Performance

After the initial installation, I removed the heatsink to inspect the contact area. The above pictures clearly show that the contact is good, especially on the right side where the thermal compound was almost completely squeezed away. The ATI logo on the core is almost visible. On the edges you can see how the suction resulting from removing the heatsink caused some thermal paste to be sucked back in.

---

For the overclocking tests I used my ATITool overclocking utility version 0.0.22. ATITool has the unique ability to detect artifacts, or flaws, in a rendered image. As defined by ATITool, the maximum stable overclock on a card is the speed at which it is able to consistently (15 minutes in this test) produce no errors, or artifacts. ATITool detects ANY artifacts, even ones which will not be visible in game. Using the human eye to detect artifacts introduces subjectivity into the test, so despite the fact that an ATITool tested overclock will be characteristically lower than a human one, I will use this.

Temperature was measured with one case side open by putting a thermal probe on the backside of the Radeon 9800 Pro opposite of where the GPU is. Idle temperature was measured after letting Windows sit one hour at the desktop. Load temperature was measured after running 3DMark2001 looped for one hour. Both at the card's default clock of 380 / 340 Mhz.

Arctic Silver Lumière was used as thermal interface material for the GPU core in all installations. Lumière is a specially engineered testing compound - it needs no settle in time to reach its maximum performance, but it's not designed for permanent use.

A 7V setting is possible by connecting the fan connector's black wire to the PSU's 5V output, and the fan connector's red wire to the PSU's 12V output (12V - 5V = 7 V).




Additionally to the 9800 Pro tests I tested the ZM-80D on a Radeon X800 Pro. The X800 features dynamic fan speeds which means that fan speed is varied based on temperature. Since it would be a bit unfair to compare those dynamic speeds to the fixed speed of the OP1 I used ATITool to run the stock fan at maximum speed during the 'fan 100%' tests.
Maximum core clocks were verified with ATITool and a number of other applications/games. Default clocks are 475/450 Mhz, temperature was measured using the on-die thermal diode of the X800. The remaining test setup was the same as in the 9800 Pro tests.




While running the ZM-80D passive gives you a completely quiet solution, the temperatures, especially under load, are unacceptable. Don't forget that on the 9800 the temperature was measured on the card's backside so actual core temperature is even higher. Zalman does recommend the addon fan OP-1 for all 9800 cards - same goes for the X800.
My suggestion is to run the ZM-80D with the OP1 add-on fan set to 7V. This gives the best tradeoff between temperatures and noise. I wonder why Zalman didn't include the 7V option in their power cables - I have to admit the 7V connection it is a bit uncommon but it will work fine.
Nothing can beat watercooling on the X800. It seems the core just loves low temperatures.

Value and Conclusion