ABIT VP6

Company: ABIT
Product: VP6 Dual Intel Motherboard
Street Price: ~ $165
Complete Motherboard Spec's: ABIT-USA Spec's
Date Reviewed: May 27, 2001
Reviewed By: Jason

Benchmarking:

When I sat down and thought about it, traditional benchmarks really don't apply to this system. Most benchmarking utilities & games are written for single processor systems, and thus wouldn't take advantage of the second processor. However after digging around some on the internet, and just from personal knowledge (I also own a BP6 which is a dual Celeron ABIT board), I managed to come up with several programs that were able to chug away pretty hard at that second processor.

• SiSoft Sandra

• CliBench Mk III

• 3D Studio MAX 4

• Distributed.net RC5 Client

I already know some people are thinking "Hey, what about Q3A, it supports SMP!" Well, my response to that is yes and no. Yes it has an option to enable support for SMP, but the implementation is so poor you really don't see any performance benefit. Besides, not many people are going to buy a dual processor system if their primary goal is to play games. Dual processor boards are made for people who know that they need the second CPU for their applications that support them. That is why I chose to stick to real SMP enabled benchmark programs that truly utilize more than one CPU.

Sisoft's Sandra has been around a long time and has constantly undergone improvements and is an excellent quick reference to see how your system compares to others. These benchmarks were taken at 160FSB (1.12GHz) since that is the fastest I could get the system running.

CPU	Memory	MultiMedia

CliBench is a free benchmark program that was written in C++. It is also one of the few synthetic benchmark programs that support SMP. It is able to test the following:

• Dhrystones - An industry standard benchmark which is an indicator for integer performance

• Whetstones - An industry standard benchmark which is an indicator for floating point performance

• Eight queens problem - A famous algorithm that depends on the latency time of the CPU. (The longer the pipeline, the bigger the latency time)

• Matrix operations - By calculating matrices you'll get a sign for the CPUs level 1 performance

• Number Crunching - Raw integer performance

• Floating point - Raw floating point performance

• Memory throughput - This shows the transfer speed from the CPU into the memory

• The hard disk's throughput and the CPU usage - The transfer speed between the memory and the hard disk. CPU usage tells you the amount of work the CPU has to do transferring this data.

FYI, I tested all but the HD test. For comparison against a single processor system, I used my BE6-II which also has a P3-700. I overclocked both systems to 155FSB (1085MHz) with 512MB RAM for this test, the 3D Studio MAX 4, and the RC5 test.

Pretty obvious the VP6 would come out on top since it has two CPU's. However you will notice that the VP6 numbers aren't exactly double the BE6-II. This is partially due to them being different systems, but also you must remember that with a dual processor board, the two CPU's are having to share the resources of the rest of the system. So depending on what your application uses the most (Memory, CPU, Disk I/O, etc) will affect how much of a realized performance gain.

3D Studio MAX 4 is used for rendering 3D models. It is one of the most popular programs used for such tasks in the business world, and is able to take full advantage of multiple processors, and even using extra computers over a network. I used software mode to isolate out the video card factor, even though I used the same video card in both machines just to be safe. To test each machine's performance, I took two standard scenes that are included with 3DS, rendered them, and recorded the average time it took to render each frame. The lower the time, the better.

3DS is a program that just hogs all of your system, it eats up RAM, CPU usage, and disk space (mostly for a swap file). If you look at the numbers, you will notice that the VP6 was consistently able to render about 42% quicker than the BE6-II. Again, this is due to the two CPU's having to share the rest of the system resources, which is a factor with any multiprocessor system. Still, this shows how efficient the system can handle such a high load and still be able to perform (no doubt 3DS has very good MPS routines too).

Lastly we take a look at Distributed.net's dnetc client. I like to test the RC5 crunching over a long period of time. If you run the benchmark, it will only give you the score for a single CPU, however if you start the program and let it run, it will give you a continuous update on the average keys per second.

The VP6 here shows almost double the performance of the BE6-II. This is because the program is almost completely CPU intensive.

Go To Page 5 To Wrap It All Up -->