Corsair Dominator 8500C5DF looks impressive already at a first glance but through the new cooler with an external fan module there should be more than what just the eyes can see. We decided check through some extensive testing.
At the end of 2006, the industry foretold that there would be a general increase in the DRAM manufacturing, because of more manufacturers in the market and increased demand, not completely unrelated to the launch of Windows Vista. For the European consumers, this has led to a bigger range of manufacturers to choose between, as well as more competing prices. Corsair is hardly a new manufacturer, and not quite unknown. It’s on the contrary, actually. Corsair was one of the first manufacturers to enter the enthusiast market by releasing memory intended for higher frequencies than the ones in the IC circuits’ original specifications. Lately, we’ve been seeing new innovations from Corsair like load indicators, programmable LEDs and more advanced heatsinks when it comes to memory, but also additional products such as flash drives, power supply units and water cooling devices.
Today we’ll be taking a look at one of Corsair’s memory kits in the Dominator series; Corsair XMS2 TWIN2X2048-8500C5DF. We’re talking about two DDR2 modules of 1 GB each, specified to run in PC2-8500 speeds. Whether the modules can handle this or not, and at which memory latencies, is something we’ll be studying in this article.
Let’s begin by deciphering the long product code.
Corsair XMS2 Dominator 8500C5DF – Specifications
|Corsair XMS2 Dominator 8500C5DF
|DDR2, 240pin DIMM
|533MHz, 1066DDR2, PC8500
|Enhanced Performance Profile (EPP)
DHX (Dual-Path Heat Xchange) heatsinks
We’ve listed the specifications of the memories and we can see that the most vital information is found in the product code.
Twin2X means that we’re talking about two memory modules, 2048 means that the amount of memory is 2048MB, 8500 is the PC-rated speed, C5 is for Cas5, D is short for Dominator and last but not least we have F, which reveals that the memories are paired with a fan. We have discussed how it’s a good idea to cool today’s high-end memories, and it is starting to feel like a necessity during high voltages and overclocking. Corsair was one of the first manufacturers to use sturdy heatspreaders for the memories, which earlier consisted of a thin metal plate, tops. As you can see on the pictures above Corsair hasn’t settled for just removing the heat from the memory circuits, but also from the circuit board that the memory circuits are seated on. We will look into this later on.
Let’s take a closer look at the memories.
At a closer look we can see that Corsair have taken one step
further when it comes to design. A tight work with the aluminum
coolers and the black PCB gives a good quality impression.
The included fan device is of the same good quality. As a
base you have a reliable aluminum profile, quality fans, metal
screws and stylish braided sleeving.
Corsair specifies that Micron’s
D-revision memory circuits are used with these units. Confirmed
overclockers knows that there is a quantity of speed rankings within
this group, and that other manufacturers sometimes specifies more
precise which circuits are used. The reason for Corsair’s lack
of information is because they perform an internal test procedure for
all circuits. This so called screening or speed-bining is more
detailed then the manufacturers own tests.
Let’s take a glance on the test system before we start the tests.
|Asus P5B Deluxe
|Intel Core 2 Extreme X6800
|Corsair Dominator 8500C5DF (2048MB)
|nvidia GeForce 7900GT
|Silverstone Zeus 850W
|Windows XP (SP2)
|Intel Chipset Driver 126.96.36.1991
nvidia Forceware 93.71
Everest Ultimate 2.80
As previous memory tests we use a Asus P5B Deluxe based on
Intel’s P965-chipset. At the frequency tests we have used the highest
possible memory multiplier to minimize the influence in overclocking
ability with northbridge or CPU.
To be able to give a wide picture of the overclocking ability for
the modules we have modified the mainboard to be able to give the
memory more voltage. As a stability tester during the overclocking
moments we have chosen to use MemTest 1.65+, test #5, and the
requirement of 5 flawless loops. Since motherboards measure the voltage
with some margin for error we measure the exact voltage with an
adjacent multimeter. Then the voltage has been adjusted within a margin of ±5mV from
the value given in the diagrams. As cooler we used Corsairs enclosed
Dominator Airflow through the whole test.
Remember that we will test the modules with voltage levels far
over what’s recommended for continuous use and what is covered by any
warranty from Corsair. Even with active cooling voltage levels over
2.4V is not recommended for more than short periods of time.
Let’s go for it and see how the modules do at low latencies.
include_once("/public_html/dia.php"); ?> do_diagram(1799); ?>
When we look at the wide spread voltage register allowed by the
modules we can certainly see that the modules have the same
characteristics as other Micron RevD-based circuits. It takes about
2.4V to reach over 400MHz with these narrow latencies, and thereafter
the scaling slightly decreases towards 460MHz. 2.9V for DDR2-modules
is far from recommended, but here you can see that there is performance to gain with higher voltages.
We move on and test 4-4-4-12.
We can see how the frequency scales really well and that the 4-4-4-12 setting is a lot better for this. As we’ve seen in previous tests higher frequencies are to prefer at many scenarios with today’s platforms. At 2.9V we reach over 600MHz, which is a real achievement for memories specified at 533MHz and 5-5-5-15.
On the next page we take a look at the specified latencies, 5-5-5-15.
At low voltages we’re able to overclock the memories an additional 50-60MHz compared to 4-4-4-12. At more than 2.4V things are starting to slow down and we can’t see any real advantages with our aggressive
voltages. In a previous article we suspected that we were closing in on the limitations of the platform, in this case a P965-based motherboard. Our suspicions are strengthened by this scenario and we will look into this at a later occasion.
Let’s move on and conclude our overclocking data on the next page.
Here we can see the difference in overclocking ability using different memory settings and voltages. The 4-4-4-12 and 5-5-5-15 settings have great advantages frequency-wise, and based on our earlier investigations of the performance of DDR2 memory, higher frequencies usually equals greater performance. For the really heavy overclocking you can take a look at voltages greater than 2.6V. At these voltages, the difference between 4-4-4-12 and 5-5-5-15 is pretty slim on this platform, which makes 4-4-4-12 the preferable setting.
Where high voltages aren’t available, there’s a more clear change in frequencies that can be reached depending on the timings. Here the 3-3-3-8 setting can be an alternative along with a moderately overclocked FSB and 1:1 FSB:RAM ratio, or 4-4-4-12 with 4:5 as the FSB:RAM ratio. With Intel’s platforms you are more limited by few memory ratios compared to AMD’s and NVIDIA’s platforms, but if you use one of the latter ones, you can make more optimal use of the memory performance.
We will summarize our experiences on the next page.
Performance and stability
As we said earlier, Corsair is one of the biggest manufacturers of system memory for computers right now, which leads to most motherboard manufacturers using Corsair memory for compatibility tests. We’ve already experienced the results of these tests several times with early motherboards, where Corsair has simply been the only memory that has worked without hassle. Stability-wise, we haven’t experienced any problems whatsoever during our time testing this memory at the specified speeds.
Thanks to Micron’s memory circuits, there’s a lot of headroom for fiddling with memory settings to find a setting that fits the system’s configuration, like FSB and CPU speed. If you can access high voltages, there’s a lot of potential for overclocking these circuits far beyond their specified speeds. During our tests, we’ve thought about OCZ’s corresponding memory, mostly about the cooling for it. We felt that Corsair’s memory responded better to higher voltages, which gives us the conclusion that they are better at dissipating heat. The included fan makes it possible to have a permanent cooling for the memory, as well as the possibilities of using higher voltages and frequencies. Finally, we want to point out that the frequencies that we’ve tested are MemTest stable and that it is highly probable that you could push it even further as long as you’re using less demanding applications.
To sum it all up, you could say that Corsair is a very safe way to go when it comes to compatibility between different motherboard manufacturers. The overclocking potential is undeniably there and if you can access high voltages, you can reach both high frequencies and fast latencies. The included fan, which can also be purchased separately and used for most other memory models and brands, is of high quality and functionality. The price of this kit in particular is $370 at the moment, but it doesn’t seem to be updated in all stores as the exact same memory (8500C5D) costs about $290 separately and the fan costs about $20, which adds up to $310 – a few tens of dollars above the corresponding memory from other manufacturers.
We would like to thank Corsair for sending this memory kit for evaluation.