Introduction : DDR4
Time Changes, so does technology; especially in the PC components arena.
While on one hand some of the components gets updated l every 8 months or so, a few of them have a longer development and implementation cycle like RAM for example.
In this review we will be taking a look at one of the newest standards in RAM’s; the DDR4 memory.
Now what is DDR4? in technical terms DDR4 is “double data rate fourth generation synchronous dynamic random-access memory” in simple laymans terms its an update to the existing DDR3 rams both in physical aspects and in terms of specifications.
On the physical side a DDR4 DIMMS has 288pins as opposed to DDR3’s 240 DIMMS.
Also the DDR4 Dimms have much higher bandwidth and runs at a much lower voltages than DDR3 Modules.
Take a look at the small info sheet below then we will take a look at the kit at hand.
So in short we have a DIMM that’s runs faster..runs cooler and packs more punch. Kind of an win-win situation for consumers..isnt?
Our coverage of DDR4 will begin with a a sample from Kingston .
We would also like to thank Asus India for providing us with a proper test platform.
Close look and Test system.
Even though we covered the kit in our un-boxing here there’s no harm looking through it again.
So here goes.
Kingston sent their HX430C15PB2K4/16 kit.
So what does the whole code means? “HX” stands for HyperX “4” for DDR4 “30” for 3000Mhz “C15” for the CL rating.”PB” is Predator Black, and the “K4/16″ is for a kit of 4 totaling 16GB/
The kit we have is the first lot out of the factory and as such these kits had the old JEDEC timings programmed in it and as such booted up at 1.5v which is considerably higher for a DDR4 kit. upon further investigation we found that only the sample lot has this problem in SPD programming. The retail kits will be programmed to boot up at 1.5V. SO if you decide to buy this kit don’t be shocked to see the rated voltage.
Also please note the the part number is not correct in the packaging; all retail kits will come with the number HX430C15PB2K4/16.
Now to the kits itself.
The heat spreaders are black in color which looks classy in my opinion.However this heat-sinks are pretty high at about 55mm; while the look very space age type i still don’t get it it why memory makers prefer to have so big heat-sinks. the kits in reality doesn’t produce that much of heat to warrant these heat spreaders also they don’t really fit under heavy tower type cooler which effectively narrows down the target customers/users of these kits. Its good to have options, the sooner the memory manufacturers realize these the better.
Out side the heatspreaders there isn’t much to tell the difference between a DDR3 and DDR4 module except the numbers of contact pins(288 in DDR4 as opposed to 240 in DDR3)
All in all a very good looking set of kits, just a little too high.
Benchmarks and Overclocking
Now to the its time to see the numbers.
Firstly i would like to thank Asus India for providing me with a DDR4 test platform to test these kits.
The test setup we used for this review are as follows
- Core i7 5960X courtesy Asus India
- Asus X99 Deluxe courtesy Asus India
- Kingston HyperX Predator 4x4GB DDR3 3000 MHz @ 1,35v provided by Kingston.
- AMD R9 290X
- Corsair Nova SSD
- Coolermaster V1000 PSU
- Custom water-cooling kit.
For benchmarks we ran a mix of workloads which puts considerable amount of stress on memory.
the benchmarks used are Aida 64 memory suite, Wprime 32M and 1024M. these two benchmarks clearly shows the performance scaling of memory subsystem across the board.
Overclocking the memory was a mixed bag.While the kit itself needs 125Strap on the motherboard to do 3000 MHz. We couldn’t do more than that. We could do 166mhz on the motherboard itself but the ever elusive 3334 MHz barrier remained..elusive.
However, we had fun with the timings.
here’s a list of what we could do with the kit along with cpu-Z validations.
2666Mhz @ 14-14-14-36
Validation link : http://valid.canardpc.com/as1zrw
2666 13-14-14-28
Validation link : http://valid.canardpc.com/zcfe00
2800Mhz 14-15-15-35
Validation link : http://valid.canardpc.com/qj537n
3000Mhz 14-15-15-37
Validation link : http://valid.canardpc.com/g28xm6
3000Mhz 15-16-16-39
Sadly we lost the screenshots of the scores here due to a storage crash, so no screenhost for this speed but we do have a validation link.
Validation Link : http://valid.canardpc.com/g5zkk6
Now that we have seen the clocking potential of the kit which is somewhat of a mixed bag. Lets see how the performance number stands.
First up,Aida 64 Memory Suite.
Please Note : for the first three graphs, higher scores means better and for the next three graphs lower score indicates better performance.
Aida Memory Read
Aida Memory Write
Aida Memory Copy
Aida Memory Latency
Here we can see the 2800MHz 14-15-15-35 providing the maximum performance across all the settings.
WPrime 32M
Here also we see the 2800MHz 14-15-15-28 setting working out well.
WPrime 1024M
However with WPrime 1024M 2666 Mhz 13-14-14-28 setting manages to produce a lower score.
Conclusion
Its pretty early for the era of DDR4, so we should see some changes in the specs of the Kits coming to retail in future.
The HyperX predator kits comes as a pretty fast kit out of the box although it disappoints a little when it comes to bumping the frequency upwards of 3000 Mhz. However that’s not enough to call it as a con for the kit since this is the first kit TechARX received for review and we don’t have enough data from other kits to compare.
However i have to point out that the retail availability of this kit is next to none so if you want to order this kit(which i think you should unless you are putting in big air-cooler on your DDR4 setup) you might have to wait out a bit.
EDIT : Kingston has intimated us that the product is available for purchase in Flipkart.
So if anyone wiseh to purchase it;which i think they should since its a very good kit.
They can purchase it from the following link : HyperX Predator DDR4 16 GB PC RAM (HX421C13PBK4/16)
All in all a pretty decent kit from Kingston; now only if they could do away with the height of the heat-sinks.
