Booting the Ballistix Tracer to Windows and using CPU-Z to check the settings, we see that they are SPD programmed for 9-9-9-24-33 at 1.5v at the JEDEC frequency of 666Mhz or DDR3-1333. However Crucial includes an XMP profile of 8-8-8-24-36-2T with 1.65v at 800Mhz or DDR3-1600. AMD systems do not support XMP profiles, so these timings must be manually set in the BIOS, but this is a relatively minor change. The memory easily reaches its rated specifications, and every Ballistix Tracer kit we've tested has also been able to use a 1T command rate as shown. Since this kit is relatively identical to our previous Ballistix Tracer Red review, we're not going to focus on benchmarks this time. However before we were unable to test for operation above 800Mhz due to the motherboard used, now that we're on a newer chipset board let's see what these sticks can do for overclocking.
Using Memtest+ to verify stability, the reference or "bus" speed was slowly increased and test #5 looped until the memory registered an error. This occurred at only 205Mhz, or 820Mhz on the memory, in other words we didn't get very far before we found the ceiling. Fortunately a small bump in voltage from 1.65v to 1.70v allowed us to continue. The memory ran fine at 840Mhz (210Mhz ref) however at 860Mhz we again began seeing errors, and neither 1.75v nor 1.80v would stabilize that speed. Therefore 840Mhz is where we're calling this one, an increase of only 40Mhz over the rated frequency.
To integrate that into our existing system overclock, a reference speed of 250Mhz is used with the default CPU multiplier value of 16x, yielding a 4.0Ghz processor frequency. The HT link multi is decreased one value to 9x for 2250Mhz SB interface, while the NB link multi is increased one value to 11x for 2750Mhz memory interface, which as we noted in previous tests really boosts performance in an AMD system. Using the DDR3-1333 setting or 3:10 ratio results in the memory running at 833Mhz (250 x 10 / 3) with 1.7v at the rated timings of 8-8-8-24-36 and 1T.
Going back to our article on memory latency, we can calculate that an operating frequency of 833Mhz equates to a 1.200ns cycle time, which at CAS8 results in a 9.6ns latency. That's just slightly better than DDR3-1866 CL9, so with a slight voltage increase and a small overclock we're seeing the same benefit as if we'd purchased faster, more expensive memory.
Let's wrap things up with some final thoughts and conclusion.