I like Winstone as a good benchmark of overall system performance. I’ve used it for years and feel comfortable with what it does along being able to understand the results, so lets take a look at it first:
Pretty impressive. I use 10% as a guide to being able to tell the difference between two systems and we are approaching 10% in both tests. To be honest, this test alone would make me rule out using the older P3 over the up to date P4. I know some will say that since the 1.5GHz P4 is almost twice as fast as the P3 800 I compared it with, that it should be significantly faster. Some will say it should be even faster yet, but just remember that pricing fits into this picture also and that the two CPUs used are similar in price.
So will the new SYSmark 2001 test give the same results? I’ve not had as much time to “play” with it yet, but I do think I like the applications it uses to test with because they are more up to date.
Anybody that knows me will tell you I’m not easily impressed, but this is REALLY impressive. Overall, the P4 is almost 23% faster than the P3, while in the ICC test it’s over 30% faster! I’d say the P3 is definitely outclassed in this series of tests, no question about it. Now, some may claim the test is rigged and SYSmark 2001 is biased toward the good points of the P4, but I say it’s more of a matter of gearing the tests toward the actual system usage with newer applications.
So lets see about the individual systems using Winstone 99, not a real up to date test but still one of the better ones.
Just like Win98SE test previously the graphics results for the P4 don’t give the scores we might expect. At just over -9% the business Graphics results are surprising to say the least. The FPU test results are interesting as I keep on hearing about a poor performing FPU (not so by these results). CPU is a bit better (and it should be with a CPU that is almost twice as fast!) and the disk scores are just a tad better.
Take a close look at the individual tests just shown, then the applications test suite results shown above – don’t show the same trend do they? This is a good example of why I feel you need to look at overall system performance running real or simulated real applications and not just the individual tests. Notice I’ve not included any pure memory benchmarks, and there is a reason for it. Who cares what the memory benches out to? Isn’t overall system performance a reflection of actual memory performance, and isn’t the overall performance what really matters? I think so.
Be the first to discuss this article!