Intel Pentium III 1 GHz
The Intel 1 GHz was launched in India with great fanfare in the first half of September, 2000. Though this processor has been around for quite some time, Intel never managed to get this processor in decent quantities out in the market. With Intel recalling its 1.13 GHz processor, the 1 GHz still retains its position as IntelÆs premier product. We received this processor as part of a kit, which included the 1 GHz processor, the Intel VC820 motherboard and 128 MB of RDRAM.
Now, this is some very serious equipment and begs to be run through its paces properly. We set about doing this with a test setup consisting of the kit along with a GeForce DDR 32 MB graphics card, the Sound Blaster Live! Platinum 5.1 and a U8 Seagate 8.4 GB hard drive. We subjected the processor to a variety of tests to examine its raw performance. First on the menu for the Intel 1 GHz was the 3DMark 2000 test. This test was run not to test the graphics subsystem, but for the more important task of getting the CPU 3D marks. The score was surprisingly low, with 3293 as the 3DMark score and 207 as the CPU 3D marks. While comparing it to the platform and the processor that this kit is trying to replace, we got some rather unpleasant results.
When we had tested the Pentium III 700e on the ASUS CUBX using the tried and tested 440BX chipset with the GeForce DDR and a generic PC-100 cas2 128 MB RAM stick, we obtained 5200 3Dmarks and 350 CPU marks. This enormous loss in performance by the Pentium 1 GHz can be definitely attributed to the RDRAMÆs high cas latencies.
Then we moved on to Quake III. It was run in a torture stress loop for an hour as a warm up for the tests ahead. The system showed no signs of fatigue or lag while exiting Quake III. Benchmarks for Quake III in normal mode for the 1 GHz set up were 107.3 fps when running in normal mode, and 62.5 fps while running 1024 x 768 High Quality mode. Compare these to 118 fps and 66 fps that the 700 on the BX chipset gave us. This proves the fact that RDRAM is the serious bottleneck here.
Though theoretically the 133 MHz FSB is not supported by the i440BX chipset, the next best solution to run this processor would be on the i815 platform. However, the i815 board is a Socket 370 board and the Slot 1 would not work with it. It is possible that if you remove the RDRAM and test it with standard PC-100 or PC-133 RAM, you will get better performance. Since, RDRAM was part of the kit, we didnÆt do that replacement, and tested this processor as it is. Another test we ran was the ZD Content Creation Winstone 2000. This test is considered as a very real-world bandwidth and stability oriented test. Think of this benchmark as the æQuaverÆ ZD-equivalent test, which pushes real world programs to their limits on a system level. It scored a very decent 30.5.
Now compare it to the 26.5 that the 700 scored and the difference can be easily understood. Since the 700 is running on the 100 MHz FSB, the system bandwidth is limited, whereas the 1 GHz is running at 133 MHz.
If you look at the PC comparison test in CHIP, October 2000, you will no doubt be reminded of a 733 MHz system on an i815 based system, which scored 29.3 in this same test.
So, it all boils down to the bus speed and RDRAM in the end. Get it running on a 133 MHz FSB without the RDRAM and the 1 GHz would truly blossom.