Posted: May 6th, 2002
Written by: Tuan Huynh
The Test
We’ll first start off with the beloved SiSoft Sandra CPU Arithmetic benchmark which is a popular benchmark among enthusiasts and companies to use for testing. Here we see the P4 2.53 GHz come out on top in terms of FPU performance. Compared to the Athlon XP2100+ at 133 MHz and 166 MHz FSB, the P4 2.53 GHz is able to outperform it by approximately 18% in terms of FPU performance. Going from a P4 2.4 GHz with either the 400 MHz or 533 MHz FSB, you’ll see that there’s only a minimal 5% increase, as the frequency between the P4 2.4 GHz and 2.53 GHz is only 130 MHz.
When it comes to ALU performance, it appears that all the processors come in within 1% of each other, as increasing the FSB doesn’t really affect ALU performance as much as it affects FPU performance.
Memory bandwidth is where the Pentium 4 shines. When moving up to the 533 MHz FSB, the Rambus running at PC800 obtains a hefty 10% improvement when compared to the 400 MHz FSB. Running the Rambus at PC1066 speeds, the P4’s memory bandwidth just totally tramples the Athlon XP’s (133) memory bandwidth by approximately 40%! Even if the Athlon XP is moved to a 166 MHz (333 DDR) FSB, the bandwidth of DDR333 will never be able to match the massive amount of bandwidth Rambus has. Even when compared to PC800, PC1066 is able to outperform its predecessor at 400 MHz FSB by 25% and at 533 MHz FSB by 15% respectively, the P4 truly shows the performance capabilities of Rambus.
When pitting the P4 2.53 GHz or any P4 in general to against the Athlon XP, the P4 will definitely pull out on top in Internet Content Creation. This is due to the fact that most of the applications used in the Internet Content Creation side of Sysmark 2002 is heavily optimized for SSE2.
Compared to the Athlon XP2100+ (133), the P4 2.53 GHz dominates by approximately 34%, even with the Athlon XP at 166 MHz FSB, the Athlon still falls behind by 30%. The Athlon XP falls behind very in this benchmark as the Athlon XP can only execute SSE instructions but not SSE2 instructions.
When looking at the P4 2.4 (400) and P4 2.4 (533), we see that for some strange reason the 400 MHz outperforms the 533 MHz version. This could be due to the fact that both subsystems are utilizing PC800 memory and Internet Content Creation thrives on memory bandwidth greatly, and if the benchmark was ran enough times, they’d eventually score exactly the same.
Sysmark 2002 Office Productivity is a benchmark that relies heavily on the I/O subsystem as it accesses the hard dsive greatly because office applications are always accessing the hard drive. Here we see the different speed grades of the P4’s not really offering any performance advantages over each other. Compared to the Athlon XP (133), the P4 2.53 GHz is capable of performing up to 11% faster, this is most likely due to the clock frequency rather then the FSB.
Next Page
|
|