Super Talent Project X 2x1GB DDR3 PC3-14400 Review

Before we actually get to some stability testing and overclocking, let's discuss exactly what our stability testing consists of and a few thoughts on stability testing for memory in general.

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 simply 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:

• Multiple loops of 3DMark 01 / 06 (30 minutes of looping the full tests each)
• Dual 32M runs of SuperPi Mod 1.5 (ran at the same time)
• 2 hours of dual Prime95 using Prime95 v25.4 Stress Testing Blend
• 2 hours of dual HCI MemTest Pro in Windows using 850MB/instance

The above testing is quite exhaustive and has been found to provide us with a system that is stable and willing to run a distributed computing program 24/7 or hours of gaming without interruption or failure. This method of stability testing will not only be used for testing the stability of the overclocks but also the rated specifications outlined by the manufacturer. We have added the Specification Stability Testing section in response to finding a few kits of DDR3 memory not being able to stably run the specified frequency/timings at the rated voltage.

One disclaimer for a lot of DDR3 memory is that manufacturers tend to test their modules on one specific motherboard and therefore cannot guarantee that any given module will run the specifications outlined on every system. This practice is obviously not ideal and many feel that if manufacturers cannot guarantee a module for the entire range of a given chipset, then perhaps the specifications should be lowered. With that said, let's see if the PC3-14400 Project X kit we have here today can run their rated frequency/timings at the 2.00v outlined by the Super Talent specifications on the Asus P5K3-Dlx. Super Talent has mentioned in the past that they test these modules on the Asus P5K3-Dlx (P35 based), but recently have stated that they test these modules on the Asus P5E3-Dlx (X38 based).


Specification Stability Testing:

With recent advances in DDR3 frequencies and tightening of timings, it came to our attention that a number of DDR3 memory kits were unable to perform specified frequency and timings at the rated voltage. Most memory manufacturers have outlined the setup that their particular modules are tested on before being released to the retail channels. Super Talent does not provide this information directly on their web site for this particular kit of Project X modules when looking at the specifications. If one digs further into their web site, however, you can find in the press release that "These two Project X Kits are built with the fastest available, carefully screened DRAM chips, and are hand tested as a matched pair on a dual channel Asus P5K3 Deluxe motherboard". It would definitely be nice if Super Talent took the time to validate their modules on more motherboards and posted their testing procedure for the modules in a location on the web site that is easy to find.

Here now is the result of many hours adjusting secondary and hidden memory timings with the Project X memory on our Asus P5K3-Dlx at the specified frequency of 900MHz (DDR3-1800) with 7-7-7-21 and 1.95v set in the BIOS. This particular P5K3-Dlx, as do many others, over-volts memory by 0.06v according to a calibrated UEI DM393 multi-meter when reading directly from the VDD output of an open DIMM slot with the system under load.

As we can see from the screenshot, things didn't quite go as planned. As mentioned, many hours were spent testing a great number of combinations of secondary timings available to us through the BIOS or through Memset 3.4 within Windows and at no point were we able to run dual 32M SPi or Prime v25.4. One instance of SPi would continually fail and one thread of Prime would also fail within a few minutes of running. HCI Win Memtest passed two hours with ease and the 3D Mark benchmarks had no problem looping for 30 minutes each. Increasing the voltage to 2.00v in the BIOS allowed the system to run through our stability testing without issue as we will see below but this means the memory was actually receiving 2.06v.

This really isn't that un-common as every motherboard is different and it is quite possible that Super Talent tests their modules in house without taking into consideration the over-volting nature of the P5K3-Dlx vDIMM circuit and thus, passing at 2.00v set in the BIOS like we did. In the end, this isn't the end of the world for these modules, it simply means that they are binned very tight and obviously some samples are right on the edge of specification. Yes it is disappointing but life will go on and the slightest of slight voltage bumps will make this kit completely stable at the specified frequency and timings.

It is now time to see what these modules do at more than just their specified timing sets. The decision to stay close to the specified voltage is an easy one. These DDR3 modules are still very new and it is hard to say exactly how much voltage is going to damage the Micron ICs and there have been no manufacturers offer a module at higher than 2.0v that we have seen. This would indicate that going over 2.0v isn't going to be highly recommend so we have gone up to 2.06v (A = actual) and tested for maximum overclocking ability at three timing sets, 6-6-5-18, 7-7-7-21, and 8-7-6-18. Here is the graph of how far we got:

After finding out that we needed a pinch more voltage to get these modules running at the specified frequency with the tight 7-7-7 timings we weren't sure what was going to happen during the overclocking. Combine that with the fact that the rated voltage is basically all we were willing to put through these modules for 24/7 stability testing, we don't see a whole pile of overclocking headroom. Of course, when the modules are already rated for ridiculously high frequency with tight timings, you can't really expect much.

At 6-6-5 the modules performed quite well running almost up to DDR3-1600 and at 8-7-6, the modules had no problem running into the 950MHz or DDR3-1900 wall that this particular P5K3-Dlx has. This is not the first nor will it be the last kit of memory that runs into the motherboard limitation at 8-7-6 timings. The P35 chipset just isn't up to the task of getting clocks a lot higher than 950MHz but there are some fortunate souls out there with P5K3-Dlx motherboards with 970MHz running and stable. It is a real shame that the Maximus Extreme was not 100% compatible with these modules as the overclocking at 8-7-6 would have been much higher.

You may also notice that the voltage is listed as "actual" or designated like this (A). As described in the Specification Stability Testing section, this particular Asus P5K3-Dlx over-volts memory from what is selected in the BIOS. For every overclock, we have tested the voltage to be 0.06v above what was set in the BIOS. In the screenshots below, Asus AI Suite presents the voltage that is set in the BIOS and the text file we have added, lists the actual measured voltage by the digital multi-meter.