Asus ME and OCZ Platinum 1600 (D9JNL)

[chris.y2k.r1]

I could use a little advice. My rig is Asus Max. Extreme, E8400 and 2x1GB DDR1600 OCZ Platinum (D9JNL). My problem is this. I have 3 of these at this moment. Well, I have 3 that work and 1 bad one. I have to return 2 of them to OCZ. So I need to return 1 bad and 1 good.

Now, 1 stick by it self with do 800 7 7 7 20 @ 1.96 no problem. But as soon as I install the second stick (Either of the two - or any combination of the 3 sticks at all) I cannot even manage 750 7 7 7 20 @ ??? for more than an hour in OCCT without failing. 1 stick will run for over 12 hours in OCCT. As soon as I install the second the problems start. So, any advice? What gets stressed when you insert the second stick? The memory controller on the NB? I've tried many different settings but I'm still struggling getting it to simply run at stock speeds. Any help would be appreciated.

Settings are (From memory...)

DDR 1501
2T
7 7 7 20 60 10 5
8 5 6 4 7 4 7
DRAM Stat Read - Enabled
Dyanmic Write - Disabled
Clock Twister Light
Channel Skew A & B - Auto
Trans. Booster Disabled 3

CPU 1.4575
PLL 1.68
NB 1.61
FSBT 1.54
VDIMM (Anywhere from 1.78 to 1.96) Right now I think I have it at 1.86
Load Line: Enabled
CPU GTL .63x
NB GTL .67x
DDR3 Channel A Auto
DDR3 Channel B Auto
DDR3 Controller Auto
SB 1.5V Auto

Like I said, I can run 1 stock at the 'stock' speeds of 1600 7 7 7 20 with 1.96v (I know that's probably almost 2v with this board, I have the 1201 BIOS and it does not over volt as bad as the earlier BIOS' did. Only by about .02v according to the meter readings I've taken). As soon as I plop in a 2nd stick - then I can't run stock speeds and be OCCT stable. I'd like to know what happens that stresses the system so bad when you add the 2nd stick.

Thanks in advance for any info.

Chris

[RyderOCZ]

Try disabling Static Read

Set the Skew to 150 advanced on both channels...try up to 300 advanced.

Set Clock Twister and AI Trans booster back to auto for now. If necessary, test AI Trans booster at disabled Relax Level 0 or 1.

Have you tested the other pair of slots to see if there is any change?

[chris.y2k.r1]

Try disabling Static Read

Set the Skew to 150 advanced on both channels...try up to 300 advanced.

Set Clock Twister and AI Trans booster back to auto for now. If necessary, test AI Trans booster at disabled Relax Level 0 or 1.

Have you tested the other pair of slots to see if there is any change?

Heya Ryder,
Thanks for the quick reply. By the way - what do you ride? I just sold my R1, I miss it, but there no place to ride where I luve now.

Yes, I tried the other slots. The white ones. No difference at all. I'll try what you suggested though.

What is Dynamic Read/Static Write for anyway?

[chris.y2k.r1]

No luck. The only way I can even get these things into windows at DDR3 1600 7 7 7 is by relaxing the transaction booster level to 8. And even then I can't complete even as little as an hour of OCCT before they crap out. 32M Super PI takes down the OS entirely. I have one stick that I can run at the rated speed and it takes 1.96 volts but is 12Hours OCCT stable.

I just got these as an RMA replacement for a previous set. What's going on with the binning over there...? I kid you not when I say this is either the 5th or 6th set of memory that I've received consecutively (OCZ memory) that simply could not run at it's rated speed. It got so bad that I gave up on the OCZ DDR2 and took a trade on a power supply instead. The power supply has been great (I have two the 600 and the 1010 and they are excellent) - but the memory.....

As a matter of fact, the system in which I was trying to get the OCZ DDR2 to run at proper speed in, currently has cheapo Super Talent DDR533 Hynix chips in it running at DDR1000 5 6 5 at 2V. I could not get the Flex or the Reaper that were both rated above DDR2 1000, to run anywhere near that (IE the 1000 5 6 5). So, I know it's not my system, as I do in this case as well. What's going on over there man? Am I just that unlucky with the OCZ memory or is the binning process being poorly managed...? Rhetorical question. What can I do... I mean, if I try yet another RMA, who's to say I'm actually going to get memory that works as it's supposed to..? Like I said, I must be very unlucky but I hope you can understand my frustration... What should I do Ryder? (FYI, both systems I speak of are in my sig).

Thanks! Seriously, thank you for trying to help. It's appreciated.

[RyderOCZ]

OCCT failing is not always memory.

Did you test the other stick singly as you did the first one? Does each stick run at 1600, but just not together?

[chris.y2k.r1]

OCCT failing is not always memory.

Did you test the other stick singly as you did the first one? Does each stick run at 1600, but just not together?

Oh I know it's not.

No, I have not yet had time to test a single stick. What I have done is put in the known good stick with one and then the other two replacement sticks. I will try running the other two by themselves to see if I can get them to run singularly.

Let me ask, if they do run each OK by themselves - then what does that indicate the problem might be??

P.S. I've tried a variety of NB, CPU, VDIMM, PLL and FSBT voltages to no avail. CPU is definitely right at the sweet spot. It doesn't like more or less really. And FYI so far the highest NB temp I've seen has been 44c. That shouldn't be hot enough to cause problems.

Thanks again.

[RyderOCZ]

How fast are you running the CPU?

[chris.y2k.r1]

4.5GHZ 5x900 24/7.

P.S. - Your sig is PRICELESS!!!!

[RyderOCZ]

How are you getting 1600 MHz ram with the FSB at 500?

Have you tried 400 FSB with a 7 CPU multi (to keep CPU near stock) ram set at 1600 with the right timings and voltages?

[chris.y2k.r1]

How are you getting 1600 MHz ram with the FSB at 500?

Have you tried 400 FSB with a 7 CPU multi (to keep CPU near stock) ram set at 1600 with the right timings and voltages?

Waddya mean how am I getting 1600 @ 500?? 500x9 with 333 strap.

No, I haven't. I have no intention of running the CPU at stock speed. I have some pretty extreme water so that would be pointless or rather a waste of $$. However, if you are suggesting I do that as a test - Sure, I can do that.

[RyderOCZ]

Yes I am suggesting you do it as a test. Both with 1 module at a time and with 2.

You are using the 1066 divider on the 1333 strap, gotcha.

[chris.y2k.r1]

Yes I am suggesting you do it as a test. Both with 1 module at a time and with 2.

You are using the 1066 divider on the 1333 strap, gotcha.

OK - NP, I will try that. I'll get back to you in about 5 hours as I don't have access to the system here at work. I have 1pm here EST, so I'll get back to you around 6pm EST - I suppose that would be around 3PM your time... correct?

[RyderOCZ]

Eastern time here.. 6pm for you is 6pm for me.

[chris.y2k.r1]

Ok - Sticks act the some in any order. 1, the other, or 2.
However I can get 1 of my old sticks stable at 1.96v 12 hours OCCT. I am about to try 400x7 and see what that does. I will report back later.

Question - Most of my expirience prior to this machine is with AMD. I know with AMD, when you add a second stick of RAM you need to up the CPU voltage due to the IMC. I would suspect that with Intel the same would be true accept you would up the NB juice and not the CPU. Is this typically true??

[RyderOCZ]

I have not found it necessary to add juice to the northbridge just because 2 sticks are in place.

However you are overclocking the NB at 400 and especially at 500, so yes more NB voltage should help.

[chris.y2k.r1]

I have not found it necessary to add juice to the northbridge just because 2 sticks are in place.

However you are overclocking the NB at 400 and especially at 500, so yes more NB voltage should help.

I've noticed that this board doesn't like anything over 1.63 NB for 500. It gets unstable over that even with NB temps in the 30s.

OK - A little update, I played with it this morning. I actually had trouble trying to back down to stock speed or rather 400x7. While getting it stable at the low speeds I managed to curropt my Vista partition so I had to restore it from Acronis. So, I was not able to test the memory at stock CPU speeds.

However, it definitely appears that the more I slow the memory the longer I can run OCCT.

So, any other thoughts at this point? I will try again to run the CPU @ stock later tonight. It just so happened that it didn't want to adjust at first. Also I am running BIOS Rev 1004. I was thinking of trying 1007, 1101 or perhaps 0905 as it seems people are having luck with that older one.

.... Hummmm ..... I know this board is picky but I had the damn thing good to go until the one stick got borked! Dang it!

[RyderOCZ]

Slowing the memory, lowers the total load on everything.

You have already RMA'd the memory and still have same/similar issues. Do you have other memory that runs with no issue at 1600? and with the overclock you want?

You have to play in the limits and running the CPU at 4G with ram at 1600 or more...is really pushing the limits.

[chris.y2k.r1]

Slowing the memory, lowers the total load on everything.

You have already RMA'd the memory and still have same/similar issues. Do you have other memory that runs with no issue at 1600? and with the overclock you want?

You have to play in the limits and running the CPU at 4G with ram at 1600 or more...is really pushing the limits.

I hear you. My issue is I always have this problem with OCZ memory. I figure, I'm not looking to run the memory outside it's rated clock/latencies. I realize that a faster CPU is going to hit the memory with requests faster. Problem is the sticks I had before these two were able to handle it. One just quit on me @ stock everything @ 1.96V (BIOS Indicated). So, I hear what you are saying but I think I should be able to make this happen.

I realize slowing the memory eases the system. Like I said, I was able to handle these speeds with the 1 good stick that I had left - 12hours OCCT no problem. I can't see why these two new sticks won't do that if the previous set would. Oh well - look, if you are out of ideas then I'll just keep at it. If I get somewhere I'll let you know. For me, I think I just need to stick with OCZ PSUs (Love 'em) and get other brands of memory. Hey, it's just my luck. Most reviews seem to favor OCZ so the only common denominator is me.

Thanks Ryder.

It's all good :)

[RyderOCZ]

Also been wondering why so much voltage... 1.9V is rated voltage...should not take more than that.
Have you tried less voltage?
What about Trfc...how do you have that set?

What Perf level are you getting/setting?

[chris.y2k.r1]

Also been wondering why so much voltage... 1.9V is rated voltage...should not take more than that.
Have you tried less voltage?
What about Trfc...how do you have that set?

What Perf level are you getting/setting?

Well... As for voltage I've done alot of testing:

As for the one original stick that still works - At 1600 7 7 7, it won't post unless I give it 1.74v. It won't run OCCT stable unless I give it 1.96v. Now, oddly enough - that's with 1 stick. If I put 2 in, 1.96 is too much and causes instability.

AI Trans Boost - That's what you mean by performance level - right? The tightest I can set it at with this memory has been DISABLED LEVEL 3. The higher I go the more stability I get.

Now if you mean performance level as in MemSet performance level - that's set to 13. You see why I'm expecting a little more here? Right now I am at DDR1333 9 9 9 24 and I am under 1 hour OCCT.

tRFC I've set at 60~72 for current testing. However in the past I have set it as high as 110. I'm going to go back to 110 and see what I can do from there.

I'm going to back up. I am going to run the system at just a little over stock speeds and then work my way up with 20 MHZ increments. I'm hoping that by starting over and writing down my settings (Which I always do) - I can get this baby stable where I want it, or atleast close to it. I don't see why I shouldn't be able to. If I can't and the memory seems to be the problem, would it be possible to upgrade to something rated a little higher like the 2000 set of Plats in hopes that in would run between 1600 and 1800 7 7 7 ???

[RyderOCZ]

Disabled

Level 3.. that is why memset says 13.

Auto should give a Perf Level of 6 at 333 FSB. Then move to 7 as you clock up and 8, etc. That will all be determined by how Asus programmed the bios.

You get more stable by going higher Levels or lower?

[chris.y2k.r1]

Disabled

Level 3.. that is why memset says 13.

Auto should give a Perf Level of 6 at 333 FSB. Then move to 7 as you clock up and 8, etc. That will all be determined by how Asus programmed the bios.

You get more stable by going higher Levels or lower?

To answer that question I get more stable by going to higher levels of DISABLED unit AI TRANSACTION BOOSTER. Like I said, right now I am at 3....

AND...

I'm currently running OCCT for 40minutes (So far) at 1600 7 7 7 and CPU @ 7 7 7 20. Would you like to know what did it?

For some reason there seems to be a relationship betweem PLL and FSBT. I found that if I raise PLL and leave FSBT alone the system gets unhappy. However if I tweak both of them at the same time (Not necessarily a 1 to 1 ratio - but close), I remain stable. Anyway - Here's where I am at right now:

PLL is 1.68
NB is 1.61
VDIMM is 1.94 (Actual)
VFSBT is 1.54

CPU 500x9 DDR 1600 7 7 7 20,

OCCT Stable for (at the time of this post) 1 hour and counting....


F*&king Aaaaa! It's about time.

I should also mention that this memory definitely needed to be burnt in. At first I could only run it at 1600 8 9 9 then, slowly, without changing anything else I could tighten up the timings after a couple of hours of stressing it with OCCT or Prime.