Introduction

Crucial is a very unique computer company when it comes to memory manufacturers. They are a division of Micron and the only direct reseller of memory modules. All other memory modules sold on the computer market are OEM. That means the company buys memory modules from the various manufacturers and then repackage the modules under their brand name. This gives Crucial a distinct advantage over their competitors as they have much more control over the quality of the parts provided.

Don't expect to find Crucial memory in most computer stores though. Over 95% of their business is conducted through their website. This allows them to provide better prices by removing the retail distribution costs. Many might be concerned that this causes unwanted delays in receiving memory from them, but from personal experience they are very fast. Orders placed by 12pm Mountain are shipped the same day and the 2nd day service rates are very reasonable. Finding the proper memory modules is also very easy with their online Memory Selector.

Crucial was one of the last companies to produce PC3200 memory for the marketplace, but they had specific reasons. Many of the companies released their PC3200 memory prior to the official JEDEC standard for the 400MHz bus speed. As a result, Crucial memory complies with all the DDR 400 MHz bus speeds standards while many other modules that shipped before it were simply overclocked DDR 333MHz chips certified to run at 400 MHz.

Specifications

The memory used in this review is two modules of Crucial 256MB PC3200. This is standard desktop DDR 400MHz memory that can be used with either the NVIDIA nForce2 or Intel 865/875 chipsets. The exact specifications are:

Two memory modules were used for dual-channel memory configuration testing. Similar modules are also available in 128MB and 512MB sizes.

Test Configuration

Testing of the modules was done on an NVIDIA nForce2 platform to allow for dual-memory configurations. The following hardware and software configuration was used:

For comparison purposes, the Crucial PC3200 modules were compared against two high performance Corsair XMS3200-512C2 memory modules. The Corsair modules are actually overclock certified DDR 333MHz modules that use a lower timing setting than the Crucial modules. They are also 512MB DIMMs compared to the Crucial 256MB DIMMs.

Benchmarks

For testing of the memory modules performance, two primary benchmark programs were used: FutureMark's PCMark2002 and SiSoftwar Sandra MAX3 version 2003.7.9.73. Both of these are synthetic benchmarks that have specific routines that test the memory subsystem to give us a good idea as to the performance of the memory. During the testing, we tested the Crucial memory at the default SPD timings as well as the fastest stable timings that the chips could support, 6-2-2-2.5T.

First we look at the results for FutureMark's PCMark 2002. For this benchmark we are only looking at the composite memory score.

Looking at the results from the PCMark2002 memory score, there is less than 1% difference in the scores from the standard timings of the Crucial PC3200 memory and the high performance Corsair XMS modules. The faster memory timings on the Crucial memory boosted performance over the standard timings, but the performance boost is less than half a percentage point.

For more detailed memory test scores, SiSoftware Sandra MAX3's memory bandwidth benchmark was used. Based on information from AnandTech's recent article entitled "Searching for the Memory Holy Grail - Part 1", the tests will be run with the buffering on as well as the buffering off. Buffering is a technique used by the CPU and chipsets to increase the performance of the system during small repetitive memory writes, something common in most benchmarks. By turning off this buffering, benchmarks are able to see more realistic memory performance differences of the subsystem. In addition, we also look at the memory results of using a single memory module versus a dual-channel configuration.

Let's take a look at the buffered results:

Once again, the numbers from the buffered Sandra memory benchmark results show very little difference in the memory bandwidth between the similar configurations. Here the memory timings are showing anywhere from a .3 to .9% difference between the 8-3-3-3T and 6-2-2-2.5T settings. Even more interesting is that the single versus dual-channel configurations show approximately a 2% performance gain. So let's now take a look at the unbuffered results:

Once the buffering is turned off, we see a much greater range of scores between the configurations. In the case of the memory timings, there is still only approximately a 1 to 2.5% difference in the scores. This is still statistically small enough to be unnoticed by the user at all. The difference between the high-performance Corsair memory and the Crucial memory is also only 1%. What the unbuffered test results really show us is how effective the dual-channel memory configurations can be. The test results show a 15 to 19% performance increase through dual-channel that is definitely significant.

Overclocking

A number of people out there try to get as much performance as they possibly can by overclocking their processors and memory. By increasing the clock speed of the parts, they are able to get some extra performance from their parts at no monetary cost. There are of course drawbacks to overclocking parts. Stability is one of the key issues that users must deal with when overclocking, so they try to find the best parts available.

Crucial's parts are not designed or certified for overclocking. As a result, Crucial does not endorse or condone overclocking of their memory modules. User who overclock their Crucial memory modules will void any warranty on those parts. As a result, overclock at your own risk.

For testing of the overclocking of the memory modules, I wanted to find the highest memory clock that would provide stable operation. The problem is the Athlon XP 2500+ is already pushed to the 200 MHz frontside bus to generate a 400 MHz memory clock. Pushing this processor further will result in an unstable system because of the CPU rather than the clock. To get around this, the processors multiplier was reduced down to a 6x setting to reduce the load on the processor. This setting should allow me to test the limits of the memory modules rather than the limits of the CPU.

The testing was done by first increasing the front side bus in 5 MHz increments, increasing the memory bus by 10 Mhz. Once the memory bus had stability problems, I dropped back down 5 MHz and switched to 1 MHz steps. This was done until the final stable speed was located. That final frontside bus speed setting was 218 MHz that generated a 436 MHz memory bus. This is a 9% increasing in the clock speed of the memory, which is respectable but not as high as other high performance memory modules are capable of. Benchmark results showed an 8.8% memory performance increase.

Conclusions

Overall, the Crucial PC3200 memory modules performed very well in all the benchmark tests. While it was still slower than that of the Corsair XMS3200 high-performance memory, the difference in the benchmarks did not show a very significant speed difference. When you factor in the price of the Corsair compared to the Crucial memory modules, the Crucial memory becomes even more attractive. Of course, if you are looking to overclock your memory beyond the 400 MHz mark, the Crucial memory is probably not the best choice. While less expensive than the high-performance memory modules from other companies, it does not have the same level of overclocking potential.

Manufacturer's Site