Looking at our previous overclocking experience with Micron D9PFJ based memory, the ICs perform their best using even primary timings. So to keep things quick and simple, we will narrow our testing procedure down to five simple timing sets while altering the voltage between 1.35 and 1.85V.
Similar to all other D9PFJ memory we tried at even primary timings, the Tracers show close to linear voltage scaling that starts to fade only by around 1.8V. Above that mark you can either force settings to work by going nuts on the voltage or you could raise tRCD and tRP up one notch to extend the range of linear scaling. Using the gentle approach and a lighter stability test, we managed to get our memory to participate in a calculation of 32 million digits of Pi at the following settings:
Sadly, the fun starts to fizz out once the frequency passes the 1067MHz mark. Even though, technically, we could pull a HCI Memtest up to 1100MHz and call it fully stable, minor boggles with normal operation were causing a lot of problems. Since adjusting the subsettings does not improve the situation, we believe it to be a design and/or a compatibility issue. Still, as proven in one of our previous articles, 1000 8-8-8 will show roughly the same performance as 1200 10-12-12, so low frequency numbers shouldn’t upset Ivy Bridge users too much.
Overall, our sample of Tracers shows slightly better overclocking results than the more expensive Elite, but it doesn’t necessarily mean it’s a superior product – most likely we just struck a better sample.
Discuss this article in the forums