Testbed and Methods

All tests were carried out on two platforms: Intel and AMD. We tested the systems in open stands as well as with closed system cases. We used the following hardware components to assemble our testbeds:

• Mainboards:
  • ABIT AN8 SLI (nForce 4 SLI), Socket 939, BIOS v.1.9;
  • ABIT AA8-DuraMAX (Intel 925X), Socket T (LGA 775), BIOS v.2.4;
• CPUs:
  • AMD Athlon 64 3000+ 1800MHz, 512KB, Cool & Quiet Disable, (Venice, E6);
  • Intel Pentium 4 516 2933MHz Prescott, 533 MHz FSB, 1MB, (SL8J9 Malay, E0);
• Thermal paste: KPT-8;
• System memory:
  • DDR, 2 х 512MB PC3200 Corsair TWINXP1024-3200C2;
  • DDR-2, 2 x 512MB PC4300 Corsair TWIN2X1024A-4300C3;
• Graphics card: PCI-Express x16 Sapphire Radeon X800 GTO2 256MB GDDR-3, @540/1180MHz 16p;
• Graphics card cooling system: Zalman VF700-Cu, 5V, 1350rpm, 18.5dBA;
• HDD: SATA-II 160 GB Seagate Barracuda 7200.9 (ST3160812AS 2AAA) 7200rpm, 8MB;
• DVD±R/RW & CD-RW drive: Toshiba SD-R5372;
• Case: ATX ASUS ASCOT 6AR2-B Black&Silver + two 120mm case fans: Sharkoon Luminous Blue LED (~1000rpm, ~21dBA);
• PSU: XG Magnum 500W + 80mm GlacialTech SilentBlade fan (~1700rpm, 19dBA).

All the tests were conducted in Windows XP Professional Edition Service Pack 2 with NVIDIA nForce system driver version 6.82. I would like to point out specifically, that we used KPT-8 thermal paste for all our tests today. I have already mentioned the results of the efficiency comparison between the Arctic Cooling’s own and our thermal pastes. The automatic fan rotation speed control option was disabled in the BIOS Setup of both mainboards.

AMD Athlon 64 3000+ (1800MHz) was overclocked to 2700MHz at 1.575V Vcore:

The Intel CPU marked as SL8J9 and manufactured in Malaysia was overclocked to 4019MHz without raising the Vcore at all:

We used S&M utility version 1.7.6 (beta) for CPU warm-up, temperature monitoring (with SpeedFan 4.26 utility) and fan rotation speed control. We warmed up the CPU for 15 minutes in the FPU test under 100% workload.

Moreover, keeping in mind that S&M loads the CPU very heavily, which is not typical of most software applications, we performed the CPU warm-up tests with two simultaneously running copies of Super PI benchmark, which is very popular among overclockers today. The calculation of the PI value up to 32 million digits takes about 26-27 minutes on a platform like ours.

The thermal protection of the Intel Pentium 4 Prescott processors was again managed with the help of ThrottleWatch tool version 2.02:

The coolers were tested in absolutely equal conditions in two systems types: in an open stand and in a closed system case equipped with a pair of quiet 120-mm case fans. The winter weather hasn’t spared us these days, so the room temperature during our test session was pretty low: only 19^o C. So, we took this temperature value as a reference po0int on our diagrams.

We didn’t have any cooling solution price within the same $30-$32 range at that time, so we decided to compare them against the best representatives of the today’s air cooling systems, such as Thermaltake Big Typhoon ($41) and Zalman CNPS9500 LED ($65). Let’s take a look at the results now.