with just better and/or tested chips might explain why some 7600GTs are more power efficient than others.

mattthemuppet · Post by **mattthemuppet** » Tue Jun 27, 2006 4:37 pm

so when you overclock your CPU at the same voltage, you're saying that there is only a minimal increase in power consumption? That'd be pretty cool if it were true.

cmthomson · Post by **cmthomson** » Wed Jun 28, 2006 6:30 am

mattthemuppet wrote:so when you overclock your CPU at the same voltage, you're saying that there is only a minimal increase in power consumption? That'd be pretty cool if it were true.

It's true. Power consumption is linear with clock frequency, but increases as the square of the voltage.

P = c * F * V * V, where c is some constant.