Introduction
Features
Features (cont.)
Test System and Methods
Test Bed
Test Suite
3DMark05, PCMark05
Super PI 1.1
ScienceMark 2.0, Sandra 2007
Quake 4
F.E.A.R
Half-Life 2: Episode 1
Call of Duty 2
Oblivion
Multitasking
Overclocking
Conclusion

---

Test Suite

When it comes to choosing the products that will be included in our benchmarking suite, we often have a hard time making the decisions. We don’t simply use the benchmarks that we have used in the past, because they may not be as applicable for the specific component we are trying to test. We have chosen the following tools based on the way the tests are run, the kind of results they generate, and the current interest/demand for them in the community:

• Kanada Lab - Super PI 1.1
• SiSoftware - Sandra Lite 2007
• ScienceMark - ScienceMark 2.0 Benchmark Suite
• Futuremark - PCMark05 CPU Test Suite
• Futuremark - 3DMark05
• Futuremark - 3DMark06
• Id Software - Quake IV
• Monolith - F.E.A.R
• Valve - Half-Life 2: Episode 1
• Infinity Ward - Call of Duty 2
• Bethesda - The Elder Scrolls IV: Oblivion

This list may seem a bit short, and trust us, we wanted to run a whole lot more tests than we did, but due to certain time constraints (Yes, the NDA was pulled in to July 13th from July 27th) we were unable to run every test we had originally planned. We still feel that the above tests produced very clear information about the processors and their performance. Specific reasons as to why each test was chosen will be discussed when the results of those tests are presented.

Tests

3DMark06

Scalability. That is the only real reason we decided to run the 3DMark06 benchmark on the Core 2 Duo processors. The test itself is largely graphics dependent due to the heavy High Dynamic Range rendering and SM3.0 shading that needs to take place in order to receive high scores. However, the test does evaluate your processor and the results of that evaluation are factored into the overall score. Thus, we ran 3DMark06 at stock processor speeds as well as at three differently overclocked configurations to see if substantial clock speed increases would result in significant increases in the score received. (Note: Overclocking methods and more extensive results will be presented later in this review.)



There is kind of a lot of information here, but you can see how each processor has four separate results. Keep in mind that each result was generated from the average score of three tests, so you’re really looking at the results of 24 tests. You can see that both processors scale slightly as expected, considering the 3DMark06 test is not primarily CPU defined. The most interesting result we see here, however, is the relatively low score received by the Core 2 Extreme processor when overclocked to 3.52GHz. This result is attributed to the heavy screen ripping and subsequent graphics driver failure that took place each of the three times the test was run when the processor was clocked at this speed. While we would normally attribute this to the video card, we are clueless as to why the same results were not yielded during the other tests.

To briefly put the high score of 5802 into perspective, comparable results were achieved (Reference: Futuremark ORB) from systems with similar specifications (besides the processor of course) only after heavy overclocking of both the processor and video card had taken place. Said systems also had 2GB of RAM opposed to the 1GB used for our tests. We have not run any extensive tests to determine the effect each component has on 3DMark06 results, but we feel we can make the assumption that they all do indeed have an effect.