My testbed consists of the following components housed in a
Lian-Li PC-777B chassis:
Overclocking
A successful overclock requires a sum of all parts working in tandem. Components need to be carefully selected based on their potential and matched appropriately. Without the right ingredients, this venture will quickly turn into a recipe for failure. Taking this into consideration, when reviewing a kit of RAM I want to test how flexible the chips are. The first priority is to find out exactly how high a kit can scale and still boot into Windows. I'm not necessarily looking for stability at this point, as that is going to vary from system to system and also from kit to kit. After this ceiling has been discovered, I then focus on stability as it pertains to the testbed. Three things are looked for:
1) The tightest timings achievable at a CPU's stock 200MHz frequency.
2) The highest stable 1:1 overclock regardless of timings.
3) A middle ground between the two.
Once those three objectives are reached, the RAM is put through a series of benchmarks to find out which setting translates into the highest real world performance. Let's have a look at how Crucial's PC4000 Ballistix Tracers fared:
| 200HT x 11 (2.2GHz) |
2-2-2-5 1T, 2.8V |
| 219HT x 11 (2.4GHz) |
2.5-2-2-6 1T, 2.9V |
| 260HT x 10 (2.6GHz) |
3-3-3-8 1T, 2.95V |
| 300HT x 7 (2.1GHz) |
3-4-4-10 1T, 3.1V, booted but not stable |
MEMORY TESTING
Sisoft Sandra Memory Bandwidth
SiSoft Sandra (System ANalyzer, Diagnostic, and Reporting Assistant) is a 32 and 64-bit Windows system analyzer that includes benchmarking, testing and listing modules. We use Sandra's Memory Bandwidth benchmark "to estimate the maximum throughput of the system."
Everest Ultimate Edition 2006
"EVEREST Ultimate Edition is a system diagnostics and benchmarking solution for PC users, based on the award-winning EVEREST Technology. During system optimizations and tweaking it provides essential system and overclock information, advanced hardware monitoring and diagnostics capabilities to check the effects of the applied settings." For benchmarking RAM, we use Everest's Memory Read and Memory Write tests.
Super Pi Mod v1.4
Super Pi is a popular program used by enthusiasts to benchmark system performance. Owned and maintained by XtremeSystems.org, Super Pi is as much a competition as it is a benchmark, with many forums cataloging user times in the 1M and 32M calculations. For benchmarking RAM we use the 1M calculation.
Sciencemark 2.0
"Science Mark 2.0 is an attempt to put the truth behind benchmarking. In an attempt to model real world demands and performance, SM2 is a suite of high-performance benchmarks that realistically stress system performance without architectural bias." For benchmarking RAM, we use MemBench, an unbiased, synthetic benchmark that sweeps through more than 15 different publicly available memory copy algorithms to measure the peak memory performance of your CPU's caches and your memory subsystem.
Doom 3
Doom 3 is an "older" game that is still a good benchmarking tool as it highly stresses your entire system. For benchmarking RAM, we use TimeDemo1 on high quality with the screen resolution set to 1024x768.
Before jumping to the conclusion, let me take a moment to comment on latencies. There's a common assumption floating around that loosening RAM timings to achieve a higher clockspeed is not worth the tradeoff. This can be true when looking at a small increase in frequency, but as we see in the benchmarks above, abandoning the 2-2-2-5 timings at stock paid performance dividends at each speed bump. Running 3-3-3-8 may not look particularly attractive on paper, but the resulting frequency and increased bandwidth were easily able to compensate.
Warranty & Support
Crucial is highly regarded in the enthusiast community for having some of best customer support around with modules backed by a lifetime warranty. There is a wealth of information to be found on their website for pre-purchase decisions, and for post-purchase issues, support is just an email, phone call, or online chat away.
Our Recommendation
Color me impressed. Not only will these sticks ensure a smooth computing experience in RAM intensive applications, but they've proven to be very flexible as well. It's no small 'feat' for a 2GB kit to fill the shoes of tight timed kits half the size (pun intended), but Crucial has managed to do just that. Running my testbed's 4400+ X2 at stock speed, the Ballistix Tracers held impressive 2-2-2-5 latencies without touching the voltage. A slight bump to 2.9V had my processor happily chugging away at 2.4GHz emulating a 4800+ X2, with latencies still considered tight at 2.5-2-2-6. When increased past the rated DDR500 spec, the Tracers had no trouble holding steady at a frequency of 260MHz, putting my processor at 2.6GHz. For those that have long since gone cross-eyed keeping up with AMD and Intel's nomenclature, that's basically an FX-60 minus the unlocked multiplier!
Unfortunately for socket 939 owners, some DDR1 kits can be a challenge to find these days with AMD prepping a jump to DDR2 later this summer. The 2GB kit reviewed falls under this category, which is a shame after the showing they put on. If you do manage to locate these sticks however, your search will be rewarded whether you intend to run tight at stock or decide to push your processor into higher pricing tiers. The next time someone tells you that size doesn't matter, rest assured they are not packing a kit of Crucial Ballistix Tracers.
