Sisoft Sandra 2001 |
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 2001 Benchmark Results |
|
Shuttle AV45GTR |
Score |
|
Multimedia Benchmark |
|
1. |
Integer SSE - 100 MHz |
6086 it/s |
1a. |
Integer SSE - 109 MHz |
6556 it/s |
2. |
Floating-Point SSE - 100 MHz |
7345 it/s |
2a. |
Floating-Point SSE - 109 MHz |
7957 it/s |
|
|
|
|
CPU Benchmark |
|
3. |
Dhrystone ALU - 100 MHz |
2965 MIPS |
3a. |
Dhrystone ALU - 109 MHz |
3174 MIPs |
4. |
Whetstone FPU - 100 MHz |
792 (FPU)/1862 (SSE2) MFLOPS |
4a. |
Whetstone FPU - 109 MHz |
867 (FPU)/2022 (SSE2) MFLOPS |
|
Memory Benchmark |
|
|
|
|
5. |
Integer ALU - 100 MHz |
2067 MB/s |
5a. |
Integer ALU - 109 MHz |
2259 MB/s |
6. |
Float FPU - 100 MHz |
2068 MB/s |
6a. |
Float FPU - 109 MHz |
2251 MB/s |
Sandra is a theoretical power tester, as we can see, when the system is
overclocked, performance does increase by quite a substantial margin.
Interesting note, at stock speeds, the P4X266A has about the same memory
bandwidth as the KT266A. It's not surprising since they share the same memory
controller.
3DMark2001 is the latest installment in the 3DMark series
by MadOnion. By combining DirectX8 support with completely new graphics, it
continues to provide good overall system benchmarks. 3DMark2001 has been created
in cooperation with the major 3D accelerator and processor manufacturers to
provide a reliable set of diagnostic tools. The suite demonstrates 3D gaming
performance by using real-world gaming technology to test a system's true
performance abilities. Tests include: DirectX8 Vertex Shaders, Pixel Shaders and
Point Sprites, DOT3 and Environment Mapped Bump Mapping, support for Full Scene
Anti-aliasing and Texture Compression and two game tests using Ipion real-time
physics. Higher 3DMark scores denote better performance.
3DMark 2001 Benchmark Results |
|
Shuttle AV45GTR |
3DMarks |
Ranking |
1. |
100 MHz FSB |
6972 |
|
2. |
109 MHz FSB |
7391 |
|
3DMark2001 has SSE2 code
written into it and the P4 does fairly well. As we can see, the P4 is
begging for higher clock speeds, the marginal increase from both FSB and clock
speed really boost performance!
Return to Castle Wolfenstein is a relatively new
game benchmark. However, RTCW takes up where Quake III left off and continues to
form the basis of the first person shooter system stressing that QIII has become
a hallmark for. Based upon the Quake III engine, RTCS is obviously going to be
quite taxing on even a top end system. Higher numbers denote faster frames per second (FPS), and
hence, better performance.
RTCW (640x480 atdemo6) Benchmark Results |
|
Shuttle AV45GTR |
(FPS) |
Ranking |
1. |
100 MHz FSB |
40 |
|
2. |
109 MHz FSB |
43.2 |
|
RTCW (640x480 atdemo8) Benchmark Results |
|
Shuttle AV45GTR |
(FPS) |
Ranking |
1. |
100 MHz FSB |
118.6 |
|
2. |
109 MHz FSB |
129.5 |
|
RTCW (1024x768 atdemo6) Benchmark Results
|
|
Shuttle AV45GTR |
(FPS) |
Ranking |
1. |
100 MHz FSB |
39.7 |
|
2. |
109 MHz FSB |
42 |
|
RTCW (1024x768 atdemo8) Benchmark
Results |
|
Shuttle AV45GTR |
(FPS) |
Ranking |
1. |
100 MHz FSB |
110.6 |
|
2. |
109 MHz FSB |
121.6 |
|
The atdemo6 demo is very much like the nv15demo in Q3,
all CPU limited and as we can see here. The P4 does better in the atdemo8 demo, that is more of a overall system
benchmark and here, the GeForce3 Ti500 takes over.