In our tests, we used the ASUS P4T533-C motherboard which is based on the Intel 850E motherboard. We loaded the test system with 256MB of either PC800 and PC1066 RDRAM. The graphics card we used was MSI's G4Ti4600-VTD while the hard disk was an IBM Deskstar 75GXP DTLA-307020. Unless otherwise mentioned in the charts below, all i850E results were based on PC800 RDRAM memory.
The 845E results were based on the ASUS P4B533-E motherboard with 256MB of DDR266 memory. The memory timings were set to use default specifications (mainly we used what was outlined as standard specifications by JEDEC). CAS Latency was set to 2.5. Graphics and hard disk was the same as the previous setup.
Results based on the AMD Athlon XP 2600+ was performed on an EPoX 8K3A+ motherboard which was based on the VIA KT333 chipset. Again, we loaded the system with 256MB of DDR333 memory using standard specifications (CAS Latency 2.5, Memory interleaving at 4 and 1T SDRAM command rate). Graphics and hard disk drive remained the same.
Results for Athlon XP 2200+ and below was performed on an MSI KT3 Ultra-ARU motherboard which was also based on the VIA KT333 chipset. Memory settings were set similarly with that of the EPoX 8K3A+. Graphics and hard disk drive - again the same.
Results - Winstone & SPECViewPerf
The results for Business Winstone 2001 and Content Creation Winstone 2002 are shown in the charts below.
In Business Winstone 2001, the Pentium 4 2.80GHz took the lead away from the Athlon XP 2600+ by a very small margin. It doesn't look like it will retain that position for long knowing how the Athlon scales easily in this benchmark suite. However, when it comes to Content Creation, the Pentium 4 took the lead further up the scale especially when you pair the 2.8GHz processor with PC1066 RDRAM. Looks like there are still a lot of room for scalability with the Pentium 4 as it prepares itself to move past the 3GHz mark.
In SPECViewPerf, the Pentium 4 scored about 5.5% more with the 2.8GHz processor than its 2.53GHz counterpart. However, SPECViewPerf is more sensitive to memory bandwidth than pure processing and it's evident if you compare both PC800 and PC1066 results. Using PC1066 RDRAM memory, you can boost performance by as much as 13%. Just by changing memory alone, you get twice the performance gain in SPECViewPerf as compared to increasing processor frequency from 2.53GHz to 2.8GHz. Sometimes, upgrading memory will do more good than increasing processor clock speed and here's a classic example.