Print this article

Super Talent Project X 2x1GB PC3-14400


Date
: 01/22/08 – 02:41:07 AM

Author
:

Category
: Memory

Page 1 : Index

Manufacturer:
Super Talent Technology Corp.

Part Number:
W1800UX2GP

Price:
$499.99 USD (Newegg.com)

Well look what we have here, a new year, a new site design for O2, and brand new technology that is so far ahead of the pack that the pack is just a group of riders out for a Sunday cruise on the hill-side. The calender on my desk now reads 2008 and everything I have dated this week has 07 scratched out filled in by 08. Why is it that the human memory has the inability to re-train information like that within a reasonable amount of time? Perhaps it is just my memory? Either way, we are not here to talk about memory…or are we?

When discussing memory, the random access kind, there are a number of companies that resonate through the occupying air above the discussion. One of them is sure to be the US based memory manufacturing power house, Super Talent. This is especially true when the discussion is directed by computer enthusiasts, or the relatively fringe, overclockers. Super Talents foundation is located in San Jose California where their ISO 9001 certified manufacturing facility is the single largest in North America amongst memory module manufacturers. The Super Talent brand, however, stretches world wide with sales offices in Latin America, Germany, Taiwan, and China to name a few. The Super Talent specialty is of course DRAM, and that is what we will be looking at today.

We won't be looking at just any DRAM modules though, no, we will have the distinct pleasure of test driving some of the fastest rated DDR3 memory you can find today. Appropriately being marketed as "Fast and Furious", the Super Talent Project X PC3-14400 7-7-7 modules are some of the fastest certified DDR3 sticks on the planet. The distinctive black heatspreaders not only make this memory stand out from the crowd but provide an excellent cooling solution for DDR3 memory that is rated near the top of the voltage charts for stock frequencies. Super Talent impressed with their last visit to OverclockersOnline.net when we opened 2007 with a DDR2 offering. We can only hope the same holds true for 2008 with this incredible looking DDR3 kit.

Page 2 : Package & Contents

We start off today with a quick look at the package and the modules themselves before we get into some of the specifications of these impressive DDR3 Project X modules.

The package Super Talent uses is a standard plastic blister pack that is not sealed around the edges aside from the three staples we can see here. There have recently been some grumblings about this type of package in an enthusiast forum from a customer who received a package similar to what we see here. The problem was that the memory appeared to have finger prints on the modules leaving the customer to believe his memory wasn't exactly new.

The modules in this package are certainly secure for transportation but as the customer described above, there really is no way to tell if you are the first person to receive the modules. Of course, the finger prints could have simply been from the retailer or even from the factory during packaging but we believe it is time manufacturers started sealing the plastic packages like this with a tamper proof sticker of sorts. The staples seen on this Super Talent package are a start but something more substantial to prevent tampering is clearly being called for by end users.

The cardboard insert provides some marketing copy about the Project X modules and specifically about the heat sinks. We will discuss the heat sinks used on these Project X modules in the specifications page shortly. The rear of the insert is completed with a couple quotes from a pair of mainstream review sites.

Unleashing the modules from the package allows us to fully enjoy the beauty and splendor that is Project X. These heat sinks, while taking a number of cues from specific OCZ and Corsair heat sinks, certainly are unique to the rest of the Super Talent lineup and their heat sinks in the past.

The marketing theme behind the Project X line, as previously mentioned, is "Fast and Furious". With my review sample came this mouse pad sporting a Project X module and the Fast and Furious logo. I do not believe that mouse pads are going to be standard package with the Project X retail modules but be sure to check with the retailer you purchase your Project X modules from as they may have received a couple for promotional purposes.

We can clearly see the aluminum heat sinks are quite substantial. Super Talent claims twice the mass and 106% more surface area than traditional memory heatspreaders leading to superior thermal capabilities. This is all well and good but if the contact with the ICs is poor, it doesn't matter how good the heatsink is. Super Talent has also seemingly taken care of this through use of their "special thermal adhesive".

Through the magic of macro photography, we can get a good look at the thermal adhesive on the modules without having to take the heat sinks off. Super talent has used this same adhesive on other heat sinks in the past and it is likely cured to the heatsink providing excellent thermal transfer. It clearly isn't the same thick spongy thermal tape that traditional heatspreaders have used in the past. The only problem I can see is that the thermal material is not making complete contact with the IC negating some of the heat sinks superior heat dissipating capability.

Super Talent has placed two stickers, one on each side of the modules. One is simply sporting the logo for Project X and serves no purpose aside from branding. The other, however, is quite useful and necessary providing specifications and the model number for the modules. As we saw earlier in the photos of the package, the sticker on the module is the only identifying label on or in the package. This is pretty much standard practice and one that we hope doesn't change. Let's now take a look at the complete specifications of these modules.

Page 3 : Specifications

The Super Talent web site is pretty thin when it comes to details on these modules. They have a chart of the various DDR3 memory they offer but only a few kits have links to PDF files with more features and specifications, the Project X W1800UX2GP kit is not one of them. I have put together a graphic with the specifications that are listed.

We should all be pretty familiar with the main specifications of these modules by now, we have seen them a couple of times. Being rated for operation at 900MHz or DDR3-1800 with 2.0v puts these modules into the upper crust of DDR3 society. There are a couple other brands with modules rated for this operation but not many. The only down-side to such high rated frequencies and tight timings is that there is likely very little overhead to overclock these modules from 'stock'. These modules are rated for voltages up to and including 2.0v and for anyone un-familiar with DDR3, that is already 0.5v above JEDEC specification. This facilitates the need of such an elaborate heatsink.

When Super Talent claims twice the mass of a traditional heatspreader, I am not going to doubt that claim. The weight of these modules is more than memory with standard heatspreaders and with the security fence like peaks, the surface area is also going to be much more than standard heatspreaders. The heat sinks really appear to be up to the task of keeping the ICs cool and at the rated voltage of 2.0v, they will have to.

On the left is what the memory tab of CPU-Z looks like when booting up the system at Defaults with an E6850 and the Super Talent Project X modules. The memory defaults to 1.75v in the Asus P5K3-Dlx and these modules have no problem at all booting up with this kit. On the right, is the SPD programming of these modules and we can see XMP profiles which mean when paired with an Intel X38/X48 based motherboard, the user will have the option of simply selecting the XMP profile and the system will automatically adjust the settings to run the memory at the rated frequency and timings. For P35 users like we are with the Asus P5K3-Dlx in this review, we will manually have to adjust some system settings in order to run DDR3-1800 which requires a 450MHz system bus frequency. Of course, the major drawback of this kit is that in order to run these modules at the rated timings and frequencies, the system is going to have to be overclocked and not all motherboards are going to be capable of running stable at DDR3-1800.

Normally at this point in the specifications page we would take a look at the part# of the PCBs used and the ICs that are mounted to those PCBs. Unfortunately there are no markings on the PCB and the Project X heat sinks are not going to be coming off of these modules in this review. One can speculate as to which ICs are on this memory but it is safe to say it is either Micron D9GTR or the newer Micron D9GTS that has started showing up on some kits from other manufacturers. Again, the only way to find this out for sure is to remove the heat sinks as Super Talent was not able to provide this information and we won't be risking this kit to remove the heat sinks.

Page 4 : Installation

With the added weight, added bulk, and added cooling ability of the Project X heat sinks, comes added installation issues. Normally memory installation is a formality but not with this kit. We will first take a look at how the memory fits with the test motherboard, an Asus P5K3-Dlx.

You will have to pardon the dust on this setup, the maid is on vacation this week. The large Noctua NC-U6 north bridge cooler extends out to the memory slots and all but blocks the first DIMM slot. Only standard width modules fit in the first slot and since the Project X modules are only taller and not much thicker than standard modules, they have no problem fitting in the first slot. When used in DIMM slots side by side, this standard width provides plenty of room for both modules to breathe.

Switching the CPU cooler to the wide spread Thermalright Ultra-120 does not create any problems with these tall heat sinks. The memory slots on this motherboard aren't close enough to the DIMM slots to interfere. Had the memory slots been too close, then yes, this memory would not fit with this cooler or any other CPU cooler with a large wing-span.

We can clearly see that the bottom few rows of fins on the Thermalright Ultra-120 would come into contact with the extended heat sinks of the Project X heat sinks. The Ultra-120 gives clearance for memory modules but only for standard height modules. You will want to keep this in mind when picking a CPU cooler based on your motherboard if you are planning on using these modules. Now, we will quickly drop the modules into the Asus Maximus Extreme for a test fit since I happen to receive it before publication of this review.

Clearly the layout of the P5K3-Dlx and the Maximus Extreme are different because the Project X module does not fit in the first DIMM slot with the Thermalright Ultra-120 in this orientation. The added height of the modules is what is causing the issue as was predicted not more than a few seconds ago.

Sliding the modules over to the white slots alleviates the issues. Now, it is a well known fact that recent Asus motherboards are designed with performance optimizations for slot 2 and 4 which are the two white outside slots so this works out rather well. If, however, you wish to run 4 modules then the Thermalright Ultra-120 will have to be rotated in order to access the first slot with these modules on this motherboard and the Ultra-120 or Ultra-120E. Keep this in mind when picking out your components to match up to the Super Talent Project X memory.

Page 5 : Test Setup & Overclocking

We have two setups listed here because we will have a comparison to DDR2 numbers in the benchmark charts. The thing to keep in mind when looking at the performance results is that the DDR2 setup is based on the Intel P965 chipset while the DDR3 results are on the Intel P35 chipset. The two chipsets are very similar in performance and virtually identical in every way but they are in fact different. Let's first take a look at the DDR2 setup and then the DDR3 setup that will be powering the Super Talent Project X modules.

• DDR2 Bench Setup:

• CPU: Intel C2D E6600 (L629B383)
• CPU Cooling: Thermalright Ultra-120
• MB: Asus P5B-Dlx Wifi-AP Edition (BIOS 1210)
• NB Cooling: Noctua NC-U6
• GPU: Biostar 8600GTS 512MB DDR3 ForceWare 162.18
• PSU: Silverstone Zeus 560W
• HD: Seagate SATAII 80GB 8MB NCQ
• OS: Windows XP SP2 (with all updates)

  Memory:

• Buffalo FireStix 2x1GB PC2-9600

• DDR3 Bench Setup:

• CPU: Intel C2D E6850 (L724A479)
• CPU Cooling: Rosewill RCX-Z775-EX
• MB: Asus P5K3-Dlx Wifi-AP Edition (BIOS 0604)
• NB Cooling: Noctua NC-U6
• SB Cooling: Thermalright HR-05-SLI
• GPU: Biostar 8600GTS 512MB DDR3 ForceWare 162.18
• PSU: Thermaltake TT Toughpower 700W
• HD: Seagate SATAII 80GB 8MB NCQ
• OS: Windows XP SP2 (with all updates)

  Memory:

• OCZ Platinum 2x1GB PC3-10666 7-7-7-20 (OCZ3P13332GK)
• Crucial Ballistix 2x1GB PC3-12800 8-8-8-20 (BL2KIT12864BA1608)
• Super Talent Project X 2x1GB PC3-14400 7-7-7-21

I have outlined my stability testing for the memory overclocking section a number of times now in previous reviews so I will simply cut and paste from one:

My formula for testing the overclocks shown in this review consists of the following list:

• 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 v2.54 on blend mode
• 2 hours of dual HCI MemTest Pro in Windows using 850MB/instance

With any overclocking presented, we have to make a disclaimer that the results we are about to see are for a single sample of this memory. Not every kit of memory will perform this good, and some will perform much better. There are many factors that go into overclocking aside from the memory itself which include motherboard, cooling, and of course the end user. Please keep all of this in mind when looking at the results and use the overclocking results we have gathered here today as a reference to compare with other users results of this same memory. With that said, on with the show…

The results of the overclocking left me with mixed emotions. Despite understanding that these modules are already rated for rather high frequencies and tight timings, I was hoping for a little more headroom than was found. At the stock timings of 7-7-7-21 and the rated voltage of 2.00v set in the BIOS, we can see that a meager 4MHz was all that was stable for an overclock. With a slight bump in voltage, almost 920MHz at 7-7-7-21 was achieved. The 6-6-5 overclocking was very nice almost reaching 800MHz with 2.05v set in the BIOS. At 8-7-6, the modules had no problem reaching the motherboard limit of DDR3-1900 with less than rated voltage. This should be all but expected on all Micron based DDR3 memory kits and the Super Talent Project X lived up to that expectation with ease. There is however one thing that has to be pointed out when dealing with the actual voltages and what is being set in the BIOS.

It is a known fact that the Asus P5K3-Dlx over-volts memory from what is selected in the BIOS. Many users and other review sites have confirmed this information. With that said, I have listed the voltages on the chart as what was set in the BIOS (B) but on average, the motherboard provides 0.05v~0.06v more than what is set in the BIOS. This would indicate that the modules don't run the rated frequencies and timings at the rated voltage. This was consistent with some basic testing on the Asus Maximus Extreme which provides voltage readings that appear to be consistent with digital multi-meter readings taken from the motherboard components. The Project X modules would not load windows at the default timings with 2.00v and were only stable at the rated timings/frequencies with 2.06v. This equates to what we found on the P5K3-Dlx when taking the over-volting into consideration.

When this was discussed with Super Talent and our first set of Project X modules, we were provided with a second set assured that they met their in-house testing for rated frequencies/timings on the Asus P5E3-Dlx. The second set is what we have provided results from today and as we have outlined, do not meet the rated frequencies/timings without higher than rated voltage for the outlined stability testing we perform. This was explained to us by Super Talent as simply being a case of not all motherboards being the same. So keep this in mind when making your purchase decisions as these modules appear to be very tightly binned and may require slightly more than rated voltage in order to run specified timings at the specified frequency.

Here now are a few screenshots of the various overclocks. The voltage shown in AI Suite is what is set in the BIOS so actual voltage is 0.06v higher than shown.

DDR3-1560 6-6-5-15 1T @ 2.00v(BIOS) / 2.06v(real)

DDR3-1839 7-7-7-21 1T @ 2.00v(BIOS) / 2.06v(real)

DDR3-1900 8-7-6-18 1T @ 1.95v(BIOS) / 2.00v(real)

Page 6 : Memory Benchmarks

Since DDR3 is still so new and the performance hasn't really been figured out completely, we will be comparing the Super Talent Project X memory with a high clocked kit of DDR2 at DDR2-1200 with timings of 5-5-5. This comparison is basically putting the best of DDR2 up against this kit of DDR3. We will also be mixing into the benchmarks the results from a lower end DDR3 kit of memory from OCZ Technology.

In the memory key above, we have outlined the different setups that were benchmarked for the charts on this and the next couple of pages. For all of the DDR3 benchmarks, the system settings are as equal as possible providing an even comparison of the performance between the various settings. The column listed as PL in the key is the Performance Level. Performance Level plays an important role in memory performance so it was kept static at 7 throughout testing. Now, let's take a look at some results, starting with memory bandwidth:

SiSoft Sandra Lite XII SP1 – Memory Bandwidth

Lavalys Everest Ultimate 07 v4.00 – Memory Bandwidth

ScienceMark v2.00 – Memory Bandwidth

The memory bandwidth results fall in-line with what we have come to expect for the frequencies and timings run with the Project X modules. I have looked at a handful of DDR3 kits of memory now and the performance at almost all frequencies/timing sets seems to equal or better highly clocked DDR2 in memory bandwidth. This is again the case with even the lower frequency of DDR3-1600 at 6-6-6-21 out-shining DDR2-1200 at 5-5-5-8 in all bandwidth benchmarks.

SiSoft Sandra & Lavalys Everest – Memory Latency

I really have to stop using SiSoft Sandra in reviews because the results are just so different from any other program out there. They change drastically from version to version and certain timings that drastically change the memory sub-system performance don't register any changes in SiSoft benchmarks at times. Either way, The latency results from Everest show a pretty close battle amongst the DDR2 setup and the Project X at DDR3-1600. Given the higher bandwidth we just saw with the DDR3 setups, this close of a latency race gives the overall memory performance win to the Project X DDR3 setups in these synthetic tests.

Page 7 : System Benchmarks

The next dish on the dinner table for us is going to be some overall system performance followed by a couple benchmarks that tend to rely heavily on the memory system for its results. These two programs are of course the WinRAR built in benchmark and SuperPi, particularly the 32M calculation of SuperPi.

The five setups compared are the same as the previous page and will be for the next page of benchmarks as well. It wouldn't make sense otherwise. Let's have a look at PCMark 05 first followed by the WinRAR benchmark up to 500MBs. The default PCMark 05 suite was ran and like all the benchmarks we are looking at today, it was ran three times with the average being posted on the graph below.

Futuremark PCMark 05

WinRAR 500MB Benchmark

The PCMark 05 results skew a little bit off of what we saw with bandwidth which would indicate the overall system differences between 400FSB and 450FSB playing a factor. The WinRAR compression benchmark follows along the lines of the PCMark 05 numbers showing an almost steady decline in times from the top to the bottom of the chart. The most interesting numbers are that of the Project X at DDR3-1600 6-6-6 and DDR3-1800 with the lower frequency going neck and neck with the higher frequency and looser timings. Keep in mind that performance level was left at 7 for all benchmarks and most motherboards should be able to run PL6 or even PL5 for DDR3-1600 so the 6-6-6 results could be improved even more. At the same time, the 7-7-7 settings could run at PL6 as well so these setups should be very similar in performance out in the wild.

SuperPi Mod v/1.5 – 1M

SuperPi Mod v/1.5 – 32M

SuperPi 1M really isn't a good gauge of much more than system speed but if I left it out of the review I would be inundated with e-mails so I will never drop 1M results. The 32M results are more telling of how the memory is effecting system performance. Naturally the 3.8GHz time is much better than the rest and this is a combination of the CPU speed, FSB speed, and of course, running 950MHz on the memory. The comparison between the Project X 7-7-7 and 6-6-6 results is again what I will focus on. To be honest, I am quite surprised to see the 6-6-6 time lower than the 7-7-7 result but each of the three runs were with a couple seconds of each other as were the 7-7-7 results. Clearly, there is something to be said for 6-6-6 timings at 800MHz.

Page 8 : 3D & Gaming Benchmarks

We have seen how the memory performs in bandwidth testing as well as a few memory intensive programs like WinRAR and SuperPi testing. We now briefly turn our attention to the performance of this versatile memory in 3D applications and benchmarks. Unfortunately I am still left with an 8600GTS for testing 3D performance and finding very few new games that show any advantage to system speed at even 1024×768 so we will simply see how the different setups perform in the tried and true Half-Life 2.

We first look at the full spectrum of the 3DMark series from Futuremark.

Futuremark 3DMark 01/03/05/06

There is nothing really significant or unique in the results we see here with the 3DMark series. All of the results are very similar but if we look close, we can see that the higher the memory frequency, the better the scores we get. As marginal as they may be, the gains are still there. The most telling of the suite is 3DMark 01 as it is more system orientated and the DDR3 setups clearly have the advantage over the DDR2 setup.

Gaming Benchmarks – Half-Life 2

Un-like the 3DMark series of benchmarks, HL2 shows a definitive lead in frames per second over all but the highest DDR3 frequencies. This was a bit surprising as it is the first time that the DDR2 setup really showed an advantage. This may be more indicative of the P965 chipset have better performance than the P35 in HL2 but that is hard to tell. We will now wrap this review up in the conclusion with my final thoughts on the Super Talent Project X PC3-14400 memory.

Page 9 : Conclusion

The time has come to try and wrap up another memory experience in a couple of neat and eclectic paragraphs. This is normally a fairly straight forward task but this kit of Super Talent Project X DDR3 memory wasn't exactly straight forward. It was noted in the overclocking section that this was the second kit of this particular memory because the first kit didn't seem to run the rated frequencies at the rated timings with the rated voltage. The second kit was better but with the over-volting of the P5K3-Dlx to consider, it too barely missed ratings and was not 100% stable at the rated frequency/timings/voltage in the Asus Maximus Extreme.

Presuming that the Asus P5K3-Dlx was to blame for the slight instability at rated settings, the memory was tested on the Asus Maximus Extreme which is based on the X38 platform. It too was determined to require 2.06v in order for the rated settings to be completely stable. This Project X kit of memory is rated for extremely fast frequencies with extraordinarily tight timings, but it would appear that Super Talents screening process for these modules isn't quite up to the task.

Despite the slightly less than stellar performance at the rated timings and voltage, this sample of the Super Talent Project X memory excelled both at tighter and looser timing sets with slight bumps in voltage. It easily maxed out the Asus P5K3-Deluxes ability to run memory stable at DDR3-1900 with 8-7-6 timings and did so with a more than acceptable 2.01v as measured by a digital multi-meter. Combined with the exciting 6-6-5 clocking ability, these Project X modules still managed to impress with very reasonable voltages.

Advantages

• Premier cooling heat sinks that look boss

• One of the highest rated DDR3 modules you can buy
• Flexible timing/frequency sets
• DDR3 already spanks DDR2 in pretty much every performance aspect

Disadvantages

• Binned a little tight to rated timings/frequencies/voltages

• It continues to go without saying that price is still very high on DDR3

Overclockers Online would like to thank everyone from Super Talent in helping make this review possible.
As always, feel free to join us in the O2 forums to discuss this memory here