Testbed and Methods

We tested all the three SFF systems based on i845GE chipset with the following equipment:

• Pentium 4 2.53GHz CPU;
• 512MB DDR333 SDRAM (timings: 2-2-2-5);
• ATI RADEON 9700 PRO graphics card (used in some tests);
• Fujitsu MPG3409AH HDD;
• ASUS CD-S360 CD-ROM drive.

We used the above listed components to assemble two fully-fledged systems in each of the described barebone solutions: one system used the integrated graphics core and the other one used ATI RADEON 9700 PRO add-on graphics card. We ran all tests in Windows XP Professional. The BIOS Setup of the systems tested was set to show maximum performance.

During our tests we checked not only the systems performance but also such important parameters as noise level and thermal conditions.

Performance: Temperature

First of all, we will check the temperature changes. Really, what do we need a constantly overheating system for? Designing an efficient cooling solution is one of the most complicated and more important tasks of the cubic SFF system developers.

At first we assembled the systems, which used the integrated graphics core. The graph below shows the temperature of the working systems in idle state, that is when there are no applications running.

Even in this pretty primitive test the temperatures inside the system cases are very different. The highest temperature was detected inside Shuttle barebone. The thing is that there is only one cooler in this system that blows the stream of air through the system case and cools down the CPU at the same time. Of course, the systems from FIC and Soltek boast lower temperatures inside the case due to an extra cooler.

Now let’s figure out what happens to the temperature values if we load the CPU to the full extent.

Well, the CPU temperature grew up by well over 20^oC. But the overall picture remains the same: Shuttle SB51G copes with this task worse than the competing solutions. FIC IC-VG61 and Soltek EQ2000M with a classical CPU cooler onboard manage to keep the processor temperature at a lower level and hence the overall temperature inside the case appears lower. However, 67^oC under full workload is not that dramatic, because the CPU can work stably in even harder thermal situations. The only thing that we were quite concerned about in Shuttle SB51G system was high temperature inside the case. 48^oC is a pretty dangerous temperature for a hard disk drive, as it may affect the service time of the drive that is why we wouldn’t recommend to use thermally sensitive HDDs in a Shuttle barebone system, just in case.

Now let’s install an add-on graphics card and see how this change will tell on the case temperature. At first we ran the test in idle mode:

ATI RADEON 9700 PRO graphics card is a solution dissipating quite a bit of heat. The only card that can generate even more heat is probably the new NVIDIA GeForce FX, but these systems were lucky to have a cubic case too small for a graphics accelerator like that. Anyway, even ATI RADEON 9700 PRO made the cooling systems work much harder. This card managed to raise the case temperature by 3^oC and the CPU temperature by 6^oC.

Well, and what will happen in the Burn mode? Let’s check it out!

Nothing bad happened. The CPU temperature and the temperature inside the PC case grew up just a tiny bit. All three systems worked absolutely stably in this case, without getting overheated. In fact, this is the most important outcome of the entire thermal investigation. The cooling solutions of all three SFF PCs with “cube” design proved quite efficient, even under heavy workloads.

However, we still have only one leader in this test: this is FIC IC-VG61. Soltek EQ2000M fell just a little behind and Shuttle SB51G without an extra cooler for the CPU turned out the very last one.