Should I RMA this kit?

[Kantastic]

Some of you may know that I've been having issues running my Kingston HyperX 2000MHz kit of RAM any higher than 1600. It will not even run at 1800 stable (as in it won't even successfully load Windows).

I'm running it (seemingly) stable @ 1600 6/7/5/60/17/1T right now, but I feel that this should be expected for a kit rated at 2000MHz. I'm not sure if it's a bad kit or not since it passes 2-3 hours of Windows memtest86 without problems.

It won't boot at 19XX-2XXX, haven't tried 17XX-18XX.

I've tried:

DRAMv: 1.66 (1.59@1600)
Timings: Up to 11/11/11/30/3T

To RMA or not? Am I doing something wrong or is it actually the RAM that's faulty?

 

[LittleLizard]

i wont rma it as there is very few gain from 1600 with low latencies to 2000 (if its any gain). so unless your epeenes wants to say to your friends "i have an i7 with 2ghz memory" then no

 

[TheLaughingMan]

I would have to concur. I have seen different in timing and speed during benchmark and RAM tests, but very, very little real world performance. I have argued both sides of the fence about this and been proven wrong on both sides of the fence. Now I dance on it just go with the lowest timing I can get at the highest clock speed without creating an error.

I should lower the timing or up the clock on my current kit, but I haven't had time to do so recently with the new work schedule.

I mean 6-7-6-18 @ 1600Mhz is pretty damn good. I mean the kit maybe able to do 2000 Mhz in an AMD rig where you can up the voltage to 1.7 and up, but I don't know if you can test that theory.

 

[Kantastic]

I guess it's a keeper... even though it's rated @ 2000MHz on 1.65V, it's doing 1600MHz 6/7/5/17/60/1T stable on 1.59V. I lowered some other timings too, but those probably don't make much difference.

 

[Binge]

You do know that running ram at 2000MHz is actually quite a feat even if the sticks are good. You have to volt your CPU's qpi even harder to handle it. There are other tricks as well.

 

[Kantastic]

Care to let me in on some Bingey?

 

[Binge]

Care to let me in on some Bingey?

No.

 

[Kantastic]

D: That was blunt.

 

[Jstn7477]

I guess it's a keeper... even though it's rated @ 2000MHz on 1.65V, it's doing 1600MHz 6/7/5/17/60/1T stable on 1.59V. I lowered some other timings too, but those probably don't make much difference.

Much better than my G.Skill DDR3 1600 kit's 9-8-8-24/1T @ 1.6v. I would be happy to have your RAM.

 

[Binge]

D: That was blunt.

Hahahaha, I'm so E-vile. Alright so some tricks..

#1) QPI voltage at least .1V higher than your average. Most people recommend 1.4V for 2GHz memory

#2) DRAM voltage at 1.65-1.7V but no higher

#3) Uncore multi at x2 memory multi +1. Let memory multi = M. 2(M)+1 = new uncore multi

#4) Don't forget your VCore, it might need to be increased as well.

#5) Increase QPI pll and IOH vcore slightly. I believe stock is 1.12V or 1.2V. 1.25-1.3V would be no issue if you have a good mobo. Only use this if you find the previous steps did nothing.

 

[Kantastic]

Hahahaha, I'm so E-vile. Alright so some tricks..

#1) QPI voltage at least .1V higher than your average. Most people recommend 1.4V for 2GHz memory

#2) DRAM voltage at 1.65-1.7V but no higher

#3) Uncore multi at x2 memory multi +1. Let memory multi = M. 2(M)+1 = new uncore multi

#4) Don't forget your VCore, it might need to be increased as well.

#5) Increase QPI pll and IOH vcore slightly. I believe stock is 1.12V or 1.2V, 1.25-1.3V would be no issue if you have a good mobo. Only use this if you find the previous steps did nothing.

1. I've done up to 1.65V QPIv, no luck. (This was default when I used XMP.)
2. DRAM voltage I've done up to 1.66, no luck. (I was afraid to go any higher, it won't even load Windows @ 19XXMHz with 1.66, and this is with 11/11/11/30/3T.)
3. I'm lost, all I know is my uncore is on auto.
4. Your i7 clocks better and on lower voltage than mine, mine needs 1.288 to be stable @ 4.0.
5. I'll leave this as a last resort.

 

[Binge]

On uncore multi... the uncore multi is always 2xDRAMmulti.

Let's make a model

200BCLK x21 for the CPU

Memory is 2000MHz at 200BCLKx10

Uncore would then be 200[2(10)+1] or an uncore multi of 21 making the uncore 4200MHz

On a later note... never EVER run your qpi at 1.65V ever again. You will fry something with that kind of voltage. It's not even really supposed to go above 1.35V but people still push it to 1.45V at the most.

 

[Kantastic]

So...

2:8 would be 17 -> 3400
2:10 would be 21 -> 4200

I'll give that a shot like now.

 

[Binge]

So...

2:8 would be 17 -> 3400
2:10 would be 21 -> 4200

I'll give that a shot like now.

Correct sir!

 

[Kantastic]

No luck Binge! =[ I pumped the QPI up to 1.35 and returned the timings to auto, bumped the MEM up to 2000MHz, uncore freq. @ 4211 and after exiting out of the BIOS my LCD Poster/Display thing was stuck on "DET RAM", manually powered down then powered back up and got an overclock failed message. =[

 

[Binge]

honestly... I just noticed you have a GENE and that board, for me, had the weakest ability to handle ram out of any of my x58s.

I run my 2000MHz ram at 1600 cas 6-8 depending on the benchmark, and the numbers didn't really differ between the two. I'm glad you've got a set that runs timings tight with low voltage. In the end running it like that will save you heat and energy.

 

[PaulieG]

honestly... I just noticed you have a GENE and that board, for me, had the weakest ability to handle ram out of any of my x58s.

I run my 2000MHz ram at 1600 cas 6-8 depending on the benchmark, and the numbers didn't really differ between the two. I'm glad you've got a set that runs timings tight with low voltage. In the end running it like that will save you heat and energy.

This is what you should pay attention to. My experience has been like Binge's. Very little, if any difference b/t 1600 and 2000, especially with timings as tight as you are getting. Not many DDR3 sticks can get that low.