The ASUS P/I P55T2P4
The first motherboard we have tested is the ASUS P/I P55T2P4. As seen from the previous table, the motherboard will handle up to 12A, which means the K6-III 400 should be supported with no problem. See our How To: Supporting the K6-2 and K6-3 on the ASUS P55T2P4 article for the details on how to set up the board. We compared a K6-2 400MHz and a K6-III 400MHz processor to show the difference in performance that the 256K on-die cache provides for.
The test setup included the P55T2P4 board with 512K onboard cache (BIOS Level 0207), 64MB 60ns EDO memory, Western Digital 1.2GB (PIO 4) HDD, Diamond FireGL 1000 Pro w/ 8MB (BIOS Level 1.54), Windows 98 (release 1), DirectX 6.1 and Diamond 4.10.01.2359 video driver. The BIOS settings were set to ‘Setup Defaults’ and the display properties were set to 1024×768 and 16 bit color. The benchmarks run were Winstone 99, 3DMark99 Max Pro and Rage’s Incoming game benchmark. These were chosen to provide a comparision of business vs. games applications. Any number of games besides Incoming could have been chosen, but this one provides worst, best and average framerates.
|CPU||Setting||Winstone 99||3DMark99 Max Pro|
|3DMark99 Max Pro|
|K6-2 400||6.0 x 66MHz||14.6||4044||827||10.12||31.87||19.89|
|K6-III 400||6.0 x 66MHz||17.6||5036||885||11.47||32.31||22.17|
The improvement in Winstone99 scores with the K6-III is more than 20%, with a respectable 17.6 average. By comparison, we got a 19.6 score using a Celeron 400, Soyo SY-7IZB+, Matrox Millenium 8MB, 128MB of PC133 SDRAM, DirectX 6.1, (with all current BIOS and driver levels).
The 3DMark99 Max Pro results were mixed, with an impressive 24% improvement in the 3DMark score, but only a 7% improvement in the CPUMark. This should be expected, since the K6-III core is essentially the same as the K6-2. As you can see, the Incoming average FPS also improved by about 11%. Note that the Diamond FireGL 1000 Pro card is not a modern 3D accelerator, so the FPS is somewhat lower than most gamers are used to. We believe that this hardware configuration is valid, as most hardcore gamers will have abandoned their 3 year old motherboard long ago in favor of a more current model, while most casual and business users will still be using older components. The reason we used the FireGL card is that the Matrox cards and older Diamond cards could not be used with 3DMark99 Max Pro.
While running each processor, we checked the voltage regulator heat sinks and found that they were hotter running the K6-2 400 than with the K6-III 400. In both cases, they were not too hot to hold a finger on but the K6-2 400 caused them to become uncomfortably hot. The system was kept running for a full 24 hours, so it appears that there should be no long-term problems using the K6-III in most cases. Of course, as always your mileage may vary so if you decide to take the chance, monitor your system very closely for the first few days.
Be the first to discuss this article!