Print this article

OCZ 2x2048MB PC8000 DDR2 (OCZ2P10004GK)


Date
: 03/12/08 – 04:52:07 PM

Author
:

Category
: Memory

Page 1 : Index

Manufacturer:
OCZ Technology Inc.

Price:
$105+ CDN after MIR (Pricebat.ca)

The word on ‘buzz street’ as of late has been DDR3 and all the performance that it brings. Well, that is ‘buzz street’ and today, we are reporting live from reality square where money doesn't grow on trees. In this wonderful land of reality, performance isn't something that can be discarded either. That is the joy of reality. Performance is still a requirement alongside budget and cost effective purchases. As we all know, DDR2 memory prices have dropped like the titanic in the North Atlantic this past year, but what does DDR2 still have to offer us computer enthusiasts aside from value?

In addition to enthusiasts, DDR2 still has a lot of things to offer all computer users. Everyday computer users are really going to want to stick around today because what we are looking at from OCZ Technology isn't just for the benchmarking, overclocking, teenager with too much disposable income. We are taking a good look at the biggest asset DDR2 memory still has to offer the computer public…size, lots of it at an amazing price. Not only can DDR2 offer gobs of system memory with staunch performance, but with the Platinum PC2-8000 2x2GB kit from OCZ, it can be had for pretty much pennies a glass.

OCZ Technology has long been a mainstay in the system memory industry and always at the top of the list for system builders everywhere. At one point OCZ was thought of as a hardware junkie's memory supplier but that couldn't be further from the truth these days. OCZ has a wide selection of memory options at every level and for every budget. They offer lifetime warranty on all of their performance memory and back it up with some of the best customer service and technical support in the business. Let's get started with a look at this 4GB kit of high performance Platinum memory that is priced extremely well for use in anyone's Windows Vista powered system.

Page 2 : Package & Contents

The OCZ platinum series utilizes their XTC heat sinks so there isn't too much that is new as far as visuals go with this kit. The package is also something we have see a number of times so we won't spend much time there as well.

Like was just said, OCZ has been using this heat sink/package combination for a while now and there have never been any complaints. The OCZ package of choice is a molded plastic clam shell that closes securely and does an excellent job of holding the modules in place. The branding is done via a standard OCZ cardboard insert. Module identification comes by way of the sticker on the memory sticks. As the saying goes…if it isn't broke…don't fix it.

Like all OCZ memory, there is a sticker proudly displayed on the one side of the heat sinks with all of the modules identification papers intact. OCZ uses this sticker to provide the part number, speed rating, timings, and size of the memory it is attached to. I have mentioned it before but will again. This sticker is great for identifying the memory, but it is counter intuitive to what the XTC heat sink is designed to do which is provide an avenue for air to directly contact the IC. I can't think of an alternative off the top of my head but I would really like to see OCZ come up with something interesting to alleviate this design contradiction.

The OCZ XTC heat sink has become the standard heat sink for OCZ performance memory replacing the bulkier solid heat sinks of years ago. The major benefit to this heat sink design is that it is so light and provides adequate cooling without increasing size. The XTC heat sinks allow for easy installation in pretty much any setup. There is barely any added width and absolutely no added height to these modules so they are pretty much the same as a non-heat sinked memory module. The installation demonstration we have planned in just a bit will illustrate this point very well.

At the heart of the XTC heat sinks on this OCZ Platinum 2x2GB PC2-8000 memory is the honeycomb micro-convection design. It is this design that helps eliminate dead spots around the memory for heat to build up. The design promotes airflow around the entire module providing even and adequate cooling. From the side we can see these modules are double sided and the heat sink adheres to the memory through a very thin layer of thermal adhesive. This adhesive can also be seen through the honeycomb holes in the heatsink. As said, this heat sink really is nothing new from OCZ but it is a very effective cooling solution that doesn't interfere with installation of the memory.

Page 3 : Specifications & Features

Having taken a couple good looks at the sticker on the OCZ Platinum 2x2048MB PC2-8000 memory, we are pretty much fully versed on the specifications of these modules. The specification list below is pulled from the OCZ Technology web site to fill in the blanks.

We already know the memory is rated for operation at DDR2-1000 (500MHz) with timings of 5-5-5-18. The voltage that the memory is rated to run those settings is 2.1v, but there is a note that the memory is still under warranty up to 2.15v. This gives a little bit of vDIMM overhead but the word in the locker room is that these 4GB kits aren't the vDIMM lovers of the 2GB performance kits of past. At the bottom of the specifications there are two part numbers listed, the 2x2GB kit that we are looking at today, and single 2GB modules of this memory are offered.

OCZ has gone with a conservative approach with this memory and the SPD profiles. They only provide a profile up to 400MHz at 5-5-5-15 timings. This will definitely ensure that the memory will boot with any motherboard at the DDR2 default of 1.8v. Furthermore, this means that at defaults, the system will attempt to run this memory at 5-5-5-15 and not the rated TRAS of 18. Because of this, it might be required to go in and set the timings manually. The next screen shot is what the memory defaults to when the correct divider is chosen in the BIOS to run the memory at DDR2-1000. The following CPU-Z screenshot also happens to be at 1.8v too which was a nice surprise.

As suspected, the memory defaults to a tRAS of 15 and through Memset, I was able to see that tRFC defaulted to 46. With recent talks of 2GB modules requiring a tRFC of 52+ for stability at 500Mhz, I am surprised not to see a tRFC of 52 or 54. We will see in the overclocking section if they passed stability testing at stock with these default settings above.

Before moving on to the installation section, I want to talk about one more thing on the product page for this memory at the OCZ web site. Tony (AKA OCZTony, AKA BigToe) has been working really hard lately trying to get setup information about each of their memory kits on the OCZ support web site. I have noticed a lot of posts with various motherboard configurations coming from the entire OCZ staff in recent months. Tony has continued that effort with this kit offering up two links with information about 4GB kits and Asus motherboards here, as well as setup details of this kit and the Foxconn Mars motherboard here. Again, this just goes to show the type of support you get when you buy an OCZ product.

Page 4 : Installation

This installation section really isn't necessary with the ‘natural’ size of these modules but a couple photos won't hurt anyone. Plus, I have possibly the absolute worst motherboard as far as layout is concerned when it comes to CPU/DIMM slots with the Thermalright Ultra-120. You'll see what I mean.

In the test setup, the Abit IP35-Pro and the Rosewill RCX-Z775-EX, the memory clearly just keeps to itself and doesn't bother anyone. There is plenty of room around the DIMM slots on this motherboard and the OCZ Platinum memory we are working with today look rather puny with all the open space around them, now for the challenge.

At first glance, one would think that the CPU heat sink needs to be removed in order to install or remove the memory but that is not the case with the OCZ Platinum XTC heat sinks. The modules are quite easy to remove from the first two slots and as we can see, fit snug as a bug in a rug (shrug). There really isn't another setup in recent memory that would restrict memory this badly and the XTC heat sinks have no issue at all. This is the major upside to the very slim and light weight XTC heat sinks.

Page 5 : Test Setup & Overclocking

Going back to DDR2 was a little bit weird as I haven't setup the IP35-Pro for a while. It is nice though visiting a setup of old with a few new parts to see how they work. All of my bench setups have been updated with 45nm C2D E8400s and this one is no different. The IP35-Pro needed a beta BIOS update but other than that, it was ready to go with 2x2GB sticks and the E8400. Here is a complete list of the hardware used for the testing and benchmarks:

• CPU: Intel C2D E8400 (Q740A523)
  
  
• CPU Cooling: Rosewill RCX-Z775-EX
• MB: Abit IP35-Pro (Beta Bios dated 09/06/2007)
• PWM Cooling: Stock
• NB Cooling: Stock
• SB Cooling: Stock
• GPU: HIS HD3870X2 1GB (Catalyst 8.2)
• PSU: Thermaltake Toughpower 700W
• HD: Seagate SATAII 80GB 8MB NCQ
• OS: Windows Vista Ultimate (with all updates/hot fixes)

Memory:

• OCZ Platinum 2x2048MB PC2-8000 (OCZ2P10004GK)
  
  
• 2x OCZ Reaper HPC Edition 2x1024MB PC2-9200 (OCZ2RPR11502GK)

The OS has been setup as a typical Vista user might have it setup. The sidebar is going to be left enabled with a number of gadgets running on it, the full Vista Aero package will be intact throughout testing, and all of the systems services are left running in the background. We could have stripped Vista down to look and run more like XP but that would defeat the purpose of the whole testing. We want to see if 4GBs of memory performs better than 2GBs on a typical loaded Vista system. The benchmarks might not be the best method to determine this but it is the only objective method we have. Subjective performance will also be commented on throughout the benchmarks and in the conclusion.

Stability testing memory is a commonly asked question in the computer forums and very seldom do you ever hear the right answer. Okay, so there is no real right answer but you never hear the correct logic for an answer. A lot of people like to pick an application and hang their hat on it. If it passes this applications stability test, then it is stable to them. I have personally tested so many kits of memory that I know enough to never trust a single program. Instead, I like to rely on an entire suite of memory stability tests in order to fully deem a kit of memory, overclock, or setup; stable. This type of stability would be for 24/7 daily use or in a system crunching for a distributed computing program like F@H, Rosetta, or WCG. The following is what constitutes stable to me for a 4GB kit of memory:

• 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 using HyperPi 0.99b)
• 2 hours of dual Prime95 using Prime95 v2.55 on blend mode
• 2 hours of quad MemTest Pro in Windows using 800MB/instance for three and the fourth using all available memory

As mentioned, this is not the end all say all for memory stability but I have used this menu for a long time now through DDR2 and DDR3 with very good success in finding a completely stable setup. Before we even get to some overclocking, let's first see how low of voltage we can get this kit to run the rated frequency and timings.

Despite this kit being rated for DDR2-1000 5-5-5-18 at 2.1v, it clearly has no problem running at much less voltage. We can see in the screenshot that Everest in the side bar is reporting vDIMM at 1.84v and this was confirmed by measuring the VDD output of an open DIMM slot with a calibrated digital multimeter. Keep in mind this is a single kit and results are going to vary from kit to kit but the OCZ Platinum 2x2GB PC2-8000 is off to a great start. Let's now see how this particular kit overclocked at a various timing sets.

As it turned out, the OCZ Platinum 4GB kit clocked really well at low voltage and ran spec with ease well below rated voltage but they just didn't scale with voltage at all. I found no headroom going over 2.00v at either CL5 or CL4. Trying to POST at CL3 was also a futile effort. Overall though, this memory performed quite well here despite the lack of overclocking. Some kits of this memory will likely reach DDR2-1100 completely stable at 2.0~2.1v but the majority of users should be aiming for around the DDR2-1050 mark with this memory. DDR2-800 at 4-4-4 was a bit of a nice surprise and really gives this kit the versatility running a variety of FSB frequencies and memory ratios.

DDR2-825 4-4-4-18 2N @ 2.00v (Actual)
……………..
DDR2-1070 5-5-5-18 2N @ 2.00v (Actual)

Page 6 : Memory Benchmarks

For the benchmarks we decided that a 2x2GB, 2x1GB, and 4x1GB comparison would be interesting. To be honest, we weren't expecting much of a difference in these benchmarks amongst the three setups but one never knows, that's why you play the game. The 2x1GB setup will be compared with the OCZ Platinum 2x2GB kit at DDR2-1000 and DDR2-800 and the 4x1GB will be compared at the overclocked 5-5-5-18 timings of the 2x2GB OCZ Platinum.

The memory key above outlines the various settings that were used for all of the benchmarks. As we can see, we have gone with two different CPU speeds. The stock clocks of the E8400 at 3GHz as well as overclocked to 4GHz. The only drawback of using the IP35-Pro for these types of comparisons is the fact that we can't change the tRD (Performance Level or PL) in the BIOS or Windows. This immediately puts the DDR2-800 4-4-4-18 results at a disadvantage to the other setups at 3GHz. So keep this in mind when we go through the results.

SiSoft Sandra Lite XII – Memory Bandwidth

Lavalys Everest Ultimate 07 v4.00 – Memory Bandwidth

ScienceMark v2.00 – Memory Bandwidth

We start off with a look at bandwidth and we can see that the DDR2-800 setup just doesn't stand a chance up against the DDR2-1000 configurations running at a higher tRD. There is however a couple of interesting bits of information here. First off, the 2x2GB kit seems to be providing the slightest bit more bandwidth than the 2x1GB kit and the 4x1GB appears to be performing about the same amount better than the 2x2GB kit at the higher frequencies. The differences are very minor in both instances but there does appear to be some consistency to the results.

SiSoft Sandra & Lavalys Everest – Memory Latency

Again, PL6 VS PL5 leaves the DDR2-800 results looking like second class citizens but the rest of the numbers are pretty much identical. The latency numbers don't appear to have any rhyme or reason to them and aren't showing any discrepancy between 2x1GB, 2x2GB, or 4x1GB setups.

Futuremark PCMark Vantage – Memory Bench

Like the rest of the results in the bandwidth section, the PCMark Vantage Memory benchmark numbers are about the same across the board with one exception. The DDR2-800 result actually fared better than the DDR2-1000 numbers. I guess this goes to show how much stock we can hold in PCMark Vantages memory testing suite. Of course, the memory test in Vantage could be more indicative of real world results and we might not want to dismiss it so quickly.

Page 7 : System Benchmarks

Moving on to more system orientated benchmarks we have maintained the same setups for testing as we did with the bandwidth testing. These same five setups will of course be used for the 3D and gaming benchmarks as well. It just wouldn't make sense if it didn't.

WinRAR 500MB Benchmark

The first real twist in the numbers shows up here in WinRAR with the 2x1GB kit handily out-performing the 2x2GB kit of memory by 20 seconds. The 4x1GB kit also beats the 2x2GB kit by a solid 8 seconds in the benchmark up to 500MB. The 4x1GB kit performing better in WinRAR is not much of a surprise based on bandwidth results and the fact that WinRAR heavily relies on memory bandwidth for its power but I am not sure why the 2x1GB kit performed as well as it did. Both kits have double sided modules and the bandwidth numbers seemed to favor the 2x2GB kit if anything so this is a bit interesting.

Futuremark PCMark Vantage

The default test suite of PCMark Vantage tests all kinds of system abilities from hard drive performance to multi-tasking speed as well as image manipulation and web browsing speed. Because of the full system performance scale, it is no surprise to see the 2x1GB of memory is the loser here. This perhaps is the only and most telling benchmark of the bunch as it actually would come close to testing how 4GBs or memory do in a system compared to 2GB. Again though, 4x1GB outperforms the 2x2GB kit so it really does look like 4x1GB might marginally outperform 2x2GB. Of course with 4x1GB you are limited to 4GB where as the 2x2GB kit could be upgraded to 8GBs and there is some discussion in forums of users still having issues with certain motherboards and all four DIMM slots being used. It isn't as common and the problems aren't as severe as it was back in the Socket 939 days but it is still understood to be beneficial to go with 2x2GB kits over 4x1GB.

Lame Front End – WAV => MP3 Encoding

We added the WAV to MP3 encoding test just to see what a real-world task would perform like in this environment with the various sizes and configurations of memory. Almost everyone using a computer at one point has converted a CD to MP3 for use on their MP3 player so this task seemed like a very relevant test. As we can see there might be a slight advantage to the 2x2GB setup versus the 2x1GB kit but the difference is marginal at best and could simply be chalked up to margin of error between tests. It might not be conclusive but then again, nothing we have seen up to this point can really be considered concrete evidence. All the results are pointing in the direction of more memory being better but a large difference just isn't showing up in these benchmarks.

SuperPi Mod v/1.5 – 1M

SuperPi Mod v/1.5 – 32M

The SPi results again show very little information about these different setups but the 4x1GB setup out-performs the 2x2GB kit like it has all day long. The rest of the results are too close to even consider being different. As mentioned numerous times, these benchmark results just don't show any kind of conclusive results to which setup is better. Running the OS on a day to day basis is easily quicker with the 2x2GB kit over the 2x1GB setup though. The system feels a lot snappier, windows open a bit quicker, there aren't near as many slow downs, and the network is more responsive. Keep in mind, this is on a typical users Vista installation with full visual effects and desktop gadgets running.

Page 8 : 3D & Gaming Benchmarks

This is the section I have really been looking forward too. All of the benchmarks up to this point, with the exception of maybe PCMark Vantage and the MP3 encoding test, are just that…benchmarks. They may have no bearing on real world performance and don't really replicate the difference between 2x2GB and 2x1GB. The gaming benchmarks are a whole different story. The 3DMark results will again just be a measuring stick method but the actual gaming benchmarks done are of actual games and will reflect the differences in performance when playing the game. The system has a very powerful video card and we will be running the games at the typical resolution and detail levels that we would when playing the games to accurately portray the system environment.

Futuremark 3DMark 01/03/05/06

There isn't really much to discuss here as the results seem to go back and forth amongst the different versions of 3DMark. The only real consistent results are the fact that 3DMark 01 runs like garbage on windows Vista. This upsets me greatly because as of right now, CrossFireX drivers don't work in Windows XP and 3DMark 01 is easily my favorite 3DMark benchmark. How is a guy suppose to bench 01 with a pair of HD3870X2s in XP ATI? Cough up XP CrossFireX drivers, I don't care how much Microsoft is paying you not to.

Gaming Benchmarks – Crysis

The first of the three games we are going to look at it is Crysis. The settings are listed above, medium level details and DX10. The game looks very nice at these settings and based on the results we can see is perfectly playable. A few slow-downs every now and then but for the most part it is excellent. The other thing you will notice is that we have absolutely zero difference amongst the results. The average and low FPS is what we are most concerned with and any of these differences listed here are as equal as it can get without fudging the numbers. Now granted, even at these settings Crysis is going to be GPU limited but this is what most people are going to be playing Crysis at with a high-end video card and going from 2GB to 4GB equates to no gains. Testing with a lower end video card is just going to GPU limit the performance further and the system will play even less of a role in performance.

Gaming Benchmarks – UT3

Next up is UT3, the most system dependant game of the lot. Here we actually seem some variation in the results but it isn't exactly what we were expecting. The lowest average FPS is with the 2x2GB kit at DDR2-1000. The DDR2-800 numbers actually give a higher average. Looking at all three runs that were averaged out to get the numbers posted on the graph indicates nothing out of the ordinary. All three runs for the DDR2-1000 2x2GB setup were less than 71FPS for average FPS. It simply looks like UT3 might prefer tighter memory timings over frequency.

Gaming Benchmarks – COD4

COD4 has been something I have recently become addicted to and I am not that big of a gamer. The multi-player portion of the game is as fun as watching monkeys fling their own feces at each other at the zoo. The benchmark ran for the results today was an actual recorded time demo of a multi-player free for all match I played on the Ambush level. It was just over 5 minutes long and the FPS were captured throughout with FRAPs. The time demo was then played back three times on each setup with the results averaged for the graph above. Again, we see very little, if any at all, discernable difference between the setups. The one thing to point out is that the 4GHz average FPS results are not much higher than the 3GHz FPS averages which indicate very little reliance on the system. With that said, the memory plays even less of a role.

All of the gaming benchmarks results we have looked at today are quite indicative of what the average gamer is going to see with a similarly spec'd system. System performance plays a very small role in gaming performance at a certain level and with typical resolutions and detail levels, the GPU is still the primary performance piece of hardware in your system. The fact that OCZ and other memory companies advertise their memory as ‘designed for gamers’ is really just advertising 101. Gamers spend the most on computer hardware, upgrade more often than anyone else, and have the most disposable income…direct advertising dollars at them. I can't fault OCZ or anyone else for this strategy, business is business.

The reality of it is that this 2x2GB kit of memory really would show less than negligible performance gains over a 2x1GB kit of generic ram running at DDR2-800 or perhaps even slower. The price difference isn't that much over those same generic kits though and there are a lot of benefits I will discuss in the conclusion. Sure, we could skew the benchmark numbers by running the gaming benchmarks at low details and low resolution but that wasn't the point of today's benchmarks. I wanted to see if the memory would play a role at typical settings and it obviously doesn't. That not to say there isn't a large benefit to the 2x2GB kit. The 2x2GB kit may also play a larger role in a system of less power than the one tested on today. This might be explored in up-coming reviews or articles.

Page 9 : Conclusion

Today was a bit of a unique journey for me as a reviewer and writer. Typically reviewing is all about basing judgment on objective testing results. There is a level of subjectivity when it comes to case reviews and certain aspects of any review really. At some level though, objective testing and result analysis comes into play and is the primary source of information for passing judgment. With this OCZ Platinum 2x2GB memory, however, the objectivity aspect of the review completely negates all of the positive aspects of this memory. If objectivity is all we had to go on, this memory would be a waste of your money as would all 2x2GB, or 4x1GB kits of memory…which most definitely is not the case. It is just hard to see the advantages of large amounts of memory unless you are sitting at the machine using it.

Lately I have found myself using Windows Vista for a lot of reviews because of drivers or manufacturer request or what have you. Up until I received this kit of memory, I hated Vista, with a passion. No matter how stripped down I got Vista, it still wasn't as snappy as my tweaked Windows XP installation. That all changed when 4GB of memory entered the scene. I never even thought of using 4x1GB of memory but that would have done the trick as well. The system completely comes alive once you go from 2GB to 4GB.

Going from 2GB to 4GB was like bringing my over-powered, under-performing system out of its shell. There were no nagging little delays jumping through menus. Switching between tasks was no longer annoying. The network latency which was new to me in Vista all but vanished. Photoshop and Excel became useable tools again and after re-configuring the scratch disks and the paging file, Photoshop was quicker than it had ever been. I now want to grab a second kit of this memory and see how much 4x2GB will speed things up. Of course, all of this is not visible in any benchmarks, as we saw today.

The fact of the matter is this. If you are building a new machine, especially a mid-range to upper end machine like I did the review on today, 2x2GB is the only way to go. The prices of 4x1GB of even PC2-6400 memory comes out to be more than this OCZ Platinum PC2-8000 2x2GB kit and with the 2x2GB kit you have all the advantages of potential upgrade to 8GB and no potential headaches of running 4x1GB modules. Really, there is no reason not to be running 2x2GB kit of memory at this time if you are using Vista. Sure, the gaming and benchmarks benefits just aren't there, but the more important things associated with everyday use of the system benefit greatly from the larger amount of memory. Of course, because of the subjectivity of these claims I have to say…in my opinion.

Advantages

• Price of this kit is great, wonderful, and beyond impressive
• There is a bit of overclocking headroom and the ability to run 4-4-4 timings
• Windows Vista daily performance increased 10 fold going from 2GB to 4GB of memory
• Photoshop, Excel, and movie editors all noticeably benefit from increased system memory
• OCZ XTC heat sinks allow this memory to fit under any massive CPU heatsink

Disadvantages

• There is just no gaming performance increases going to 2x2GB from 2x1GB, even with a high end video card and system
• You will wish you had always had 4GB of memory if you are running Vista and feel foolish for waiting so long like myself

Overclockers Online would like to thank all of the OCZ staff for helping make this review possible.