So what about an application based benchmark? I’ll start with Winstone 2001. What it does is profile two different types of system usage with multiple windows open at one time. Both Business and Content Creation test system performance by running actual programs while doing normal operations with real data via a script. Business Winstone includes business applications, Content Creation Winstone programs are more what you’d expect to use if designing a web site.
Note that I added an overclocking speed of 142MHz (picked from available speeds of up to 146MHz so that I could keep the actual CPU speed close to the other test). Also note it didn’t fair too well either.
The BIOS has a setting on the ‘Combo’ page of System performance ‘Maximum’ or ‘Normal’. Maximum sets the CAS to 2 and speed to Fast, disables the USB and integrated AC97 sound. Normal allows you to set the CAS and timing settings manually or by SPD and enables USB and sound (which I disabled for all tests).
All results show a nice progressively faster score, except the Content Creation using PC1600, 200MHz FSB and the BIOS ‘Max’ setting. By the looks of it I’d say it is the CAS setting in combination with the 200MHz FSB that gives that one odd result.
Content Creation has about a 10% difference between the highest and lowest scores, but there is only about a 6% difference between the highest and lowest for Business Winstone. So, if Content Creation better profiles your system use than Business Winstone, you would see more benefit from a faster FSB and memory speeds, along with more aggressive BIOS settings.
SYSmark 2000 is similar to Winstone, except it only runs one program at a time. The mix of programs is a bit different, but the concept is the same: Benchmark actual programs doing what they normally do. I don’t know about you, but I usually have six to ten windows open at a time, so Winstone will better profile my use. However, SYSmark is better for looking at individual programs since only it is running at the time of the test.
Again, from the highest to the lowest score the difference is about 8.5~9%. No real surprises here, a bit more predictable then Winstone scores were though – maybe due to the difference of only one program open at a time? If so, maybe when running more programs at a time is a type of usage that DDR SDRAM will work at its best. Also, PC2100 with a 266MHz FSB along with more aggressive timings will show an advantage over PC1600 and a 200MHz FSB.
Be the first to discuss this article!