We could, of course, have simply tested a pair of MP 1900+ processors against a pair of MP 2000+ processors and been done with it. But as a rule, we like to throw in some sort of wildcard to make things more interesting. Last time, it was a massive processor round-up involving no fewer than two samples each of four different models of processors, tested in both single and dual configurations. This time around, we decided to go another route. We tested a dually rig against a, umm, singly rig. Read on to find the method to our madness.
Is two better than one?
The thing about SMP systems is, they're not always nearly as cool as you might think. To be sure, a properly done dual processor system has some massive horsepower, and AMD has definitely done it right; take a look a the "AMD dually school" section of this article to see what we mean.
But the problem is taking advantage of all that horsepower. Certainly a busy server can use all the processing power that you can throw at it, but what about a workstation? Only some applications (typically very high-end, expensive applications) can take advantage of more than one processor by themselves, and while many users work in more than one program, many don't do so in a way that will benefit from SMP.
This issue has grown more murky with the passage of time. Currently there are only two Athlon SMP chipsets, the 760MP and its successor, the 760MPX. These chipsets were designed for the server market, where stability is much more important than implementing the latest and greatest features. As a result, these chipsets don't have many of the features of their newer single processor counterparts, such as the VIA KT333 and its support for DDR333 SDRAM and ATA/133 hard drives.
Aye, there's the rub. The greater memory bandwidth and other advantages afforded by these newer single CPU chipsets means that it's likely that a dual processor system will be slower than a single processor system at some tasks. For a consumer building his next computer and trying to decide between a single or dual processor configuration, this is a potential dealbreaker.
To help people in such a situation come to a decision, we decided to test our 760MPX system against a brand spanking new KT333 configuration featuring Soyo's new KT333 Dragon Ultra (expect a full review of this board very soon). Just to make things as hard as possible on the 760MPX (hey, it's got two CPUs, stop feeling sorry for it) we tested with 512MB of DDR333 capable of running at CAS 2, and a hard drive with an ATA/133 interface.
With this comparison, we should be able to determine which situations will favor two processors, and which situations will favor the faster memory and hard drive interfaces afforded by the newer single processor chipsets. Of course, we also test the Athlon MP 2000+ against the Athlon MP 1900+, to see what improvements the bump in clock speed brings.