![]() |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| « Back - Print | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Page 1 - Introduction/Test Configuration The last few months haven’t seen many large changes in the memory market other than confirming the change in standards from DDR to DDR2, a large part of which can be accredited to the AM2 launch which had both Intel and AMD supporting a unified memory solution again. The popularity of Core 2 chips among enthusiasts has also greatly spurred the popularity of DDR2 because it is the only memory standard compatible with the, as of right now, performance-crown-holding processors. Using DDR2 as the new unified memory solution means that memory manufacturers can begin focusing on one technology instead of two. In other words, DDR2 memory has become the point of emphasis for pretty much all memory manufacturers. As a result of this, memory manufacturers are following a market trend that we have seen in the past and will most likely see again in the future with DDR3. It is a market trend that pretty much always develops with the introduction of a new memory standard: manufacturers begin aggressively looking for ways to increase the clock frequency of their products while simultaneously implementing tighter timings. In this article we have intentionally limited ourselves to analyzing the overclocking performance of only two categories of DDR2 memory. The modules we are testing are all classified as DDR2-667 (PC2-5400) and DDR2-800 (PC2-6400). For the article, we chose only memory which fit into a price range available to a large majority customers. Although DDR2-1100 kits might look appealing, the truth of the matter is that they are generally in a price range that makes them suitable to a small portion of consumers. In other words, only a few people can actually afford them. For this analysis, we chose to analyze memory kits that have a total of 2 Gigabytes. The reason behind this decision is that Dual Core CPUs are able to perform better in multitasking paired with 2GB of memory instead of 1. Below is a list of the modules we will be testing in this article:
PC2-6400 Category
Test Configuration and Procedure
The motherboard we chose was from the upper mid-range segment of the market: the ASUS P5B Deluxe which is based on the Intel P965 chipset. In order to test stability of the modules after they were overclocked, we ran a pure calculation test, SuperPi 32M, and a graphical application, 3dmark 2001. To simplify the number of overclocking results we had, we limited ourselves to timings of only 4.4.4.12. All others timing formations were not included in our results. In order to allow the modules to reach their maximum frequency, we gave them as much voltage as they demanded. However, we were limited by the maximum voltage level our motherboard could supply, 2.45V. The differences in voltage among the modules came from the diverse characteristics of the various chips used for each kit. In order to have the modules perform to the best of their ability, we properly cooled all modules during the tests with an 80mm fan. In the article, we will begin by introducing the products we are testing from each manufacturer. After all products have been introduced, we will analyze the results of our tests and compare overclocked frequencies against the default values. Page 2 - Mushkin Mushkin sent us a total of three kits; one kit was for PC2-5400 and two were PC2-6400. Below are the main technical characteristics of the modules:
All of the modules come in plastic packaging. The cardboard background of the two PC2-6400 modules adorns the new green and white Mushkin logo, which in our opinion looks refreshing. All of the modules were equipped with Mushkin’s “Frostbite” heatspreaders, which are larger than standard heatspreaders. Each memory kit from Mushkin that we will be testing belongs to a different category. There are three categories in all. “HP” stands for High Performance, “XP” stands for Extreme Performance, and “EM” stands for Enhanced Memory. The last one, as the name suggests, is aimed at the value market segment. The main differences between the modules are the different chips used in them, which in turn offer different levels of performance.
The overclocking results of the two PC2-6400 memory kits come out to be particularly interesting. The XP2-6400 ended up needing the highest voltage our motherboard could offer, 2.45 Volts. Its overclocking performance, however, turned out to be quite impressive, a result of the Micron Fat Body D9 memory chip that the module comes equipped with. The Enhanced Memory kit we tested, which as we said before is classified as value RAM, turned out to have an overclock of 17.5% while running at timings faster than the default values. Page 3 - OCZ OCZ sent us two kits that belong to two categories. Below are the technical specifications for the modules:
Again, OCZ uses standard plastic for its packaging. Both kits use the unique Platinum Mirrored XTC Heatspreader. The honeycomb heatspreaders are based on OCZ’s latest design and supposedly allow for increased ventilation and heat dissipation. The Platinum Edition also comes certified for use with ATI Crossfire Technology. Aesthetically, the heatspreaders are pleasing, especially the Special Ops Edition which uses a unique camouflage design. As is customary with OCZ products, both kits come with a lifetime guarantee and extended voltage protection which allows the modules to handle voltages up to 2.2V ±5% and still be covered by our Lifetime Warranty.
In spite of the quality heatspreaders and aesthetical features, we were not able to reach very special overclocks with the OCZ kits. The maximum frequency we were able to reach was only an improvement of 5% and 11%. Page 4 - Geil Geil also supplied us with two kits for our testing. The following are their technical specifications:
As always, unlike other manufacturers, Geil set itself apart by packaging memory kits in compact cardboard boxes. Although pretty much the same externally, there is a slight difference between how the two modules are placed within the packaging. The PC2-5400 modules come packaged in cardboard while the PC2-6400 comes packaged in plastic. Both kits have the same heatspreaders and have the same orange color. The labeling on the packages are only different in order to distinguish the model numbers between the two kits. There is also a slight discrepancy in the voltage information. While the boxes’ label says a medium value of 2.1 Volts is required, the website advises using 2.3-2.4Volts. It is also important to note the low latency value of the PC-5400 modules, which are at 3-3-3-4.
The Geil kits’ overlocking results turn out to be quite interesting. Both models are sensitive to voltage and end up requiring the same amount, 2.45V. Their maximum frequencies also turn out to be quite similar to each other at around 500 MHz. The most interesting aspect of the results is the overclock we see from the Geil PC2-5400 Ultra, which ends up having a maximum frequency 54.6% higher than its default value. The PC2-6400 module had a smaller overclock of 23.7%, and its maximum frequency ended up being less than the PC2-5400 Ultra. Page 5 - Teamgroup We also received two Teamgroup memory kits for the overclocking analysis. The first kit had speeds of DDR2-667 while the second was DDR2-800/1000 with timing differences relative to the frequency the modules are running at.
The memory kits come in relatively simple plastic packaging. The red “X” which covers much of the modules stands for “Xtreem”, which is Teamgroup memory meant for overclocking and overclockers. The heatspreaders used for the modules are not very efficient and have no clamps holding them in place. Instead, they are simply glued to the chip using a sheet of adhesive thermal paste.
In this case, the major difference which allows the modules to reach high overclocks is the Micron D9 memory chip which the modules use. The D9 memory chips are usually power hungry, and so we again reached our maximum voltage limit of 2.45V. For the Teamgroup DDR2-667 the increase in frequency was 69.6% while for the dual-certified DDR2-800 we had an overclock of 15% (DDR2-1000 which is 500MHz) and 43.7% (DDR2-800). In reality, Teamgroup’s default timings are more aggressive at 400MHz (3.4.3) and more relaxed at 500MHz (5.4.4), so for that reason we can say we have a middle value equal to 29.3% for the tests run at timings of 4.4.4. Page 6 - G.Skill GSkill sent us a single PC2-6400 memory kit for this analysis. Below are the technical characteristics of the F2-6400CL4D and a link to the manufacturer web page of the product.
The packaging for GSkill products has varied a lot recently. For this memory kit, the modules came packaged in plastic all of which was placed inside a cardboard box. The heatspreaders, which are blue brushed metal, are held in place using brackets and are decorated with the G.Skill logo.
The modules proved to be particularly sensitive to higher voltages and we couldn’t go over 2.10V. The maximum frequency we could reach with our testing came out be at 450 MHz, an increment of 12.5% compared to the default speed. Page 7 - Cellshock Cellshock is a company that has rarely targeted its products to overclockers. These modules aren’t widely available in stores or online, so you might have a tough time getting your hands on them. Nevertheless, we still benchmarked the modules and listed the results for readers to see.
The German modules came modestly packaged in plastic with little decoration. Both modules were laid on top of each other and slightly inclined. As the manufacturer indicates on their website, it is no secret that the modules use the Micron D9 GMH chip. Although Cellshock doesn’t go as far as other companies to offer a Lifetime Warranty for the memory kit, their warranty period of 5 years should be sufficient considering that current high-end memory will probably be obsolete by that time.
We were able to reach a maximum frequency 30% higher compared to the default speed. Staying true to the trend for our other D9-equipped modules, the Cellshock DDR2-800 required the maximum Voltage our motherboard could provide in order for the overclock to be stable. Page 8 - Overall Results and Analysis
Analyzing the results of our testing, it turns out that nearly all memory kits were able to reach standard DDR2-800 speeds, the only major exceptions being the OCZ PC2-5400 and Mushkin XP2-5300. It is also interesting to see the substantial alignment of the maximum frequencies of the Cellshock PC2-6400, Mushkin XP2-6400, and Geil PC2-5400. This is almost no doubt a result of the fact that all of those modules use the Micron D9 GMH chip. The proposals from Team Group turned out to reach the highest frequencies out of all modules we tested. The Team Group modules, which are also based on the Micron D9-GMH, not only lived up to the memory chips’ reputation, but went beyond the performance expectations. After seeing the alignment of the Cellshock PC2-6400 kit, Mushkin XP2-6400 kit, and Geil PC2-5400, we expected the Team Group proposals to offer similar frequencies to those modules since they are all based on the Micron D9. However, they ended up providing even higher frequencies.
Analyzing the percent of increase compared to the default frequency, the Team Group DDR2-667 had the largest overclock, reaching speeds of DDR2-1150. The memory was able to run 69.7% faster than its rated limited, clearly illustrating its impressive overclocking headroom. Having the smallest overclocking capacity, both OCZ memory kits found their way to the bottom of the chart. The Geil PC2-5400 also underwent a large gain in frequency, ending up having a 54.6% overclock compared to default timings. It is important to note that most of the memory kits that had the largest increases in frequency also had the lowest rated default speeds. Page 9 - Conclusion The test results we received today leave plenty of room for reflection. Looking at the DDR2-667 category, only two kits out of the four tested reached and exceeded DDR2-800 (400MHz). It is highly probably that the overclocking success of the modules that exceeded DDR2-800 can be accredited to the Micron D9 chips which both the Team Group DDR2-667 and Geil DDR2-667 kits used. The OCZ kits we tested all had rather constrained increments in overclocking. It seems that the primary focus for the modules was to offer a sure guarantee of running at rated speeds and voltages while being aesthetically pleasing. In other words, they weren’t really meant for venturing far past default frequencies. Pricing for the modules greatly varied according to manufacturer and speed. Unfortunately, we weren’t able to find the pricing of the Cellshock modules simply because of a lack of online retailers actually selling the product. If anyone does happen to come by the price with a link, please feel free to drop us an email. Please also note that these prices aren’t the manufacturer suggested retail prices. Instead, they are the lowest price we could find for these modules online. To view our sources for the pricing, please download this excel file. All prices are for 2 x 1Gb kits. As we said in the introduction of the article, pretty much all modules tested are available at an affordable price. The lowest priced memory kit with a Micron D9 chip was the Geil PC2-5400 Ultra, coming in at $224. Incidentally, the Geil PC2-5400 Ultra offered very impressive overclocking performance for its price. The Mushkin XP2-6400 was the most expensive solution tested, and although it did offer a decent overclock, the performance it offered was virtually the same as the Geil PC2-5400 Ultra. The two Teamgroup kits, which ended up having the highest maximum frequencies, were the second and third most expensive kits. The Teamgroup DDR2-667, however, still ended up being a sub-300 dollar solution. As can be plainly seen, the results of our tests make available an overclocking solution to pretty much every segment of the market. Modules based on the Micron D9 chip dominated the overclocking tests, and it seems that companies still have not been able to come up with a proper alternative memory chip that is as versatile and offers comparable performance. The Teamgroup kits had the highest overclocked frequencies; however, the Geil PC2-5400 in our opinion was able to offer the highest overall performance for its price. As is evident from reading the article, the primary focus of this article was sheer overclocking of memory kits in the mid-range memory segment. We hope to have a more in-depth performance article for these modules written soon. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Printed from Hardware Upgrade: http://www.hwupgrade.com Copyright 1997 - 2008 Hardware Upgrade S.r.l |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||