All of the marketing hype and cool names mean nothing if the product doesn’t live up to the expectations of the buyer, so this is where the TWINX will earn it’s keep.  Our test system was based on the Abit NF7-8X, featuring the nForce2 chipset, AthlonXP 2400+, and a Radeon 9700 Pro.  The memory was installed in a dual channel configuration, since we have already covered the single versus dual channel performance of this board in a previous review.

Our first test was to check for proper detection of the SPD settings on the NF7-8X.  Leaving memory options set to By SPD, the TWINX loaded completely stable into Windows at the rated 2-2-2-6 setting.  With the stock settings working properly, we adjusted the memory options and bus dividers to find the maximum speed these modules would run, and put all of the results into this handy chart below:

The maximum stable speed of DDR 450MHz is quite impressive, but this leads to problems with the asynchronous bus speed associated with the current Athlon CPU’s.  More on this later.  Achieving this speed required a mild bump of DDR voltage to 2.7V, but no instability was noted during hours of testing and demo loops.  Going above 450 caused random lockups, and at 456 MHz, the system would not post, even with the weakest timings available.

Moving on to the performance testing, we begin to see the advantages of the increased memory timings.  The TWINX was tested at the stock SPD settings (2-2-2-6), and with the maximum overlocked speed (2.5-3-3-7), to illustrate the performance boost given by the stricter timing and asynchronous versus synchronous CPU bus.  One important thing to note is actual decrease in performance when running the memory at speeds higher than the CPU front side bus.  This continues to be an issue for Athlons while we await the higher native front side buses.

Good scores in 3DMark2001SE, with the slight decrease in performance when running the asynchronous bus.  The NF7 v1.0 is known to have some problems breaking the 180MHz CPU bus, which was the case with our sample.  Nonetheless, the strict timing of the TWINX delivered higher scores than we originally achieved with this board, which can be seen here.

Moving from synthetic to a real world benchmark, we see similar results.  The tighter timing of the TWINX gives us a slight boost in speed on this extremely demanding simulator.

Jedi Knight’s highly modifed Quake3 engine is my personal favorite of the current benchmarks, given the huge variety of titles that currently run the Quake3 engine.  No changes to the trend here, just good overall performance.

The longer I do this, the less I have come to rely on this benchmark.  These numbers really don’t mean a lot to the average gamer, so look at them, enjoy it, and move on.

After completing the testing phase of the TWINX modules, I can say that I was quite impressed with the strict timings are good overclocking capabilities of this memory.  With the emergence of dual-channel DDR motherboards, the use of matched memory becomes almost a neccessity to ensure maximum speed and stability.  Corsair’s TWINX kit offers a simple out of the box memory kit that takes the guesswork out of the memory solution, delivering awesome performance and overclocking potential.  Checking GoogleGear, there was a slight price premium (aprox. $10) for the TWINX over ordering two 256MB stick individually, but such is the case when looking for the best performance.