Test resultsThe methodology for our tests is quite simple: we attached a probe to both memory sticks of our dual channel setup, and monitored the temperature for both memory modules of the setup when it was idle for 20 minutes, and after a full load of 20 minutes.
The load was applied with a double instance of
Orthos, a dual core version of the better known
Stress Prime 2004, or SP2004 in short. The voltage used while testing the ATi memory kit is 2.15V in bios. The motherboard does overvolt a little, the real voltage was closer to 2.2V.
We tested 4 situations:
With their original heat spreaders
Without heat spreaders
With the HR-07 coolers attached
With the HR-07 coolers attached and 80mm over the top.
Let's find out how efficient the cooler is, and start by turning the case fans down to their minimum, which is about 6V. With the case fans at low, the case temperature is approximately 37°C (after the stress testing). The memory results are split up into two sections, one for each of the memory modules.
Memory module closest to the CPU:
Memory module furthest from the CPU:
As you can see, these are some extraordinary results here. First thing that catches the eye is that with low airflow the standard heat spreaders do not make a difference. On the contrary: the naked memory chips were actually cooler. The explanation for this result is probably that heat is trapped between the heat spreaders, and conductive heat transport is not really taking place in this case.
The HR-07 makes an impressive showing from the start. Even with low airflow temperatures drop at least 6°C and up to 13°C in relation to the memory without heat spreaders or with standard heat spreaders. When we add an 80mm fan, the results are even more impressive: the Thermalright cooler manages to shave off another 2-3°C.
Let's turn up the case fans to 12V now (case temperature is stable around 36°C now), and see what happens:
Memory module closest to the CPU:
Memory module furthest from the CPU:
On the bottom of our graph we've got another outcome now: the differences between "naked" memory and memory with stock heatspreaders is gone, as conductive heat removal is taking place.
But look at the performance of our test candidate! The added airflow makes the HR-07 really shine here, making it perform another 2-3°C better than before. The results with an additional fan placed upon the cooler, improves things even more, but the differences are small when you compare them with the results at low case fan speeds.
As a side note, please take a look at the differences in temperature between the modules that are placed close to the CPU, and those on the far side of the CPU. Because of the placement of the case fans at the rear of the Lian-li PC70 case, the stick closest to the CPU is always the cooler one. Notable exception is the result when we placed a fan onto the HR-07 setup. This placement probably disturbed the entire case airflow, making the stick furthest from the CPU suddenly cooler.
What was the voltage you have been using?
50° isn't a temperature that I think is capable of limiting overclocks on memory but should change if you give the memory the 2.4+ extreme people use.