FEATURED Kingston HyperX Predator 16GB DDR4-3000 CL15 - HX430C15PB2K4/16

[Woomack]

Today I have Kingston HyperX Predator 16GB DDR4-3000 CL15-16-16-39 2N 1.35V memory kit. I think it's interesting option on the quite limited DDR4 market.

Couple of photos at the beginning. I like how Kingston changed already well known Predator heatsinks.









Kingston programmed two XMP profiles in this memory so we can run it at DDR4-2666 CL14-14-14-36 or DDR4-3000 15-16-16-39. Both at 1.35V. We could expect to see one 1.2V profile but 1.35V is perfectly safe so I see no problem here. Actually higher voltages also won't damage memory. Of course there are limits

Memory is based on Hynix MFR the same as most available DDR4 at the moment but not all can make higher clocks.

XMP #1 - 3000 15-16-16-39 2N 1.35V




XMP #2 - 2666 14-14-14-36 2N 1.35V



Below also CPU-Z window with all profiles.

 

[Woomack]

Test Results

XMP #1 3000 15-16-16-39 2N 1.35V


XMP #2 2666 14-14-14-36 2N 1.35V


OC 3200 16-16-17-16 1N 1.35V


OC 3000 14-14-15-15 1N 1.45V


OC 3000 13-14-15-15 1N 1.55V



As you see, this memory can easily make DDR4-3200 clock at 1.35V. Even though timings look high then performance is good as limiting factor is mainly cache clock.

Performance of this memory is really good and also price isn't really high. DDR4-3000 C15 kit cost about $350-370 so not much more than most DDR4-2400/2666 kits.

I will try to add some more results but I have limited time as I have to send this kit back.

 

[EarthDog]

Good to know that, with the EVGA board at least, I am in the ballpark with my results...I can't say as much for the MSI/ASRock boards I used these on though.

Did you get a production run set of sticks (1.35v XMP)... looks like (mine was 1.5v). I am also able to go down to 1T on these throughout testing too even at 14-14-14-xx 1.35v.

The XMP profile I have also runs 15-16-16-35... they made some changes there and in other things (tFAW, tRFC,).. interesting.

Thanks for showing them!

 

[Woomack]

I got the latest sample from Kingston but there is limited quantity dedicated for reviews/tests in EU so I have to send it back.
As I mentioned in the 1st post, this memory has 1.35V profiles for 2666 and 3000. I saw that early kits had 1.5V but it wasn't really required. All of them are probably on similar IC.

 

[EarthDog]

They are... and they told me it was 1.35v but it was pre prod run that I have so I needed to manually set it. But yeah, it looks like they changed some sub timings from my XMP profile (as one would expect).

Im just glad it was the boards holding these things back and not the sticks, LOL! But yeah, when the EVGA review publishes, you will see things pretty much in line with yours at this point.

This kit is also on sale at newegg for $410...

 

[Woomack]

$365 on Amazon with free shipping I also saw it somewhere for ~$350 but don't remember where.

This memory is scalling almost like G.Skill Ripjaws 4 that I was testing before. Actually 3200 profile made on G.Skill works fine on Kingston and both perform the same.

I expect to see something like 2666 13-13-13/12-13-13 and 3000 13-14-14/14-15-15 depends from used voltage.

My setup has one weird issue. When I set higher voltage than 1.48 or 1.58V ( depends from some other settings ) memory is not working at 3200+ clock. What is weird I can run 3000 up to 1.72V without issues on 2 different kits. It's the main reason why 3200 result is at CL16. I bet this memory can make tighter timings at higher voltage.

 

[Witchdoctor]

Interesting stuff, thanks for taking the time Woomack

 

[Woomack]

I've added 3000 CL13 and CL14 results to the 2nd post.

Btw. WD, is your G.Skill kit still in RMA ?

 

[kamstka]

I achieve




Nothing more I tried to increase the voltage And timing
But nothing more than 3200 .

 

[Woomack]

Nice kit you got there. Mine needed higher voltage to run stable at 3200 and tighter timings but at the same time didn't like anything above ~1.45V so I couldn't stabilize it.
That 3200 CL13 is stable in any longer tests or only in AIDA64 ?
I have only 5820K for tests so bandwidth will be always lower than on 8 cores. Here you actually see that good, 4 more threads and ~20GB/s higher read bandwidth.

also to the OCF

 

[kamstka]

it stable in AIDA64 and Daily use Like :
Any Video Converter Ultimate and Xilisoft Video Converter Ultimate For hours
I noticed Any Video Converter
Very sensitive to Overclock cpu ram and Voltage ..
i Try to convert mkv "8G" to mp4 .. If the Program Crash After minutes before Complete
The Overclocking not full stable you need to increase the voltage Even if you pass AIDA64
You will face Crash from time to time In the system . . While if pass Any Video Converter
Will be full stable..

A week ago I have 5820K and Crucial 4x8GB
for cpu maximum reach 4.3 Anything more than that is not stable
same thing in Cache Ratio 4.1
While in memory maximum reach 2666 Anything more than that is not stable
and nything above 1.45V The system does not boot . .

 

[lanparty008]