SiSoft Sandra 2004	Source: Sandra

Sandra is designed to test the theoretical power of a complete system and individual components. The numbers taken though are again, purely theoretical and may not represent real world performance.

Sisoft Sandra 2004 Benchmark Results
	Multimedia Benchmark		CPU Benchmark		Memory Benchmark
Processors	Integer SSE2:	Floating-Point SSE2:	Dhrystone SSE2:	Whetstone SSE2:	Integer SSE2:	Float SSE2:
Athlon64 3200+	14980	19780	8394	3171 FPU / 4141 SSE2	3115	3117
Pentium 4 3.2C	14724	23343	9370	2754 FPU / 6163 SSE2	5021	5019
Dual Xeon 3.06 GHz HT Enabled	47055	65392	18652	7714 FPU / 13299 SSE2	2270	2242
Dual Xeon 3.06 GHz HT Disabled	38091	44525	15296	4493 FPU / 8284 SSE2	2266	2276
Single Xeon 3.06 GHz HT Enabled	23524	32883	9397	3847 FPU / 6811 SSE2	2632	2636
Single Xeon 3.06 GHz HT Disabled	19034	22419	7891	2246 FPU / 4140 SSE2	2663	2651
Units:	it/s	it/s	MIPS	MFLOPS	MB/s	MB/s

These are some of the highest scores we've ever seen in this benchmark! Dual Xeon 3.06 GHz chips with HyperThreading really flex their muscles in SiSoft Sandra! Note that with two CPUs, bandwidth is less than with one. Not what I expected, but it almost certainly has to do with the two processors sharing the bandwidth and the load. The additional cache memory undoubtedly helped the Xeon here as it vastly outperformed its Pentium 4 sibling in everything except memory bandwidth, which is unsurprising considering the 3.2C uses the 800Mhz FSB.

Super Pi	Source: SuperPI

SuperPI calculates the number PI to 1 Million digits in this raw number crunching benchmark. The benchmark is fairly diverse and allows the user to change the number of digits of PI that can be calculated from 16 Thousand to 32 Million. The benchmark, which uses 19 iterations in the test, is set 1 Million digits.

Lower numbers denote faster calculation times (seconds), and hence, better performance.

Super Pi (1 million digits) Benchmark Results (Lower is better)
Processors	Seconds	Ranking
Athlon64 3200+	43
Pentium 4 3.2C	45
Dual Xeon 3.06 GHz HT Enabled	51
Dual Xeon 3.06 GHz HT Disabled	51
Single Xeon 3.06 GHz HT Enabled	51
Single Xeon 3.06 GHz HT Disabled	51

While SuperPi is a good measure of pure FPU power of Xeon processors, it can only address one logical processor at any given time regardless of whether HyperThreading or two processors are installed. The Pentium 4 calculates faster due to its higher 800 MHz FSB and higher overall clock speed, while the Athlon 64's differing architecture and overall number crunching power give it the nod.

POVray	Source: FutureMark

The Persistence of Vision Ray Tracer (POV-Ray) is an all round excellent package, but there are two things that particularly make it stand out above the rest of the crowd. Firstly, it's free, and secondly, the source is distributed so you can compile it on virtually any platform.

Lower numbers denote faster calculation times (seconds), and hence, better performance.

POVray Benchmark Results(Lower is better)
Processors	Seconds	Ranking
Athlon64 3200+	145
Pentium 4 3.2C	179
Dual Xeon 3.06 GHz HT Enabled	184
Dual Xeon 3.06 GHz HT Disabled	184
Single Xeon 3.06 GHz HT Enabled	186
Single Xeon 3.06 GHz HT Disabled	185

POVray is similar to SuperPi in that it can address only one physical processor at any given time. This is why the benchmark reports identical times for each test regardless of whether HT is enabled or dual Xeon processors are used. Again the number crunching power of the Athlon 64 carries the day, but the Xeon's acquit themselves well nonetheless.