 |
 |
 |
||
|
|
||||
 |
|
|
| Intro and Corsair TWIN3X2048-1800C7DF | ||||||
|
We're in the early stages of a major transition in the PC arena, much like in 2004 when Intel launched their first core logic chipset with support for DDR2 memory. This time around, however, Intel is pushing DDR3 memory with their current crop of desktop chipsets. When DDR2 memory first hit the scene, we recommended users to tread lightly as the technology needed some time to mature. Back then DDR2 had much higher latency than standard DDR memory and it couldn't scale to high enough clock speeds to negate its latency disadvantages. And in many situations DDR2 memory actually performed below the bar set by standard DDR, not to mention it was much more expensive. In time though, DDR2 desktop memory scaled to well over 1GHz and its early latency disadvantage was erased by huge bandwidth gains. But it took quite a while for that to happen. In fact one could easily argue that the complete transition to DDR2 memory on the desktop took over two years and didn't happen until rival AMD introduced their socket AM2 platform in May of '06. The landscape today is very different. There are some obvious parallels, but here we are only four months removed from the launch of Intel's P35 chipset - the first desktop chipset to support DDR3 memory - and already the market is rife with high-end memory kits that dramatically outperform the initial offerings that arrived alongside Intel's new platform. DDR3 memory kits are still substantially more expensive than even some of the fastest DDR2 kits, and they certainly don't represent the best value in desktop memory, but with the amount of innovation going on in this space we thought it was time to pull together a handful of fast DDR3 memory kits to see just what they could do. We got our hands on five, dual-channel, 2GB kits from Corsair, Kingston (2), OCZ, and Super Talent each with different rated speeds and timings, ranging from 1333MHz to 1800MHz. Our intent is to show you how performance is affected by different clock speeds and timings using Intel's P35 chipset with a processor that has a 1333MHz front side bus speed.
"The Dominator Series TWIN3X2048-1800C7DF G is a 2048MByte kit of DDR3 SDRAM DIMMs built using Corsair’s latest high performance heat sink with Dual-path Heat Xchange (DHX) technology coupled with a Corsair Airflow Fan. This part delivers outstanding performance in the latest generation of dual-channel DDR3-based motherboards. It has been tested extensively in popular DDR3 motherboards to ensure compatibility and performance at its rated speed. This memory has been verified to operate at 1800MHz at latencies of 7-7-7-20 at 2.0V VDIMM."
The memory chips on both sides of the PCB are outfitted with their own dedicated, oversized heatsinks to help cool the chips as well. Having two paths from which heat can be drawn out of the modules is what Corsair is calling DHX technology, or Dual-Path Heat Xchange. And the Dominator fan included with the kit is the third component that aids in cooling, which in turn enhances stability, longevity, and overclockability. In addition to DHX, Corsair is aggressively binning the memory chips used on the Dominator kits and is using only select samples th at can operate reliably at their rated latencies and frequencies. The culmination of Corsair's efforts is a 2GB memory kit that's rated for operation at over 1.8GHz with 7-7-7-20 timings. We should note, however, that the TWIN3X2048-1800C7DF memory requires 2.0v to operate at its rated speed whereas many other DDR3 kits require only 1.7v to 1.8v.
|
| OCZ OCZ3P1600EB2GK | ||||||
|
Representing OCZ in this round-up is the company's OCZ3P1600EB2K 2GB memory kit. As its name implies, the OCZ3P1600EB2K kit is rated for operation at 1600MHz (800MHz DDR). It's also rated for 7-6-6-20 (CAS-TRCD-TRP-TRAS) timings and requires a 1.9v feed.
"The PC3-12800 EB (Enhanced Bandwidth) Edition is built with ardent enthusiasts in mind, featuring the ideal balance of sheer DDR3 speed and enhanced timings. This memory is specially designed and optimized for the latest generation of ASUS motherboards. As one of OCZ’s legendary Enhanced Bandwidth (EB) editions, the DDR3-1600 Platinum EB increases effective memory bandwidth through the optimization of memory latencies between the system memory, the chipset, and memory controller. At 1600MHz, the PC2-12800 EB edition is rated for 7-6-6 timings on ASUS platforms."
The matched paid of 1GB DIMMs used in the OCZ3P1600EB2K kit are adorned with OCZ's XTC (Xtreme Thermal Convenction) heat spreaders, which feature a perforated honeycomb design that's meant to increase surface arean and allow for air to circulate through the heat spreader to the PCB and chips beneath. This kit also features OCZ's EVP (Extended Voltage Protection). What EVP does is allow overclockers and power users to increase the memory voltage to as high as 1.95V without invalidating OCZ's Lifetime Warranty.
|
| Kingston KHX11000D3LLK2/2G | ||||||
|
Kingston's KHX11000D3LLK2/2G memory kit isn't reated for clock speeds quite as high as Corsair's, OCZ's, or Super Talent's, due to the fact that the DIMMs are built using different memory chips. Whereas Corsair, OCZ, and Super Talent employ Micron chips, the Kingston KHX11000D3LLK2/2G kit and the KHX11000D3ULK2/2G kit we're going to show you on the next page use Elpida chips.
"Kingston's KHX11000D3LLK2/2G is a kit of two 128M x 64-bit 1GB (1024MB) DDR3-1375 CL7 SDRAM (Synchronous DRAM) memory modules, based on sixteen 64M x 8-bit DDR3 FBGA components per module. Each module pair has been tested to run at DDR3-1375MHz at a low latency timing of 7-7-7 at 1.7V. The SPDs are programmed to JEDEC standard latency DDR3-1066Mhz timing of 7-7-7 at 1.5V. Each 240-pin DIMM uses gold contact fingers and requires +1.5V. The JEDEC standard electrical and mechanical specifications are as follows:"
Although the company's website shows DDR3 memory with a different heatspreader design, the KHX11000D3LLK2/2G kit we received features the same blue, aluminum heatspreads as the rest of the HyperX product line. We point this out only to avoid confusion when browsing the Kingston website, as the heatspreaders shown here are more than capable of handling the heat produced by these DIMMs. In fact, during testing they didn't get hot enough to cause any concern. As its suggests, this 2GB kit is comprised of a pair of march 1GB DIMMs rated for operation at PC3-11000, or DDR3-1333. Timings of 7-7-7-15 are supported at this clock speed when the DIMMs are supplied with 1.7v.
|
| Kingston KHX11000D3ULK2/2G | ||||||
|
At first glance, the Kingston KHX11000D3ULK2/2G memory kit looks just like the KHX11000D3LLK2/2G shown on the previous page, save for the different decals affixed to its heatspreaders with the kits model number and specifications. But the KHX11000D3ULK2/2G is actually quite different, at least from a performance standpoint.
"Kingston's KHX11000D3ULK2/2G is a kit of two 128M x 64-bit 1GB (1024MB) DDR3-1375 CL5 SDRAM (Synchronous DRAM) "ultra low latency" memory modules, based on sixteen 64M x 8-bit DDR3 FBGA components per module. Each module pair has been tested to run at DDR3-1375MHz at a low latency timing of 5-7-5 at 1.8V. The SPDs are programmed to JEDEC standard latency DDR3-1333Mhz timing of 8-8-8 at 1.5V. Each 240-pin DIMM uses gold contact fingers and requires +1.5V. The JEDEC standard electrical and mechanical specifications are as follows:"
Like the KHX11000D3LLK2/2G memory kit we showed you earlier, Kingston's KHX11000D3ULK2/2G kit is rated for operation at PC3-11000, or DDR3-1333 speeds. This memory kit also features the same basic, blue aluminum heatspreaders and is comprised of a matched pair of 1GB DIMMs. The Kingston KHX11000D3ULK2/2G kit, however, features much lower supported timings of 5-7-5-15 at 1.8v (as opposed to 7-7-7 on the KHX11000D3LLK2/2G kit). As we mentioned in our introduction a few pages back, one of the major complaints voiced regarding early DDR3 memory samples was their relatively high latencies. As CAS 5 at 1333MHz though, Kingston's kit features some of the lowest latencies of any high-speed DDR3 memory kit.
|
| Super Talent W1600UX2G7 | ||||||
|
While the underlying chips used on the W1600UX2G7 memory kit may be similar to those used by Corsair and OCZ (and many other memory manufacturers for that matter), Super Talent does a number of things to differentiate their W1600UX2G7 from the rest of the pack.
"The W1600UX2G7 is an extreme performance 2GB kit of low latency DDR3-1600 DIMMs that supports 1600MHz+ clock speeds. It is ideal for gamers, power users and overclockers, and is perfectly suited for motherboards based on the new Intel P35 chipset. This kit has been tested and is guaranteed to operate at 1600MHz at very aggressive 7-7-7-18 latencies at 1.8 volts. It has been tested as a matched pair of modules in a dual channel motherboard to ensure ultimate reliability, compatibility and performance. These modules are clad in our custom black cast aluminum high-efficiency (HE) heatspreaders for optimum thermal performance. Efficient cooling is essential to optimize speed and to preserve the life of the DIMMs."
There's no mistaking Super Talent's W1600UX2G7 kit in light of all the others we've shown you here in this article. The Super Talent W1600UX2G7 DIMMs were the only ones we looked at to feature a heatsink on only one side of their PCB. As you can see in the picture above, there are no chips on the other side of PCB, so there is no need to affix a heatsink there and increase costs. Also notice we're not referring to the W1600UX2G7's heatsinks as a heat spreader. While it is thin and mostly flat, the heatsinks used on these DIMMs do feature small raised fins that increase surface area and dissipate heat more effectively than completely flat heat spreaders. As you've probably surmised by its name, the W1600UX2G7 memory kit is rated for operation at DDR3-1600 speeds and is comprised of a matched pair of 1GB DIMMs, for a total of 2GB. Other specifications include support for 7-7-7-18 latencies, but at only 1.8v.
|
| Test System and Stock Performance | |||||||||||||||||||||
|
Compatibility Testing: Before we sat down in front of our test machine to evaluate the DDR3 memory kits represented in this article, we installed them into some of the systems we had available at the time to see if there were any compatibility problems to speak of. We tried the sticks in two different systems, powered by the P35 chipset. We tested these modules on the following platforms...
With the exception of the Asus Blitz Extreme, which we used for the benchmarks in this article, to quickly test these modules on the other board all we did was install them, power up the systems, made sure Windows booted and then we ran SiSoft SANDRA's burn-in wizard for a few minutes. We didn't experience any issues whatsoever. All of the memory kits worked perfectly in both of the platforms we tested.
How we configured our test systems: When configuring our test system for this review, the first thing we did was enter the system BIOS and set the motherboard to its default configuration. The hard drive was then formatted, and Windows XP Professional (SP2) was installed. When the installation was complete, we installed all of the necessary drivers for our components, and removed Windows Messenger from the system. Auto-Updating and System Restore were also disabled, and we setup a 1024MB permanent page file on the same partition as the Windows installation. Lastly, we set Windows XP's Visual Effects to "best performance", installed all of our benchmarking software, defragged the hard drive and ran all of the tests. Throughout all our benchmarking, we had the memory voltage set to their default values, unless otherwise noted.
We began our testing with SiSoftware's SANDRA, the System ANalyzer, Diagnostic and Reporting Assistant. SANDRA consists of a set of information and diagnostic utilities that can provide a host of useful information about your hardware and operating system. We ran SANDRA's Memory Bandwidth and Latency tests on a Core 2 Duo E6750 powered test bed with all five different brands / types of memory installed. The BIOS was manually configured to each kit's rated frequency and timings, using a combination of multiplier, FSB strap, and FSB manipulation in an effort to keep the CPU speed as close to its default of 2.66GHz and to keep the different configurations as close to the same clock speed as possible. The breakdown of different "stock" configurations is as follows:
As you can see, according to the SiSoft SANDRA memroy bandwidth benchmark, clock speed is king. The higher clocked kits all offered more bandwidth, regardless of their CAS latencies. The highest clock kit, Corsair's TWIN3X2048-1800C7DF, offered the most bandwidth at about 8.2GB/s, followed by the OCZ kit, then Super Talent, and then two Kingston kits.
The higher clocked kits also offered the lowest latency with the Corsair TWIN3X2048-1800C7DF once again taking the top spot. Also note, that despite having a CAS latency of 5, the Kingston KHX11000D3ULK2/2GB has only marginally lower latency than the similarly clocked KHX11000D3LLK2/2G kit which is rated for operation at CAS 7. |
|||||||||||||||||||||
| Stock Performance (Cont.) | ||||||||
|
For our next round of benchmarks, we ran the Memory performance module built-into Futuremark's PCMark05. For those interested in more than just the graphs, we've got a quote from Futuremark that explains exactly what this test does and how it works...
"The Memory test suite is a collection of tests that isolate the performance of the memory subsystem. The memory subsystem consists of various devices on the PC. This includes the main memory, the CPU internal cache (known as the L1 cache) and the external cache (known as the L2 cache). As it is difficult to find applications that only stress the memory, we explicitly developed a set of tests geared for this purpose. The tests are written in C++ and assembly. They include: Reading data blocks from memory, Writing data blocks to memory performing copy operations on data blocks, random access to data items and latency testing."
The memory performance module included with PCMark05 is also influenced by the CPU's clock speed, as is evidenced by the OCZ kit outperforming Corsair's despite a 200MHz clock speed defecit.
We continued our testing with some low-resolution F.E.A.R. tests. Despite the fact that this is a game benchmark that can be used to test the relative performance of video cards, frame rates are strongly influenced by processor speed and available memory bandwidth, especially at low resolutions, which is how we ran the tests to get the frame rates listed below.
The CPU's clock speed also played a major role in our low-resolution F.E.A.R. benchmarks, wher the OCZ OCZ3P1600EB2GK and Super Talent W1600UXG7 finished first and second, respectively. This test is also heavily influenced by CAS latency, which the Kingston vs. Kingston comparison proves. The CAS 5 KHX11000D3ULK2/2G kit finished the test 17 FPS ahead of the CAS 7 KHX11000D3LLK2/2G kit. |
| Overclocked Performance | ||||
|
For our next set of numbers, we focused on the maximum stable overclock of each memory kit while set to their respective rated timings. Using the same Core 2 Duo E6750 CPU, we raised the Front Side Bus speed while concurrently lowering our processor's multiplier. We tried to keep the CPU's clock speed as close to the same speed as possible for each kit. For these tests, we locked the PCI Express clock to 100MHz, raised the CPU voltage to 1.4v, lowered the CPU multiplier to 6x or 7x, and kept the memory voltage to 2.2v for all of the kits. Ultimately, we ended up with different settings for each kit, which was unnavoidable. Here is the exact breakdown of how each kit fared in the overclocking tests:
As you can see, we have an approximate 300MHz swing in CPU frequency between the kits. As for the memory frequency, the Corsair kit topped out at over 1.9GHz, followed by the OCZ kit at 1.73GHz, Super Talent at 1.71GHz, and both Kingston kits at 1.47GHz. Please keep in mind, these speeds were attained at each kits default timings. Had we raised (or lowered) the CAS latency for each kit here, their peak overclocked speeds would have been different.
All of the memory kits showed respectable increases in bandwidth while overclocked. The Corsair kit broke the 9.2GB/s mark, while the others hovered in between about 7.4GB/s and 8.4GB/s.
Overclocking the kits also had a beneficial impact on latency. While overclocked, each kit shaved a few nanoseconds of their latency scores, with the Super Talent W1600UX2G7 showing the most improvement (-5ns). |
| Overclocked Performance (Cont.) | ||||||||
|
We also re-ran the PCMark05 memory performance and F.E.A.R. benchmarks again with all of the memory kits overclocked on our Intel Core 2 Duo E6750 powered test bed...
As you probably expected, PCMark05 also reported substantial gains for all of the memory kits we tested while overclocked.
Due to its significant increases in CPU and memory frequency, the Corsair TWIN3X2048-1800C7DF kit showed a huge improvement in the F.E.A.R. benchmark, besting its stock score by 59 FPS. And because it was the highest clocked kit of the bunch, it put up the fastest overall score as well. |
| Our Summary and Conclusion | ||||||||||||||||||||
|
Performance Summary:
At like clock speeds and latencies, most memory kits will perform at very similar, if not identical levels. But as clock speeds are raised, and hence available bandwidth is increased, or latencies are decreased, a clear performance pattern emerges. It's no coincidence that the fastest memory kit we tested here, Corsair's TWIN3X2048-1800C7DF also happens to support the highest official clock speed. Corsair's TWIN3X2048-1800C7DF also overclocked to the highest levels, easily breaking the 1.9GHz mark without relaxing its timings. In contrast, Kingston's low-latency CAS 5 KHX11000D3ULK2 kit showed a marked performance improvement over the other DDR3-1333 kit with a CAS latency of 7, but even with timings that low it couldn't catch up to the higher clocked kits.
Corsair TWIN3X2048-1800C7DF:
With a rated speed of 1800MHz, the Corsair TWIN3X2048-1800C7DF is one of the highest clocked DDR3 memory kits currently on the market. And despite supporting an already high default clock speed, it also proved to have plenty of overclocking headroom left and ultimately put up the best performance overall. All that performance comes at a price, however. At about $640, the Corsair TWIN3X2048-1800C7DF kit is also one of the most expensive. That's a large pill to swallow for a 2GB memory kit, but in the current DDR3 landscape you do get what you pay for. If money is no object and you're building yourself a high-end Intel-based rig, you can't go wrong with the Corsair TWIN3X2048-1800C7DF kit - it's fast, highly overclockable, and backed by one of the most respected companies in the business.
OCZ OCZ3P1600EB2GK:
In typical OCZ fashion, the OCZ3P1600EB2GK memory kit strikes a nice balance between performance and price. While still expensive in the grand scheme of things (the fastest DDR2 memory kits are available for a fraction of the price), the OCZ3P1600EB2GK kit represents what we think is the best value here, relatively speaking of course. For about $550, the OCZ3P1600EB2GK kit offers solid rated performance, somewhat lower latencies, and a good amount of overclocking headroom. Corsair's kit hit higher clock speeds, and one of Kingston's offers tighter timings, but the OCZ3P1600EB2GK kit hits somewhat of a sweet spot in between and should be considered for your next build should it require DDR3.
KIngston KHX11000D3LLK2/2G:
It's easy to let the Kingston KHX11000D3LLK2/2G memory kit get lost in the mix. It doesn't offer the highest clock speeds, nor the tightest timings, and its performance in the overclocking department left something to be desired. Looking at the KHX11000D3LLK2/2G kit's specifications and performance results alone, there is no one area where kit truly shines. But Kingston knows this too, and its the reason why the KHX11000D3LLK2/2G kit is the least expensive of all the memory evaluated here, and by a large margin. At about $439 the Kingston KHX11000D3LLK2/2G kit at least $110 cheaper than the other kits we tested. If you want an affordable 1333MHz DDR3 memory kit, the KHX11000D3LLK2/2G will serve you well, but we'd recommended investing a bit more if you're going to the way of DDR3 at this time.
Kingston KHX11000D3ULK2/2G:
For non-overclockers, the Kingston KHX11000D3ULK2/2G is an intriguing product. With its low latencies and support for DDR3-1333 speeds, this kit offers great stock performance for those not looking to push clock speeds into the stratosphere. Unfortunately, the KHX11000D3ULK2/2G is hard to find at the moment, and even if you can find it, it's usually priced somewhere in the neighborhood of $700. That's simply too much to pay for a memory kit that can't hit nearly the same clock speeds of less expensive offerings.
Super Talent W1600UX2G7:
Super Talent has been making some noise in the enthusiast memory space as of late. And based on the performance of their W1600UX2G7 kit, the attention the company has been getting is well deserved. At approximately $559, the W1600UX2G7 kit, like OCZ's OCZ3P1600EB2GK, strikes a good balance in relation to its competition. The W1600UX2G7 kit also overclocked well and is readily available at many popular on-line outlets. If you're going to jump onto the DDR3 bandwagon, this memory kit is worthy of your consideration.
Discuss This Article and Share Your Thoughts With Others!
|
;)
;)
We've talked about Corsair's Dominator memory kits and their unique cooling in the past, so we'll paraphrase a bit here. Although based on DDR3 memory, the TWIN3X2048-1800C7DF kit shown here shares the same cooling technology as its DDR2-based Dominator branded counterparts. There are four key things at work in Corsair's Dominator memory kits that result in the product you see pictured above. With the Dominator series, the memory's PCB has been heightened and there are thermally conductive vias and traces running through the PCB and connected to the ground plane, to help dissipate heat from within the board itself. Heatsinks are then bonded to both sides of the upper portion of the PCB to further aid in heat dissipation.;)
;)
;)
;)
