The difference in power consumption is probably because the card runs at a lower GPU voltage.

This is why I avoid passive GPUs. Even with the fan undervolted to 5V, and the card massively overclocked, my Asus 8600GT (similar to the OC version reviewed) stays under 80C. At that voltage, I can't even hear the fan over the ambient noise in my room.

It would be interesting to compare the test results to the XFX GeForce 8500 GT passive. Personally, I could live with having an 8500 instead of an 8600 if I get better temperatures and lower power usage.

this passive solution dont look to fancy, the yellow stuff is covering half the heatsink. the good thing about passive sinks is that their kind of overkill, so just adding a fan blowing at the card works miracels for the temps.

as an exaple, got a crappy passive card in an office pc, yesterday i noticed that themp was quite extrem, so i mounted a 12cm noctua at 585rpm blowing from 15cm away and temps dropped ftom 83 to 47 celsius

Nice review! It would be interesting though to see the temperatures on the card if you add a 2nd 120mm fan (undervolted) to the chassis (intake) and direct (part of) the airflow straight at the graphics card. To my mind, this is a more likely setup for most of us - I for one would not run this card in the test setup (although I appreciate this is done to thermally stress the card and see how effective the cooling solution is).

At last a passive 8600 review! Anyway, thanks SPCR!
IMO the card would operate at lower temps in a case with high negative pressure, air incoming through the fins connected to the heatpipe.
I understand the reason for testing stuff in the same old case, but IMO a NSK2480 with two 800 rpm Slipstreams could do wonders for the temps.

the 8500 GT is no good at ANY game though...

Frankly the consistency of power draw across that large loaded temperature range surprises me. Did you just measure at one temperature/fan setting and then assume the rest were the same?

I ask because it seems from other reviews like the power draw/leakage current of the 8x00 GPUs is very sensitive to temperature -- quite the opposite of what those results show.

What a crappy passive heatsink.

I agree, Gigabyte passive solution for the HD 2600 that I have seems to be better to me (=Silentpipe II, single slot).
The 512 MB memory and included HDMI adapter are good "extras", probably it can be overclocked more than the HD 2600 also with adequate cooling.

I only have 256 MB and no HDMI adapter included.
Still I wouldn't like to switch.

Any review of the AC S1 and S2 products in the que.?

With the S1 on that product I wouldnt think that it would hit 50C with the Nexus @ 5v.

FYI, my X1950PRO with my Nexus running at 502rpm is idling at 43C with an ambient of (I'm guessing) 18-22C.


Andy

I was considering this card for a case with positive air pressaure. Heat from the GPU escaping directly out of the case had been nice.

Finally, I didn't go that way. Instead I went for the ordinary version (not OC) and replaced the cooler with Zalman VF900-Cu. At 5V that cooler is inaudible and the capacity is more than enough. However, for that old cooler to fit on this new GPU, you have to use some extra washers.

With some airflow in VGA-department I have these temperatures (ambient 24C):
Idle: 45C
Load (RTHDRIBL or playing games): 55C

A passive VGA-card, not running hot, is still to be seen.

Yeah, I totally agree.

If an Accelero S1 or S2 were compatible with the 8600, it would be nice to see a comparision.

Agreed, as above.

I have a passive card in my HTPC and I run an extra case fan to make sure the graphics card gets the air it needs. I need to upgrade that card and would be interested to see how this one performs with a bit more direct assistance.

I'd say temps would probably improve just by a bit of ducting from the existing intake...

From the section on power consumption:

If I use the numbers in the tables on the same page I get a power usage increase of 188 - 151 = 37 W, not 27 W. The conclusion that the power usage of this card under load is lower than the OC version doesn't follow from the measurements in the article. Did I miss anything?

basfromasd --

No you didn't. You caught a dumb math error on our part.

Unfortunately it wasn't just one error but 2 -- the 37W you calculated is correct, but the 35W power increase shown for the 8600GT OCGear being compared was also wrong -- it should have been 45W. The second error actually came from the original 8600GT OCGear review. Both errors have been corrected in both reviews. Thanks for the catch! And, no, it doesn't really change the text content -- at max load, the fanless card drew 7W less.