Athlon XP 2400+ Mobile @ 2.4 Ghz / 1.9v
Abit NF7-S v2.0 @ 209 x 11.5
Gainward GF4 Ti4200 @ 297/545
The server runs F@H 24/7 so temperatures generated by the CPU are the most extreme this system ever sees. I captured the high/low log of Motherboard Monitor prior to installing the ZM-RF1 to provide a baseline for comparing.
I then installed the fan kit and adjusted the Fan Mate to maximum. The fan is rated for 1800rpm at its highest speed, +/- 10% and as you can see, it only averaged 1620rpm, exactly 10% less than max. The second high/low log was captured over a period of 24 hours and indicates that the fan provided no measurable cooling improvement. The average ambient temperature did not vary much between the tests as can be observed by the chipset temp, which relies on air cooling.
Logic tells us there should be some benefit to increasing airflow over the Reserator. Thinking that the lack of temperature variation may be caused by running the CPU at 100% the entire time, I opted to perform this next test. Leaving the fan setting on high, I stopped the F@H service, reducing the CPU utilization to near 0%. I then noted how long it took for the core temp to register a 10° drop. After recording the results I then restarted F@H, disconnected the fan and waited for the temperature to reach its previous high. Once it did I repeated the procedure but this time without the fan running.
As you can see, the temperature registered in MBM5 as well as the rate of decrease over time between the RF1 at max speed and when disconnected appear as mirror images of each other. Again the results indicate no performance benefit when it comes to using the RF1 vs. running fanless. In both instances, the core temperature was within 1° variance at exactly the same time intervals. So far it appears that the RF1 does not help to maintain lower temps, nor does it help in reducing temps more quickly. There is one more test I want to try however.
To eliminate the possibility of some other source of airflow affecting temperatures, I covered the Reserator with a cardboard sleeve. I started the F@H service and verified the CPU load at 100%, noting that the temperature reached (and maintained) a constant 50° C for a period of half an hour. Then I reconnected the power cord for the fan and ran it at maximum speed for an hour, and recorded the change in temperature. In this environment, the RF1 was able to reduce the CPU core temp by only 2° C.
Zalman put together an effective and nearly inaudible cooling solution with the Reserator 1 Plus, but unfortunately this didn't carry over to the RF1. In testing, the optional Reserator fan kit showed very little improvement in CPU temperatures, and what benefit it did provide was only observed after cutting off all other airflow to the cooling tower. In my opinion, in a "real world" scenario, the Reserator would normally not be placed in a location were it received no airflow at all (other than from the fan), unless the user opted to put the unit in a closet or cabinet of some sort. And if such is the case, well you're going to be glad it's enclosed, because in order to get that 2° C temperature benefit, you have to run the fan at full speed, which produces 32dB, effectively negating the "noiseless" properties of the Reserator system.
I've personally tested several Zalman products, and up to this point it's been my experience that their stuff is well designed and performs well, so it is very surprising to me to obtain these results. I feel that I've given the ZM-RF1 every opportunity to prove itself and it has simply failed to do so. I believe that the principle behind the idea is sound, as evidenced by my own observations of the effect of moving air on the Reserator cooling performance by placing the unit in the direct path of my home's central air return duct. However I can only presume that the volume and velocity of air generated by the RF1 is not substantial enough to make much of a difference. The 2° C improvement recorded in test #3 is a far cry from the 9-11° C benefit that Zalman claims. Users of the Reserator 1 could most likely achieve similar results simply by placing the tower near a source of moving air such as the rear of their case, in the path of most exhaust and power supply fans.
Thanks to Zalman for providing the ZM-RF1 for review.