OCZ 2x1GB DDR3 PC3-12800 Platinum EB Review
Share:
Manufacturer’s Part Number: OCZ3P1600EB2GK
Warranty: Lifetime Warranty through OCZ Technology
There have been a lot of headlines plastering the technology news sites in recent months regarding quite a familiar name. OCZ Technology Inc. has been hard at work putting together a complete business profile that now includes one of the most renowned names in power supply manufacturing, PC Power & Cooling. In addition to the PC Power & Cooling acquisition, OCZ has recently added a premier system builder to the newly formed OCZ Technology Group, Hypersonic PC Systems. It seems to have all come so suddenly but OCZ has been cooking this dinner for a while and it is quite exciting to see it happen.
But what about the memory part of their business you ask? Did you think OCZ would let these new acquisitions change what got them where they are today? Most definitely not. It is business as usual in the memory halls at one of the leading SDRAM memory manufacturers. With the gradual movement to DDR3, we are guaranteed to see plenty of new and exciting products coming from OCZ Technology Inc..
On the docket today is a sample from the OCZ DDR3 Platinum shelf in the form of a PC3-12800 Enhanced Bandwidth Edition kit. These modules are rated for operating frequencies up to 800MHz which is DDR3-1600 at, as an added kick, timings of 7-6-6-20 on Asus motherboards. These modules sport the well known XTC heatspreaders and come in a hand tested 2x1GB package. The gavel is in my right hand and ready to pass judgment on these rather impressive sounding modules so let’s get started with the opening statements.
Specifications:
The memory we will be looking at is the 2x1GB version of these modules but as we can see from the part numbers to the lower right, we can get this memory in 1GB or 2 GB sticks as well as dual-channel kits of 2GBs and 4GBs respectively. To the left, the graphic outlines the specifications of the PC3-12800 Platinum EB modules including a rated operating voltage of 1.9v with warranty provided up to 1.95v. This will allow plenty of room for some decent overclocking and should bring some exciting results in that section.
The Enhanced Bandwidth editions of OCZ memory are similar to the standard version of a specific model but with a small adjustment to the rated latencies they can run at the specified frequencies. By decreasing the latency of the memory the bandwidth is increased, hence, the name Enhanced Bandwidth edition. These particular EB modules are rated for 7-6-6 timings on Asus motherboards as opposed to 7-7-7 at the rated operating frequency of the standard modules.
The muscle behind the OCZ Platinum PC3-12800 EB Edition modules is likely going to be Micron D9GTR. The heatspreaders have not been removed as damaging the modules would just be senseless violence which Hardware Canucks does not condone. The thinking behind the fact that they are D9GTR is based on how they ‘act’. This will become clear in the overclocking section later on. A few key specifications of the D9GTR IC is that it is designed for operation at 533MHz with timings of 7-7-7 at 1.5v. The other important feature of D9GTR ICs is that the density of the ICs offers the ability for 1GB modules with only eight banks which means 1GB modules come with only a single side populated. Double sided modules will therefore be 2GB in size and potentially offer high clocking 2GB modules which are difficult to find in the DDR2 world.
Other features of this memory include the proprietary, and easily recognized, platinum-mirrored XTC heatspreader. XTC stands for Xtreme Thermal Convection and has been used on OCZ modules since their introduction around December of ’05 if memory serves me correct. It doesn’t seem like that long ago but I guess two years have passed since the XTC inception. It looks like the critics of the XTC heatspreader were a little off in their prediction that the XTC heatspreader was going to be irrelevant for any amount of time.
The final point worth discussing is the fact that this and every kit of OCZ performance memory is hand tested before packaging ensuring reliability and a high level of quality. If for any reason the modules should fail, OCZ provides one of the most comfortable and easy RMA services there is providing hassle-free life time warranty for their DDR3 memory. The OCZ service isn’t just for when your memory goes bad, their technical support is also top drawer. The OCZ Support Forum is full of helpful guys including Eric, Tony, and other OCZ staffers. There is even a setup guide for this particular memory on the Asus P5K3-Dlx motherboard from Tony that can easily be found in the support forum.
Package:
Some of the evidence presented so far has begun to paint a picture of raw performance, but OCZ also has the ability to combine performance with looks and these OCZ Platinum DDR3 modules are no exception to the tried and tested rule.
The molded plastic blister pack is common place when it comes to packaging memory and for good reason. The package provides ample protection for the product inside. The lack of any distinctive model numbers to the package itself allows for use universally across the entire range of products without incurring additional costs identifying what is inside for each model. This simple yet effect combination of features has led this type of package to be the accepted method for memory delivery and OCZ sees no reason to mess with what is working, for good reason.
The front face of the package provides us visual access to the modules and the label on the memory enables us to know exactly what the product is. The rear of the package simply exposes the back-side of the cardboard insert and provides the end-user with information including some marketing copy, warranty information, and support information. One inclusion that would be nice to see is the address to the OCZ support forum as it is such a great resource for their product users.
Removing the cardboard insert and opening it up at the fold reveals specific installation instructions in five different languages and a couple images to aid in the explanation. Inclusion of instructions in this manner is pretty much the standard for performance memory. There is not a whole lot that can be done with the memory package and there simply is no need for anything more elaborate than what we have seen here. OCZ has done a good job over the years refining their memory package and the professional presentation is more than obvious when looking at this DDR3 kit.
Memory Overview:
Memory is not what you would call an elaborate piece of computer hardware. There is no extensive visible layout of resistors, capacitors, ICs, and other components like we see with a motherboard or graphics card. Instead, all we are presented with to look at is the heatspreader that covers the relatively simple PCB. Despite this lack of complexity, a few photos of the modules are still necessary to discuss a few features before we get into the installation of the memory.
A standard OCZ specifications sticker is present on the left side of each module identifying exactly what type of memory it is including the rated timings and full part number. Notice here how the timings are listed as 7-6-6-24 while on the web site OCZ states these modules are rated for 7-6-6-20. Perhaps a last minute adjustment to the specifications was made after the labels were already produced. The prominent feature of the module exterior is the platinum XTC heatspreader and the large Z located front and center. As mentioned earlier, this design is certainly not new but remains current and fresh providing a very nice esthetic presentation to the memory.
Another feature to note about the OCZ Platinum heatspreader is that there is plenty of access to brush the warm air influenced by the memory away. The top of the module is not enclosed, trapping heat in and because of this, the height of the module including the heatspreader is no more than a standard stick of RAM. The low profile footprint of the XTC heatspreader has led to hassle free installation of OCZ memory for a long time and that trend will certainly continue with these modules.
Earlier in the specifications section it was outlined that the density of the Micron D9GTR ICs allowed for 1GB modules to be produced on a single side of eight banks. Here, we see that in the flesh with only one side of the modules containing ICs. The backside of the memory is going to be pretty much void of any hardware and simply provides a flat surface for the rear heatspreader to be attached to for esthetic reasons. The PCB used with these modules is made by Brain Power which is typical of pretty much all OCZ performance memory since the early DDR days. Brain Power produces high quality PCBs for memory manufacturers to build off of and you will never hear any enthusiasts complain about the quality. These PCBs are labeled B63URCA 0.71 but no further information could be found aside for the fact that OCZ is not the only manufacturer using this PCB for their DDR3.
Memory Installation & Test Setup:
The opening arguments have been heard in their entirety and we will now hear from the prosecutors witnesses. The prosecutor calls “the ease of installation” to the stand.
When looking at the modules a little earlier, it was noted that installation would not be an issue and as the images show above, that statement was 100% true. The XTC heatspreaders really do not add any kind of bulk to the modules at all. It goes without saying, that any motherboard/CPU cooler combo that will fit standard memory modules, will have no problem with this OCZ Platinum memory.
Once the memory was installed, booting with the P5K3-Dlx set to its BIOS defaults was not a problem. With many DDR2 performance memory kits, it had become a known issue that a number of memory modules would not boot with the BIOS defaults because the voltage being supplied to the memory was only 1.8v and a lot of the kits needed 2.0v or more requiring users to try single sticks in various memory slots or having a stick of generic memory simply to POST and raise the voltage in the BIOS for the performance memory. So far, all the DDR3 kits we have tested seem to show no signs of this issue on the P5K3-Dlx. The SPD programming or motherboard may be the reason for this issue seemingly disappearing but it might just be that these Micron DDR3 ICs are more tolerable to low voltage allowing systems to boot.
Either way, upon booting with BIOS defaults set, the memory registers in at 667MHz with 7-7-7-20 2T timings on the 1:2 ratio using a 1333FSB E6850. Obviously manual adjustments are going to be required to get this memory running at even the rated frequencies. This is one of the major drawbacks of high-performance DDR3 memory, overclocking the system bus is required to get the memory to run at full speed because of the highest FSB:RAM ratio being only 1:2. This means that the memory runs at twice the frequency of the FSB so a 1333FSB (333MHz) processor like the E6850 can only run the memory at the fastest frequency of 667MHz (DDR3-1333) without increasing the FSB. As a result, to run DDR3-1600, the FSB has to be running at 400MHz at a minimum.
Obviously with 1600FSB (400MHz) CPUs just around the corner, this issue will disappear for users running these processors because the system FSB will be defaulting at 400MHz providing the ability to run the memory at 800MHz (DDR3-1600) without having to overclock anything.
Test Platform:
Motherboard: Asus P5K3-Deluxe Wifi-AP
Processor: Intel C2D E6850
Processor Cooling: Thermalright Ultra 120 w/AD1212MS-A73GL 2050RPM/80.5CFM
Memory: OCZ Platinum EB 2x1024MB PC3-12800
Power Supply: Silverstone Zeus ST56ZF
Video Card: Biostar 8600GTS 512MB
Additional Fans:
120mm AD1212MS-A73GL 2050RPM/80.5CFM
120mm Vantec Stealth SF12025L 1500RPM/50CFM
Hard Drive: 1 x Seagate 7200.9 80GB SATAII 8MB cache
OS: Windows XP SP2 (with recent updates)
A small disclaimer: The motherboard used for testing the abilities of this OCZ Platinum DDR3 has been modified to provide more voltage to the MCH (Memory Controller Hub), also known as the northbridge. This was required because the P5K3-Dlx used for testing has a limited vMCH setting in the BIOS that restricts voltage to 1.70v. Through extensive testing with this motherboard, it was proven that higher MCH voltages led to much higher memory overclocks. There are many P5K3-Dlx motherboards that clock memory better than this one but keep this information in mind when going over the results. In the screenshots in the following section, the voltages used for each overclock are outlined in a small visible text file.
The only other piece of information worth noting is the use of additional fans. It is the popular belief of many that any and all memory should have some sort of active cooling to remove the heat being generated to not only aid in overclocking but to help keep the memory at safe operating temperatures for a long and healthy life. Heatspreaders aid in removing heat from the ICs but that heat still has to go somewhere and that is where active cooling steps in. At all times during this review, the memory was actively cooled with a 120mm fan overhead. We can now finally get to the best part of memory…overclocking.
Stability Overclocking:
The best but most difficult part of any memory review is the overclocking. There are so many variables to control and manipulate in order to make memory run at the absolute fastest frequency it can. The Asus P5K3-Dlx is an admirable motherboard and with the knowledge of this motherboard we have, maximizing this memory shouldn’t be a problem. First we will discuss the methodology for what is, and isn’t stable.
Overclocking Methodology:
There are so many potential timing sets we could overclock that months would be chewed up just doing the stability testing. Unfortunately time constraints on reviews don’t allow for this so we will have to stick to the three main timing sets that Micron D9GTR seem to run well. After a short period of time, it was evident that we were working with Micron D9GTR based on how it responded to these timing sets and how the memory scaled with voltage.
Memory stability is always something to be debated but the outline below has been found to be 24/7 distributed computing stable in 99% of all setups tested, without having to test each overclock for a week at a time. Distributed computing projects such as Rosetta@Home, Folding@Home, World Community Grid, and more, are the toughest test for stability for a computer. Running 24/7 crunching for one of these great causes is a sure way to find holes in an overclock, unfortunately it takes a considerable amount of time to use them for stability testing so we use the list of programs below to determine stability of an overclock:
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 Orthos Beta on blend mode priority 9
2 hours of dual HCI MemTest Pro in Windows using 850MB/instance
The above testing is quite exhaustive in its own right and as just mentioned, will provide us with an overclock that is stable and willing to run a distributed computing program 24/7. In order for a result to show up in the graph below, it had to pass the outlined stability testing without any errors. The voltages listed in the graph are of that which is set in the BIOS but I have found that the P5K3-Dlx tends to over volt the vDIMM by about 0.05v from what is set in the BIOS. It is hard to say if every P5K3-Dlx does this but it definitely required to mention when testing a kit of memory on it. Without any further delays…let’s look at the results of the OCZ 2x1GB DDR3 PC3-12800 Platinum EB!
Stability Overclocking:
As we can see, all overclocking was done with the 1T setting in the BIOS as the 2T setting seems to provide in-consistent results. Whether the memory is actually running at 1T or 2T command rate is quite debatable but one thing is for certain. 1T tends to over clock slightly lower than 2T but also provides greater bandwidth so keep that in mind when digesting the results above.
These DDR3 modules are really quite impressive offering a wide range of frequencies and timings thanks to the Micron D9GTR ICs that is believed to power them. Both 6-6-5-x and 7-6-6-x scale remarkably well with voltage but we can’t overlook the fact that 8-7-6-x does not seem to scale the same. This is in fact a motherboard chipset limitation with very few Intel P35 motherboards being able to completely stabilize DDR3 over 950MHz. Individual benchmarking is possible much higher but loosening the timings to 8-8-8 or even 9-9-9 does nothing for improving stable overclocking past 950MHz. Increasing vDIMM again, has no effect on stability past this point. Below are just a few of the screenshots displaying the above results:
Benchmark Methodology:
Now that we are fully aware of what this memory is capable of running, we can find out what the performance is like in a few benchmarks at the frequencies we just saw in Stability Testing. As an added bonus, we have decided to throw in results from a DDR2 platform with some high clocked OCZ Reaper PC2-9200 memory. Here is a quick summary of the hardware used on the DDR2 platform:
Motherboard: Asus P5B-Deluxe Wifi-AP
Processor: Intel C2D E6850
Processor Cooling: Thermalright Ultra 120 w/AD1212MS-A73GL 2050RPM/80.5CFM
Memory: OCZ Reaper HPC 2x1024MB PC2-9200
Power Supply: Silverstone Zeus ST56ZF
Video Card: Biostar 8600GTS 512MB
Additional Fans:
120mm AD1212MS-A73GL 2050RPM/80.5CFM
120mm Vantec Stealth SF12025L 1500RPM/50CFM
Hard Drive: 1 x Seagate 7200.9 80GB SATAII 8MB cache
OS: Windows XP SP2 (with recent updates)
This above setup will be shown as the transparent blue bar (2) in all of the benchmark graphs.
The grey bar in the graphs (1) indicates the ‘stock’ timings of the OCZ Platinum EB PC3-12800 memory. The timings had to be set manually in the BIOS and the FSB had to be raised to 400MHz in order to run the rated frequency of DDR3-1600, but the secondary timings were all left on AUTO for these results. We lowered the multiplier to 8X for this setup as well as the DDR2 setup in order to provide results that should be attainable by most C2D based systems.<div style=”float:right;margin:12px;”:> </div> Obviously not everyone will be running at 3.2GHz and some may be running higher, but we figured 3.2GHz was a nice average frequency to run the CPU at.
The last three red results in the graphs (3 / 4 / 5) are of the overclocked settings that we found to be stable previously. Unfortunately there is no way to change the memory frequency without adjusting FSB and CPU frequency so the results aren’t really comparable to each other, but we can use them to see what type of performance is going to be had at the various combinations of CPU/RAM frequencies and timings.
The testing methodology for the results we are about to see is as follows:
a/ Windows is installed using a full format.
b/ Intel Chipset drivers and accessory hardware drivers (audio, network, GPU) are installed followed by a defragment and a reboot.
c/ Programs and games are then installed followed by another defragment.
d/ Windows updates are then completed installing all available updates including .NET Framework followed by a defragment.
e/ Benchmarks are each ran three times after a clean reboot for every iteration of the benchmark, the results are then averaged.
For all results, all settings are set in the BIOS and no adjustments are made in Windows. The same theme, and other system options are identical between the two systems as well as between the overclocked results with the OCZ Platinum EB PC3-12800 kit. The benchmark versions are all listed in the graphs and were the most up to date versions at the time of testing.
The only other settings to note are the Performance Level and Chipset Strap. For all benchmarks, Performance Level was manipulated to be a constant 7 and Chipset Strap remained at 266. The Performance Level setting is for the chipset latencies of the motherboard. A lower Performance Level setting equates to greater performance and conversely, a higher Performance Level value provides lower performance. This setting can greatly influence the results of certain benchmarks and a constant PL was determined to be the best way to go.
We have tried to outline the testing method as best we can here but if there are any further questions about settings or the OS setup, please do not hesitate to ask in the discussion thread which is linked to at the bottom of the review. So without further delay, let’s see how the benchmark witnesses will influence this case.
Memory Benchmarks:
We first take a look at a couple different memory benchmarks that will provide a well rounded image of the type of bandwidth we can expect from this memory at the various overclocks.
According to SiSoft Sandra, the OCZ Platinum EBs are off to a good start showing superiority over the DDR2 setup at the same processor frequency. Everest seems to back Sandra up in all of its tests except the Read bandwidth giving a sizeable advantage to the OCZ Platinum DDR3 at default specifications.
Moving down to Science Mark, we can see the equality between the two setups continues with the DDR2 setup pulling slightly ahead but not enough to draw any conclusion other than that of a dead heat between the two. It is obvious that these two setups are running different chipsets. That can play a role in the comparison but from a general point of view, the OCZ DDR3 Platinum PC3-12800 modules at their rated frequencies appear to be about equal to DDR2 at 600MHz with 5-5-5 timings in bandwidth capabilities. The next piece to this puzzle will be whether or not the latency is as equal.
Unlike what we saw with bandwidth, the latency benchmarks show a small but definitive advantage to the DDR2 setup. Both Everest and Sandra infer the latency of the DDR3 setup at 800MHz 7-6-6-20 is slightly behind that of 600MHz @ 5-5-5-8 of DDR2 flavor. It is hard to say at this point whether or not the small latency difference will equate to any real world advantage.
System Benchmarks:
The next round of benchmarks will look at the system as a whole and then focus in on the memory sub-system. First up is PCMark 05 which tests all of the subsystems of a computer including hard drive, memory, graphics, etc.. Memory doesn’t tend to play a massive role in PCMark 05 results but with the inherent differences between the DDR2 and DDR3 systems, we might get a bit of a surprise. The built in WinRar benchmark can be sensitive to memory timings and frequency so we might see some results identifying this.
As expected, the PCMark 05 scores are all but the same between the two setups at 3200Mhz. WinRAR, however, does seem to show some definite benefits to higher memory frequency. The DDR2-1200 and DDR3-1600 times are almost identical but when compared to the first red result, they are less than 10 seconds behind and over 300MHz of CPU frequency less.
SuperPi is one benchmark that really shows off the power of DDR3 especially in the longer bench of 32 Million digits. There is over a 20 second difference which is quite substantial. Of course, SuperPi doesn’t really relate to a whole lot of real world performance but for those that are into benchmarking SuperPi, DDR3 is a must have as the advantage is pretty substantial. The last of the benchmarks we will look at deal with 3D performance.
3D/Gaming Benchmarks:
Gaming benchmarks generally show very little impact from memory adjustments from past experience. With that said, we will still take a quick look at the benchmark results from the full Futuremark 3D series as well as a couple games.
In all four 3DMark benches, the DDR3 setup outperforms the DDR2-1200 system but like in all other benchmarks, the margin of victory is thinner than Calista Flockhart with the flu. In all benchmarks we have looked at today, any and all differences between the two setups with the CPU at 3200MHz could simply be from the fact that we are running two separate motherboards. The 3D results are no different including in Half-Life 2.
As mentioned, the trend continued with the gaming benchmarks that has be present throughout testing today. The retail version of Crysis also happened to make its way into this review by an afternoon but as we can see from the results, even at 1024×768 with all detail levels on low, the 8600GTS 512MB is still a bottleneck…what a beast of a game. That about wraps up all of the evidence presentation in this case, stick around a couple more minutes for the judgment.
10 – Conclusion:
We have seen a lot of information here today and it is quite evident that DDR3 memory is ready for the big show and the OCZ Platinum EB PC3-12800 kit is a very formidable pair of DDR3 sticks. It is almost a certainty that Micron D9GTR are under the heatspreaders simply based on the overclocking as there is no other IC capable of the frequencies we saw. With that said, this particular OCZ kit is a solid offering providing all kinds of overclocking headroom and very impressive clocking ability at looser timings.
Regardless of what is used to make these modules, the performance speaks for itself. Even at the stock timings and frequency, this DDR3 kit goes head to head with a highly clocked kit of DDR2 and equals or outperforms it in everything we tested. This leads to the ultimate question, does the price justify the performance? It is no secret that DDR2 prices are ridiculously low and DDR3 the exact opposite.
What it really comes down to is where you are in your upgrade path. Building new from scratch? Then DDR3 certainly is a viable option. You will spend a little more now but be covered for future upgrades down the road. If you are simply upgrading a motherboard and already have a set of DDR2, it is really a tough decision to pony up the pennies but if you want to have the best then a kit of DDR3 will have to be on your shopping list.
Pros:
Typical Micron D9GTR-like performance
A wide range of overclocking capabilities
Will fit any motherboard/CPU cooler combination
Backed by OCZs service and support
Cons:
$$$
Cost
And did I mention the price?
Warrantied voltage isn’t enough…but then again, it never is
We like to feedback here at Hardware Canucks so please feel free to bring up any questions or comments in the: Discussion thread for OCZ 2x1GB DDR3 PC3-12800 Platinum EB
