With the event of Intel's extremely overclocker friendly P35 chipset and the impending release of the X38 chipset, many people are having a hard time deciding whether their next PC upgrade should be DDR2 or DDR3 based. For those of you who are regular readers of Overclock3D reviews, you would have seen our recent DDR2 vs DDR3 review where we put the Asus P5K (DDR2) head-to-head with the Asus P5K3 (DDR3).
However, one thing we haven't taken a close look at yet is how DDR3 performs in terms of overclocking compared with its DDR2 counterpart. With this in mind, today we have decided to take a closer look at the Kingston HyperX PC3-11000 CL7 kit used in some of our recent motherboard reviews.
For most of us, Kingston needs no introduction. Founded in 1987 and one of the largest memory manufacturers in the world, Kingston offers over 2,000 memory products for use in anything from PC's to Printers. More importantly to us, Kingston have a large foothold in the PC enthusiast sector with their high performance HyperX memory modules. Having reviewed several of the DDR2 based kits in the past (PC2-9600, PC2-8500), the HyperX linup has always impressed. Today we're going to be finding out if Kingston have managed to continue this trend with their PC3-11000 CL7 rated HyperX DDR3 based kit.
KHX:DDR3 - It's Here — HyperX DDR3 memory, the next-generation evolution of DDR memory technology. Like all Kingston HyperX products, HyperX DDR3 modules are specifically engineered and designed to meet the rigorous requirements of PC enthusiasts. HyperX DDR3 offers faster speeds, lower latencies, higher data bandwidths and lower power consumption. HyperX is available in single and dual-channel memory kits.
HyperX DDR3 features: • 1.7 Volts operation • Memory signal termination inside the memory chip (“On-Die Termination”) to prevent reflected signal transmission errors • Operational enhancements to increase memory performance, efficiency and timing margins • CAS Latencies: 5 (Ultra Low Latency) and 7 (Low Latency) • Currently available in speeds up to 1.5GHz and capacities up to 2GB kits • DDR3 memory modules are not backward compatible to DDR2 and DDR based motherboards, due to incompatible module connections (number of pins), voltage and DRAM technology. DDR3 memory modules have a different key or notch than the same-sized DDR and DDR2 modules to prevent their insertion into an incompatible memory socket. HyperX is available in single and dual-channel memory kits.
As we can see from the specifications above, the HyperX PC3-11000 kit requires 1.7v in order to run at its rated DDR3-1375 speed and 7-7-7-20 timings. Unfortunately Kingston haven't specified whether the modules are guaranteed to run higher than 1.7v, forcing overclockers to invalidate their warranty if they want to push the modules further by increasing the voltage.
More detailed information on the HyperX PC3-11000 modules can also be found in the specifications PDF available on their website. A snippet of this information is shown below:
FEATURES: • JEDEC standard 1.5V ± 0.075V Power Supply • VDDQ = 1.5V ± 0.075V • 533MHz fCK for 1066Mb/sec/pin/ • 8 independent internal bank • Programmable CAS Latency: 5,6,7,8,9,10 • Posted CAS • Programmable Additive Latency: 0, CL - 2, or CL - 1 clock • Programmable CAS Write Latency(CWL) = 7(DDR3-1066) • 8-bit pre-fetch • Burst Length: 8 (Interleave without any limit, sequential with starting address “000” only), 4 with tCCD = 4/ which does not allow seamless read or write [either on the fly using A12 or MRS] • Bi-directional Differential Data Strobe • Internal(self) calibration : Internal self calibration through ZQ pin (RZQ : 240 ohm ± 1%) • On Die Termination using ODT pin • Average Refresh Period 7.8us at lower then TCASE 85°C, 3.9us at 85°C < TCASE . 95°C • Asynchronous Reset • 1066Mbps CL7 doesn’t have backward compatibility with 800Mbps CL5 • PCB : Height 1.180” (30.00mm), double sided component
PERFORMANCE: • CL(IDD) - 7 cycles • Row Cycle Time (tRCmin) - 50.63ns (min.) • Refresh to Active/Refresh Command Time (tRFCmin) - 90ns • Row Active Time (tRASmin) - 37.5ns (min.) • Power TBD W (operating) • UL Rating - 94 V - 0 • Operating Temperature - 0o C to 85o C • Storage Temperature -55o C to +100o C
Most of Kingston's previous DDR2 HyperX kits have been based on Micron's notorious D9 IC's, however from the image above we can see that Kingston have made a switch to Elpida for their DDR3 modules. This isn't neccesarily a bad thing as DDR3 is a totally different animal to DDR2, and there's every chance that the Elpida IC's may perform as good or better than their Micron DDR3 counterparts.
Kingston seem to be one of the very few memory manufacturers that use packaging of their own design. This is certainly a good thing as the generic moulded plastic blister pack style packaging adopted by most other manufacturers offers minimal protection for the memory at best.
As we can see from the images above, the KHX11000D3LLK2/2G kit comes in a robust box constructed from anti-static plastic with a clear plastic lid. This not only protects the modules from damage (both electrostatic and physical) but also gives good visibility of the modules inside the packaging.
Since the release of the HyperX series, Kingston has always made use of distinct blue aluminium heatspreaders. Interestingly the design of the DDR3 range shown on their website differs somewhat from the 'original' styled modules we received. I can only assume that Kingston were keen to get their DDR3 modules out on the market and decided to leave the redesign of their heatspreaders to a slightly later date.
Looking back at the previously reviewed HyperX PC2-8500 and PC2-9600 DDR2 kits, the only real difference in appearance is the inclusion of the 'DDR3' logo at the left of the modules. It certainly is a shame that Kingston couldn't bring the new design heatspreaders shown on their site together with the release of these modules as it would have given us something slightly more exciting to talk about.
As with the previously reviewed HyperX kits, the PC3-11000 makes use of double sided thermal tape coupled with chrome clips to hold everything in place. I actually found it extremely difficult to remove the heatspreaders from the modules, resorting to the use of a small screw driver to seperate the thermal tape from each of the memory IC's.
Being the first DDR3 memory review here at Overclock3D the Kingston HyperX PC3-11000 kit will be reviewed as a standalone product using the hardware specified below. However, as further DDR3 reviews are added to Overclock3D, the results from these reviews will be included on the graphs shown over the next few pages.
Kingston HyperX PC3-11000
Intel Core2Duo E6850
Asus Blitz Extreme (P35)
BFG 8800GTS 640mb OC2 Edition
Hitachi Deskstar 80GB 7K80 SATA2 7200RPM 8mb
Stock Intel Aluminium Cooler
Microsoft Windows XP (SP2) 32bit - Latest Patches
To guarantee a broad range of results, the following benchmark utilities will be used:
As with all memory reviews on Overclock3D the HyperX PC3-11000 modules were subjected to a series of benchmark tests under several different configurations. The configurations used tested the memory's ability to run at high frequencies, tight timings and also ensured that that the modules could run with 100% stability at their advertised stock settings.
Stock - 1375mhz / 1.7v / 7-7-7-20
It is an unfortunate fact that many memory modules we've tested here at Overclock3D in the past have fallen at the first hurdle by not being able to run at the advertised stock speeds. Therefore as part of our standard testing procedure, all memory modules are subjected to extensive stability testing using Prime95 and SuperPi.
Thankfully the Kingston HyperX PC3-11000 kit did not suffer with this problem, managing to run at the advertised settings of 1375mhz / 7-7-7-20 / 1.7v without any stability issues whatsoever.
Low Latency - 1375mhz / 1.8v / 6-7-7-20
Bumping up the voltage to 1.8v allowed for the modules to be tightened down to 6-7-7-20. Unfortunately any attempt to further reduce the latency down to 6-6-6-20 or 6-7-6-20 including increasing the voltage to 1.9v yielded no success. I can only assume that the Elpida IC's used on the HyperX kit do not respond well to increased voltages and/or tight timings.
Max Frequency - 1472mhz / 1.8v / 9-9-9-20
With the voltage still set to 1.8v and the timings relaxed to 9-9-9-20, the furthest the HyperX kit could be overclocked to while still maintaining stability was 1472mhz. Unfortunately, as this is the first DDR3 review conducted on Overclock3D it's hard to comment on how this 97mhz overclock compares to kits from other manufacturers.
Sisoft Sandra is a synthetic benchmark utility capable of reporting and benchmarking a wide range of system components. For the Kingston HyperX PC3-11000 memory we run both the Memory Bandwidth and Memory Latency benchmarks 3 times to ensure accuracy of results.
Everest is in many ways similar to Sisoft Sandra. Focusing mainly on Software and Hardware information reporting, Everest also comes with a benchmark utility suitable for testing the read, write and latency performance of the memory subsystem.
SuperPI is the benchmark of choice for many overclockers. It's lightweight to download and can give a quick indication on how good a system is at number crunching.
As we can see from the results above, reducing the memory timings down from 7-7-7-20 to 6-7-7-20 offered a reasonable performance boost in both memory bandwidth, latency and SuperPI results. However, the SuperPI results were overshadowed by figures produced by the memory running at 1472mhz. Even with extremely relaxed timings of 9-9-9-25, the extra 97mhz overclock gave a fairly impressive 2 second lead.
3DMark is a popular synthetic gaming benchmark used by many gamers and overclockers to gauge the performance of their PC's. All 3DMark runs were performed 3 times with averages being calculated from each of the results.
F.E.A.R. is a game based on the Lithtech Jupiter EX engine. It has volumetric lighting, soft shadows, parallax mapping and particle effects.Included in the game is a benchmark facility that taxes the entire PC system. This benchmark was run 3 times to ensure uniformity of results.
Quake 4 is a game built on the Doom 3 engine. Benchmarking was performed using Quake4Bench and a custom timedemo recording along with 4xAA, 8xAF settings at a resolution of 1280x1024.
Counter-Strike:Source is a popular multi-player FPS based on the extremely scaleable Source engine. The game takes advantage of many DirectX 9.0c features, but is fairly undemanding on the GPU and tends to thrive on systems with fast CPU's and Memory.
Unlike the application benchmarks over the previous page, the performance difference between 7-7-7-20 and 6-7-7-20 goes totally undetected in Quake4, Fear and 3DMark06. This just goes to show that the increased bandwidth and lower latency figures shown by Sisoft Sandra and Everest with the memory running at CL6 don't neccesarily equate to an increase in real-world performance. However, with the memory overclocked to 1472mhz and the latencies relaxed to 9-9-9-25 we can see a small FPS advantage in CS:Source, FEAR and Quake4 with a fairly significant points advantage in 3DMark05/06.
With new DDR3 kits hitting the market on an almost daily basis, it's an unfortunate fact that the Kingston HyperX PC3-11000 CL7 kit has already been superseded by insanely fast PC3-14400 and PC3-16000 kits. In fact, shortly after we started this review, Kingston refreshed their DDR3 HyperX lineup, adding a PC3-11000 CL5 rated kit. However, taking into consideration that the KHX11000D3LLK2 HyperX kit on review today is the first DDR3 release by Kingston, the 97mhz overclock obtained on the modules by relaxing the timings to 9-9-9-25 is certainly not a bad result.
Priced at ~£280 from retailers such as CCL Online, the Kingston kit is very well priced when compared to similar offerings from Corsair and OCZ. Unfortunately availability in the UK is limited at present, but for those of us in the USA the HyperX kit can be obtained from Kingston's very own shop.
The Good • Competitive pricing compared to similar kits from other manufacturers. • Reasonable performance at stock and overclocked settings.
The Mediocre • Ability to tighten timings slightly to 6-7-7-20. • Reasonable 97mhz overclock obtained by loosening timings.
The Bad • Does not respond very well to additional voltage. • Limited UK availability.
Thanks to Kingston for making this review possible. Discuss this review in our forums.