Print this article

OCZ Platinum PC3-10666 2x1GB


Date
: 09/25/07 – 07:08:48 AM

Author
:

Category
: Memory

Page 1 : Index

Manufacturer:
OCZ Technology Inc.

Price:
$496.85 USD (NCIXUS.com)

The weather outside in many ways is a metaphor for the change that is happening in computer memory right now. Winter is making brief appearances during the overnight hours but is still a little bit away from making the primetime or afternoon time slot. DDR3 memory is definitely out there, making small appearances here and there but still very much on the fringe of cutting edge with its mainstream appearance still a few months, if not more, away.

OCZ Technology always has and always will be on that fringe of bringing new and cutting edge technologies as soon as anyone else. Going back to my winter metaphor, if you will, they are usually the ones that first blow into town for the first frost. OCZ has cemented their placement as one of the industry's best when it comes to memory from DDR right up to DDR3. Excellent technical support through their 1-800 number, e-mail, or even in enthusiast computer forums always makes sure their customers are satisfied.

Today I have the absolute pleasure of taking a look at my very first DDR3 memory kit in the form of the OCZ Technology PC3-10666 2x1GB. This memory kit is certainly not the fastest available right now but also doesn't cost $800. In its infancy, yes, DDR3 is going to cost significantly more than DDR2 memory does right now but this OCZ Platinum PC3-10666 kit is among the best priced DDR3 performance memory you can buy right now. I am anxious to get started so let's jump right in and start by having a quick look at the package and the sticks inside.

Page 2 : Package & Contents

Since DDR2 and DDR3 are so closely related, even DDR is not really much of a change physically from the latter technologies; I was not expecting much of a change in the DDR3 platinum package from previous kits I have had over the years.

The package has shrunk a little bit which is nice to see but for the most part it is the same blister pack / cardboard insert we have all grown to know and love. The clear plastic package allows full disclosure of the modules inside and we can see they are definitely secure as always in the molded plastic package. The design of the insert is fresh and easily identifying this kit as DDR3.

As mentioned, if the sticker on the modules that you can easily see wasn't enough, the definitive 3s throughout the front-side of the insert and the DDR3 logo seen here at the bottom of the insert let us all know what is being offered.

The rear of the package is about the same despite the slightly smaller size with review site logos down one side and brief marketing copy down the middle. The Lifetime Warranty is outlined here as well as brief quotes about DDR3 and the OCZ quality testing. We are also provided the 1-800 number for OCZ technical support which still impresses me after all these years. OCZ still sits at the top of my personal favorite memory manufacturers list for their great tech support and effortless RMAs.

By pulling the cardboard insert out, we can see that it serves more purpose than just marketing eye candy. There are brief installation and handling instructions available in five different languages.

The OCZ PC3-10666 Platinum modules do not look much different at first glance but upon further investigation, there are a few things that have been changed from than the previous generation of DDR2 Platinum memory. The XTC heatspreader is still the default heatsink OCZ is using and the obvious change is the addition of the 3 to the large Z in the middle. The pin count between DDR2 and DDR3 is the same but the key locations are slightly different to avoid any confusion, but other than that, from the outside; these sticks do look quite similar to any number of other DDR2 Platinum kits.

Obviously the sticker on the modules lets us know that they certainly are not DDR2 modules but the label and information it provides has not been altered from previous generations of OCZ Platinum memory either. It is good to see that nothing has changed as the part number and basic specifications are a welcome sight on memory modules.

The XTC heatspreaders have become the face of OCZ memory as they are unique in design and very efficient in function. The heatspreaders secure to the memory IC's with a very thin thermal tape we can't even see in the photo above and the heatspreaders do not extend very far from the memory PCB. This means that these modules should have no issues when it comes time to install them in a motherboard, regardless of the CPU cooler being used.

They may look the same, but knowing that cutting edge DDR3 memory resides in-between the sexy platinum XTC heatspreaders just makes these modules look that much better.

Page 3 : Specifications

As always, we will take a brief look at the specifications of these modules despite having just seen the meat and potatoes of what this kit is all about on the previous page via the sticker on the XTC heatspreaders. First, here is a brief excerpt from the OCZ web site:

OCZ has also published a short but informative FAQ about DDR3 memory that can be viewed here. The biggest things to note are that DDR3 JEDEC standards have voltage set to 1.5v as opposed to 1.8v with DDR2 as well as a few other features such as DDR3 being fully terminated.

We have pretty much covered everything listed above already but it is noteworthy to mention that the memory is rated to run at DDR3-1333 (667MHz) with only 1.8v. OCZ also warranty's the modules up to 1.95v which will hopefully provide some overhead for overclocking.

A quick look at the SPD table of the memory in CPU-Z indicates rather conservative timings at low frequencies which will hopefully allow the memory to boot in all motherboards. As we learned from a lot of performance DDR2 memory, the need for voltage above JEDEC standard leads to some issues booting in motherboards. I checked to see if there were any issues booting from the default BIOS settings on the P5K3-Dlx with this kit and there certainly wasn't.

The OCZ Platinum PC3-10666 memory boots up at DDR3-1070 @ 8-8-8-20 on the 1:2 divider with my E6600. I assumed then that with a 333MHz FSB CPU, the memory would boot up on the 1:2 divider which would put it at the default DDR3-1333 but at the same 8-8-8-20 timings so I set the FSB to 333 and memory frequency manually to DDR3-1333 but left the vDIMM and memory timings to Auto. When the system booted, the Platinum memory was running at DDR3-1333 but with 9-9-9-24 timings and measuring the voltage on the motherboard, vDIMM was at 1.75v. So it is hard to say whether or not there will be issues with a 333FSB processor but I suspect not. If anyone wants to send me an E6850 or the like for testing, feel free to send me an e-mail :).

At this point I would normally pull the heatspreaders from a module to see what ICs are being used but with DDR3 prices being as high as they are, I really do not feel like potentially damaging a perfectly good kit of RAM. These are my only modules at this point and I may need them for future reviews so the heatspreaders will stay on. From a quick part number search, however, I have seen that most people are referring to these modules as Elpida based but no exact part numbers are provided. We can see that the PCB appears to be labeled as B63URCB 0.70 but for the life of me, I cannot find any information about this model number, if it is even a model number.

Page 4 : Installation

I alluded to the installation of the OCZ Platinum memory as being very straight forward and quite simple due to the size of the heatspreaders, or lack thereof I should say, but I am still going to take a few photos of the modules installed in the two DDR3 motherboards I have at my disposal.

The Biostar TP35DS-A7 has a really nice layout and the memory fits like it should without issue. Even with the Thermalright Ultra-120 and its long wingspan, the memory has plenty of clearance in the closest slot and really shows no signs of problems in any orientation on other motherboards.

The same story is told on the P5K3 Deluxe which actually provides so much room for the memory slots that the CPU cooler fins don't even extend out over the memory. Clearly any memory modules are going to have no issues in this motherboard so the OCZ Platinum modules are not going to be an issue. In all honesty, these modules are hardly larger than a module without heatspreaders so I would venture to say that they will fit in every DDR3 capable motherboard without problems.

Page 5 : Overclocking

This is the section that really makes or breaks a kit of memory. After all, all memory at the same frequency and timings will perform the same. What sets a kit apart from the others is the ability, or lack thereof, to overclock…at least for me it is. The following is a complete list of components used for the DDR3 overclocking and benchmarks. Below that is a list of the hardware used for the comparison DDR2 benchmarks.

DDR3 Bench Setup:

DDR2 Bench Setup:

My stability testing consists of a number of programs including the following list.

• Dual 32M runs of SuperPi Mod 1.5 (ran at the same time)

• 2 hours of dual Prime95 using Orthos Beta on blend mode priority 9
• 1 hour of dual MemTest Pro in Windows using 800MB/instance
• Multiple loops of each 3DMark 01 / 03 / 05 / 06

The above outlined stability testing has not failed yet to find stable 24/7 overclocks of memory. It is quite time consuming and rather daunting a voyage for a beginner to embark on but for the experienced memory overclockers reading, you know just how well the above line-up is for memory stability testing and it is quite worth it in the end. The screenshots at the bottom of the overclocking section outline what this memory was capable of with the outlined testing above. The graph just below is indicative of just dual 32M SuperPi run at the same time. Dual 32M SuperPi is not the same stability as the list above but is fairly close and is a quick and easy way to find a good starting point for your memory overclocking.

Dual 32M SuperPi Mod 1.5

We can see that the overclocking of these modules was a bit less than I desired. I was really hoping for quite a bit more headroom based on other reviews I had seen of this memory. I initially did the testing of this kit on the Biostar TP35D3-A7 Deluxe and thought that the poor memory overclocking was due to the motherboard. I proceeded to go out and purchase an Asus P5K3-Deluxe to make sure it wasn't the motherboard and we can see that I clearly wasted my money as the Biostar memory overclocks were virtually the same.

This particular kit of OCZ platinum PC3-10666 memory just doesn't have a lot of headroom, there really isn't much else to say. Loosening timings up to 8-8-8 or even 9-9-9 had little to no gains in frequency and neither did voltage. At around 1.95v (actual), the memory stops responding positively to volts. Here are a couple screenshots of the long term stability testing I did at the best 24/7 overclocks I could manage.

DDR1221 6-6-4-15 2T @ 1.80v (actual voltage)

DDR1408 7-7-6-15 2T @ 1.80v (actual voltage)

We will now do some benchmarking at the stock frequencies and timings as well as overclocked and compare the results with a couple of the common DDR2 overclocks found throughout enthusiast forums.

Page 6 : Memory Benchmarks

Trying to get an accurate comparison between DDR3 and DDR2 was a little bit tricky because I don't have an Asus P5K-Dlx motherboard which would give us a direct P35 chipset to P35 chipset set of results with the only difference being that one board is DDR2 and the other DDR3. Instead, I had to rely on my Asus P5B-Dlx which uses the P965 chipset. Both are very similar in performance so it is still a very good comparison but keep the motherboard differences in mind when going through the results.

The next challenge I faced was getting similarly common memory speeds to run for the benchmarks. Because I needed a FSB of 333MHz in order to run the OCZ Platinum DDR3 at stock frequencies, I had to the run the DDR2 memory at DDR833 instead of DDR800 as I had wanted to. We will also pretend my E6600 is an E6850 today as 333*9 (3.00GHz) is what an E6850 runs at stock. In the overclocked realm of the above chart, I again had trouble matching up the frequencies I wanted because of the limited memory dividers available with the P5K3-Dlx when staying on the 266 strap. I managed to come up with all five sets of benches on the 266 strap both on the P5B-Dlx and the P5K3-Dlx. The performance level was also stationary at 7 throughout benchmarks on both motherboards. This should provide a pretty equal playground to judge performance of the two setups based on the results of the benchmarks.

SiSoft Sandra Lite XII – Memory Bandwidth

Lavalys Everest Ultimate 07 v4.00 – Memory Bandwidth

ScienceMark v2.00 – Memory Bandwidth

All three programs come up with various numbers for read and write bandwidth but the difference between the setups appears to be pretty much the same across the board. All three programs are measuring the overall bandwidth of the DDR3-1350 @ 7-7-6-15 to surpass that of the DDR2-1200 @ 5-5-5-8 with the CPU at 3600MHz. Even at DDR3-1200 6-6-4-15, the OCZ platinum almost keeps up to the DDR2-1200 @ 5-5-5-8. So far, DDR3 is looking quite promising and this entry level performance kit is going toe to toe with a fairly hefty overclock of DDR2-1200 @ 5-5-5-8.

Lavalys Everest / SiSoft Sandra – Memory Latency

Memory latency on the other hand doesn't seem to quite be up to speed yet for DDR3. Both Everest and SiSoft Sandra show a decided advantage to both DDR2 timing sets over the DDR3 setup. I am anxious to see how the better bandwidth but higher latency of the OCZ Platinum DDR3-13333 performs in the rest of the benchmarks.

Page 7 : System Benchmarks

After seeing the bandwidth numbers and how close the stock results were, it will be interesting to see how that converts to performance in a more system wide performance benchmark. Before looking at the results, I have provided the settings key so we know exactly which results belong to which setup.

First up is PCMark 05 which performs a complete suite of tests on the system including memory, CPU, and hard drive tests among others.

FutureMark PCMark 05

It is obvious right away that the memory really doesn't play much a role in the overall PCMark 05 score or the memory is performing pretty much even. There is, however, an almost 100 point advantage to the DDR2 setup at 3.6GHz over both DDR3 timing sets at 3.6GHz. This could be chalked up to the chipset and other motherboard factors as well.

WinRAR 500MB Benchmark

This is where things start to get interesting. The P5B-Dlx setup really seems to pull ahead in the WinRAR benchmark. I simply let the benchmark run until it reached 500MB and recorded the time for each setup. The results at 3.6GHz are separated between the DDR2 and DDR3 setups by a near 30 seconds. What we can also see is that the DDR3-1350 result is a little bit faster than DDR3-1200 at tighter timings. This is a good sign that bandwidth rules all with DDR3, much like it did and still does, later on in the DDR2 life cycle.

SuperPi Mod v/1.5 – 1M

SuperPi Mod v/1.5 – 8M

SuperPi Mod v/1.5 – 32M

The SuperPi results seem to contradict the bandwidth results we saw earlier leaving me to believe that the P965 chipset is a little bit faster at 400FSB / 266Strap than the P35 at 450FSB / 266Strap / 7 Performance Level. Normally, he who has the most bandwidth at the same CPU speed has the best SuperPi time…at least on the same system. When comparing P965 to P35, it is hard to say whether the memory or the chipset is causing the performance difference. Either way, the time differences are minimal and it appears that entry level DDR3 kits are ready to compete with high end DDR2 memory.

What I find most interesting about the SuperPi results is the fact that at 450*8/DDR3-1350 7-7-6-15, the 32M SPi time is only 7 seconds faster than 400*9/DDR3-1200 6-6-4-15. I was fully expecting the higher FSB and higher memory frequency would make more of a difference than the slight timing adjustments. It goes to show that timings on DDR3 are still an important factor.

Page 8 : 3D & Gaming Benchmarks

The last of the benchmark results we are going to look at are the 3D synthetic and gaming benchmarks. I will run the entire suite of the Futuremark series, 3DMark 01 through 06 as well as a few gaming time demo benchmarks. I have chosen a couple of games that tend to lean on the system for performance and one game that usually doesn't show any gains from system performance increases.

FutureMark 3DMark 01/03/05/06

If there ever was such a thing as a dead heat, this is it. In order to find any differences among the results, you have to really be looking, especially between the DDR2-1200 and DDR3-1350 numbers, even in 3DMark 01. It is no real surprise to see similar results through the latter 3DMark 03, 05, and 06 but I thought for sure we would see some advantage to the DDR3 setup in 01. There is a 35 point advantage to the dark grey result but that is hardly conclusive.

Gaming Benchmarks – Half-Life 2 / FarCry / Battlefield 2142

Moving over to the actual gaming benchmarks, we see all is in order until the Half-Life 2 results. The DDR2-1200 result is by far higher than any others. All benchmarks were ran three times and averaged with no exception in HL2. All three runs were around the 280FPS mark so this is no abnormality, HL2 just appears to favor the DDR2 setup. The one expectation that held true is the lack of difference between any of the results in BF2142. That game continues to show no performance advantage to system speed and completely GPU dependant.

Page 9 : Conclusion

We have looked at quite a few numbers and a very nice comparison between the infant DDR3 memory and its experienced older brother DDR2. The bandwidth to be had by DDR3 is simply incredible. Especially when you consider the fact that the PC3-10666 OCZ Platinum memory we tested today is at the bottom of OCZs high-performance offerings. Moving up that ladder will provide much better performance at a lot higher frequencies.

So enough about the DDR2 vs DDR3 comparison, how did these particular PC3-10666 OCZ Platinum modules perform? I can't say I was overly impressed with the overclocking. The P5K3-Dlx has been considered the standard motherboard for DDR3 overclocking and even with it, the overhead of these modules from stock speeds just wasn't there. They did however run a little bit tighter timings and a little bit faster with very little extra voltage which is always a welcome sight.

The overall performance of DDR3, even at these frequencies, is already surpassing that of DDR2. If you are building a new system that you will want to keep current for months to come, you might want to start looking into DDR3 and a kit like this 2x1GB OCZ Platinum PC3-10666. The price is still pretty high for an entry level performance kit, but offsetting the cost of a new motherboard down the road if you buy DDR2 right now, puts things into perspective.

Advantages

• Cutting edge technology available right now

• Voltage requirements to maximize this kit will allow for a long life and full warranty
• DDR3 performance is already surpassing DDR2, even this early in its life-cycle

Disadvantages

• Price for DDR3 right now is still insanely high

• Very little overclocking headroom with this kit, of course, it is luck of the draw
• Not a lot of motherboard choices in the budget range, Biostar being the exception

Overclockers Online would like to thank OCZ Technology for the review opportunity.