“Hey, who's Corsair?” - This is a question that I would hope any PC Builder would know the answer to. Those who can't are clearly living under a rock somewhere in the nether regions of the Australian outback as today the Corsair brand are utterly huge. While the manufacturer's product portfolio has seen huge diversification by offering power supply units, cases, coolers and soon enough premium lager (I kid), their sole speciality was once performance memory.
So long as we have known, Corsair have been ambitious with their products from the days of low latency BH5 and high frequency TCCD DDR memory, to the infamous and highly regarded Micron D9GMH kits on DDR2. Of course, the party didn't stop there as DDR3 has steadily evolved, seeing kits with tighter timings, higher frequencies, lower voltages or all of the above. Today we will be taking a long hard look at one of Corsair's Dominator GT sets for the Socket AM3 platform.
In essence, what we have here is a 4GB dual channel set of memory, sporting a recommended clockspeed of 1600MHz. Of course, clockspeed means little when a kit's latency parameters are about as slack as a benefit fraudster's work ethic...
Thankfully, our sample kit is paired with a very respectable timing configuration of 7-7-7-20. For those who are less clued up on latency parameters, this is very tight indeed. Enough jibber jabber for now, here's a table of specifications to look at.
|Model||2 x 2GB Dominator GT (CMG4GX3M2B1600C7)|
|Intel XMP Profile||No|
|AMD BEMP Profile||Yes|
Active RAM Cooler
Sounds like a product with potential to us. Let's get to the photoshoot...
Packaging & Initial Impressions
The Corsair Dominator GT 1600 comes packaged in a fairly standard plastic outer shell. For those (like me) who regularly need treatment from stab wounds while trying to open packaging such as this, you're in luck as this particular set opens with ease.
This particular lineup of memory sports a stylish black and red colour scheme and tall heatsinks to boot. Be aware that certain heatsinks may not have sufficient clearance to work alongside this memory.
Finally, this is the included active memory cooler. This consists of two 60mm fans, powered by a single 3 pin connector. With four bolts, the unit fixes itself to the RAM using two brackets. The installation itself is very simple although you may have trouble fitting the unit with certain CPU coolers.
Next stop, overclocking!
AMD Phenom II X4 965 Black Edition "C3"
AMD Phenom II X6 1090T Black Edition
ASRock 890GX Extreme3 Motherboard
4GB Corsair Dominator GT DDR3-1600 C7 AM3
Samsung Spinpoint F1 320GB SATA II Hard Disk Drive
XFX Radeon HD 5670 1GB GDDR5 Graphics Card
Samsung 22x DVD+/-RW SATA
Windows 7 Home Premium x64
The Dominator GT set will be compared against memory at DDR3-1600 CAS9 and DDR3-1333 CAS9. Overclocked results will not be included in the comparison for reasons mentioned below.
Our overclocking attempts were involved two targets. First of all, we wanted to determine the maximum stable frequency that can be obtained with the recommended timings of 7-7-7-20. Once determined, we loosened the RAM up and shot for the highest frequency possible.
It must be said however that we encountered serious trouble with our Socket AM3 testbed. Initially we could not overclock our memory so much as a few MHz increments above the nominal Base HTT and no level of memory adjustments would remedy the problem. Thus, it became clear that our Phenom II X4 965's memory controller was to blame. After waiting patiently for a Phenom II X6 1090T to arrive but once again, anything north of DDR3-1640 was a no go. The symptoms were the same across two different AM3 motherboards as well. Two rubbish memory controllers - wonderful.
Suffice to say, we had no more time or patience left to source another AMD processor and so we placed the Dominator GT set into our Core i3/i5 LGA1156 testbed.
After a number of 5-10MHz increments and a couple of memtest86+ runs, we finally found ourselves at a maximum stable 7-7-7-20 1T frequency of 1722MHz. Not a bad effort
For anyone raising an eyebrow right now, I can confirm that the CAS Latency readout of 10 is a misread as both BIOS and Memtest86+ confirm a timing configuration of 7-7-7-20. This required a mild Voltage boost of 1.675V.
Moving on from here, we continued to expedite our memory overclock until we reached our motherboard's stable BCLK limit of 183, for a resulting frequency of 1830MHz. This was achieved with the same voltage of 1.675V and a timing configuration of 8-8-8-24. To this stage, the set of memory was showing little sign of reaching it's limits so when could assert that they could have potentially reached closer to the magic 2000MHz.
Once again, the claimed CAS Latency by CPU-Z is indeed incorrect. A confirmed 8-8-8-24 timing configuration was indeed applied.
On our Core i3 platform, the two overclocks show proportionate increases in bandwidth in SiSoft Sandra's memory performance test. Both overclocked results included Quickpath frequencies of ~2700-3000MHz and as such have not skewed the results in any way. Let's get to the real testing phase...
Synthetic Benchmarks Part 1
SiSoftware Sandra (the System ANalyser, Diagnostic and Reporting Assistant) is an information & diagnostic utility capable of benchmarking the performance of individual components inside a PC.
The CPU arithmetic test ascertains the processor's capabilities in terms of numerical operations. Two subtests named Dhrystone and Whetstone are carried out respectively. This is not a measure of latency and thus higher is better
Well how boring is that? A flat graph...
This is of course a heavily CPU dependent benchmark and so this is very much to be expected. Small gains are seen when comparing the three frequency and timing configurations although once you factor in error margins, these are truly insignificant differences.
The CPU Multimedia Test focuses on CPU based operations that may occur during multimedia based tasks. The magnitude of the score depends on the processor's ability to handle Integer, Float and Double data types.
A similar story can be found here. Again, the nature of this test means that (within reason) RAM does not have a huge bearing on the overall result.
Next we examined Memory Latency...
On an AM3 platform with a Phenom II @ 2000MHz NB, these results are very much what you'd expect for RAM of this calibre. Given that the IMC has been left at nominal frequencies, it is nice to see that ~13.5GB/sec has been achieved.
WinRAR's embedded Benchmark focuses on the processor's File Compression capability.
WinRAR appears to take great advantage of tighter memory timings and higher memory frequencies. The gains shown are much higher than we would have expected.
Synthetic Benchmarks Continued
CPU Queen is based on branch prediction and the misprediction penalties that are involved.
The nature of this particular test yielded very boring results. The operations involved in this benchmark do not have a significant dependency on memory.
PhotoWorxx as the name may suggest tests processors by means of invoking functions that are common to Photo Manipulation including Fill, Flip, Crop, Rotate, Difference and Colour to B&W conversion.
Photoworxx on the other hand seems to love memory bandwith, with a significant jump from 1333C9 to 1600C9. Timing configurations seem to have little effect on it.
This is an integer based benchmark that will test the CPU and Memory by means of the CPU ZLib compression library.
Again like CPU Queen, ZLib is not so dependent on memory.
Everest Ultimate Edition includes read and write benchmarks. Do note that the testing methods vary from SiSoft Sandra's bandwidth test and as such, the results between the two programs cannot be compared directly.
Next up, Futuremark and Games.
PCMark Vantage is Futuremark's flagship "System Wide" benchmark. With a large focus on day to day operations, it's an excellent means of judging the capability of a computer as a whole.
This particular benchmark took a real shine to both memory frequency increases and tighter timings in the aforementioned tests. Aside the peculiar dip in the Memories test with 1600C9 memory, the results are very much as you might expect them.
3DMark Vantage is Futuremarks flagship gaming oriented benchmark at present and is considered to be a demanding one at that. Our tests were carried out under the "Performance" prefix.
As the name suggests, it is a GPU dependent game and as such, the (lack of) difference in scores with varied memory configurations are a testament to that.
DiRT2 is a very recent race driving game, known for it's Direct X 11 support. Let's crank up the settings and give it a whirl...
Call of Duty - Modern Warfare 2
Modern Warfare 2 is among the most popular games available at present. With plenty of explosions and densely (polygon) populated maps, it should prove to be an interesting test for our test setup.
Microsoft Flight Simulator X
Flight Simulator X remains to be a terribly demanding game for it's age. Known for being very demanding on the CPU but also requiring a level of GPU power in the process, we thought it'd be interesting to see how it faired.
None of our games really showed dramatic increases in performance between the three configurations (if at all in some cases), indicating that under typical gaming conditions, the speed of memory doesn't make a significant impact on frames per second. Let's tie this up shall we?
Given the results, we're happy to report that the Dominator GT DDR3-1600 is a fantastic set of memory. It accompanied our Phenom II X6 1090T perfectly, showing a small but notable lead over cheaper CAS9 1600MHz and 1333MHz sets. It's tall heatsinks serve their purpose and the additional active cooler (while loud) ensures that your new modules never catch a fever.
However I do believe that it's important to mention our AM3 overclocking woes once again. Even though we proved that Dominator GT RAM are capable at much higher frequencies, this was not possible on our AMD testbed. While our sample processors aren't exactly the sharpest tools in the shed, there are many in circulation that perform similarly. As such you will want to be certain that your Phenom II processor has a strong memory controller before investing in premium RAM, which will otherwise be wasted.
Our results showed that at the same frequency, a CAS9 set is just a smidgen behind our CAS7 Dominator GT's, and it's possible to find these for almost £60 less. Ultimately our conclusion returns to the reasons behind investing in a Socket AM3 system in the first place. Given the value for money nature of Socket AM3, it really is a contradiction to invest in expensive memory that (unless overclocked) will only offer a fractional performance difference over cheaper sets. Thinking laterally, the £60 premium invested in Dominator GTs could go towards things that may offer a more noticeable difference in performance; choosing Core i7 over Phenom II, a Radeon HD 5850 over a HD 5830, perhaps another HDD for RAID0 or a new sound card?
As far as Corsair and the Dominator GT is concerned, kudos are in order for developing this set of memory. As our Core i3 testbed indicates, if your processor's memory controller is up to the task then there are performance gains to be had. If it's fast RAM that you want, then this could be it.
- Overclocks well
- Optional cooler is effective
- Excellent styling & aesthetics.
- £60 premium over non Dominator GT equivalent
- Optional cooler is far from quiet.
Thanks to Corsair for the sample today, you can discuss this review in our forums.