Corsair Dominator 3x2GB PC3-12800 C8 Memory Review

Stability Testing & Overclocking

Our stability testing methods for memory have been discussed at length in the past, so we are going to simply rely on the explanation from previous reviews. Here is a cut and paste job from previous memory reviews.

Memory stability, what constitutes stable? What is not considered stable? These questions get hotly debated in enthusiast forums all over the internet like little brush fires on the fringe of an inferno that play havoc with forest fire crews. Everyone has their own opinion about stability, especially when it comes to memory stability. For some, stable means they can do whatever it is on their computer without it crashing, blue-screening, or restarting; whether that means gaming or just surfing the internet. To this user, stable means simply using the computer as they normally would.

Of course, that is not what we would call stable and we do a lot more thorough testing before labeling a memory frequency/timing stable. One of the toughest types of programs on a system has been found to be distributed computing projects such as Rosetta@Home, Folding@Home, World Community Grid, and more. Running 24/7 crunching for one of these great causes is a sure way to find holes in a system if there is truly some instability, unfortunately it takes a considerable amount of time to use them for stability testing so we use the list of programs below to all but guarantee the system to be 24/7 distributed computing stable:

• 8 threads of 32M SuperPi Mod 1.5 (ran at the same time using Hyper Pi 0.99b)
• 2 hours of 8 threads of Prime95 using Prime95 v25.7 Stress Testing Blend
• 2 hours of HCI MemTest Pro in Windows using all available memory
• Multiple loops of 3DMark Vantage (60 minutes of looping the full test suite)

With the transition to Windows Vista 64-bit and the i7 platform, our stability testing has evolved slightly. SuperPi 32M is ran with 8 threads now and 3DMark Vantage is the 3D testing program of choice over 06/01 of past reviews. What we have found thus far is that the 8 threads of SPi 32M are an even better test for memory stability than in the past with 2 threads on a dual core. We have all but come to the conclusion that if 8 threads of 32M SPi ran at the same time passes, the system will pass all other stability testing. So when you are clocking memory on the i7 powered platform, start with HyperPi and 8 threads of 32M, it really makes testing nice and quick to narrow down to your overall stability. Now let's get to some of that stability testing.

We will continue our stock stability testing section into the i7 world and with this kit, it really was necessary because the EVGA X58 SLI motherboard didn't have XMP compatibility. Getting the memory up to stock frequencies was just a matter of setting the memory ratio to 2:16 and the vDIMM to 1.60v (1.58v under load measured with a DMM). Yes this kit is rated for 1.65v but even at 1.60v we had no problem at stock clocks. We then adjusted timings to best match what the XMP profile provided and we were off. At first boot we got a blue screen on Windows load. Back into the BIOS, a slight nudge in VTT (+200) so that it ran at 1.26v under load measured with a DMM. Here is the screen shot from that testing, which went smooth as butter after the VTT bump.

As we said, this kit definitely had no problem running the stability testing, even with voltage being undervolted by 0.07v from the rated specifications. Clearly we will have some head room with this memory. The slight bump in VTT voltage is 100% expected and will be part of overclocking memory, or simply running memory, on the i7 platform. VTT can also be labeled QPI voltage or UNCORE voltage on some motherboards. This is what the on-die memory controller gets its voltage from so naturally, it will play an important role in memory clocking. Let's move on to the overclocking now and see what we can squeeze out of this kit.

Since this is our first i7 based memory review, overclocking the memory was a bit of a feeling out process as we weren't even sure what ICs these modules had, so we weren't sure what timing sets to work with. For simplicity sake, we stuck to the tried and true 7-7-7, and 8-8-8 timing sets. We also had to decide whether we were going to run the memory over the 'safe' 1.65v outlined by motherboard manufacturers in regards to CPU damage that may be caused from going beyond it. As it turned out, that was a rather easy decision to make, and was actually made for us.

To open, we definitely had some head room with this kit of memory, but we certainly weren't blown away. The biggest factor in the limiting of overclocks was the simple fact that our sample really did not like voltage at all. Anything over 1.65v was not stable, and in most cases, even 1.65v set in the BIOS created instability. On our 8-8-8 clocks, everything was clipping along great, better than great in fact. We started clocking from 800MHz at 1.60v and almost hit 900MHz shortly. After that, we spent hours upon hours banging our head against a wall. Opening up timings, vDIMM, VTT, setting 2T, and every combination of them; we could not get this memory past the 900MHz or DDR3-1800 range. Even for benchmarking stability, 910~915MHz is about the limit for this kit. The key is the fact that it just doesn't like volts. With almost 900MHz achieved at a rock solid and low 1.60v, we certainly can't complain, but we would have loved to go further.

The story is pretty much the same at the 7-7-7 timing set. Up to the 770MHz mark we were still at 1.60v set in BIOS and once we found instability we started nudging voltage. We got a little bit of a response getting up to 786MHz stable at 1.65v but anything over that was unstable. It seems these modules are particularly sensitive to voltage and if we were able to identify the ICs it would like be Elpida based. The question has been posed at the Corsair forums in the appropriate thread and has been since early December with no response so we are left to speculate. The problem is that Corsair changes IC's so often on the same model that what we reviewed today might not even be the same memory that is available in retail channels a week from now. The IC's on memory are what define its clocking characteristics and they could completely change with an IC change. These modules are Rev 2.1, so any modules with newer revisions could have different ICs and our overclocking results are pretty much useless for comparison.

As it turns out, yes, the Corsair Dominator TR3X6G1600C8D do provide decent overhead in the clocking department and allowed us to tighten timings up nicely as well as move to 1T, but the voltage restrictions really prevent this kit from being a benchmarkers memory. For everyday use, however, they are very solid and easy to get up and running in the EVGA motherboard. They also give you plenty of room to move around to find the best BCLK for optimal CPU frequency.

These overclocks, or very similar ones, will be used in the upcoming benchmark section. We have decided to try and compare the 7-7-7 and 8-8-8 clocks at the same CPU and UNCORE frequency so the memory clocks had to be adjusted slightly to accommodate due to memory ratios and BCLKs needed to achieve each.