Official Hardware Sponsor

Testing Methodology

I got to do something with this review that I really wanted to do with my Thunder K7 review, but didn't have the resources or time for. For this review I am expanding beyond the motherboard that we are actually looking at and comparing multiple dual processing platforms. The Tiger MP just happened to be the motherboard used in the AMD based systems ;-)

I also took this opportunity to really look over our benchmarking suite and methodology here at 2CPU.com. We have been less than thrilled with the amount of relevant benchmarks available for multi processor systems, but there are so few out there. Sandra seems to be getting long in the tooth, and with all of the different versions of it floating around the 'net it it increasingly difficult to get consistent, reproducible results. Now Sandra has been a mainstay here in the labs for a long time now (they actually added the multi processor support for us) and everyone is very familiar with it so it is not going anywhere soon. Just make sure to check versions and whatnot, and also realize what you are looking at is only one benchmark.

I have been a fan of Dr. Damage over at the Tech Report for some time. Without a doubt, the man writes some of the finest, most in-depth reviews I have ever read, and he does it all without boring me to tears! Through reading some of his reviews and a few email exchanges, I decided to add some new benchmarks to the suite. I will describe each benchmark before we get into it, but I will tell you that in the majority of the numbers you will see; Higher is better! I will also make the actual spreadsheet with all of the raw benchmark numbers available at the end of this review. So if the graphs are a little blurry, feel free to grab it.

Test Systems

The systems that I used here for testing presented me with a few problems. The first thing I ran into was that I could not standardize all the systems across the board. Everything except the AMD based system was pre-built by Dell (how much more "standard" can you get?!?) and actually in use where I work. Swapping out video, audio and drives was not an option for me.

The second major problem was the configuration of the machines lent themselves to very specific duties. Two were CAD workstations and one was a server. Now how do you compare those?!? The answer was simple really. I stuck to benchmarks that isolated the three main components in the system; memory, processors and the motherboard/chipset connecting them. Any benchmark that was even remotely video or disk dependent got nixed. What I wound up with is a suite of pretty intensive tests that present a good mix of real world data and synthetic numbers. All benchmarks are free to download, so feel free to test your own system to see how it stacks-up.

Here is the run-down of the five systems compared in these tests (all BIOS settings were set to "optimized defaults" where applicable. No serious tweaking was done to any system.):

SiSoft Sandra >>