Could my i7 be defective? memory controller?

[jay2tall]

So 2 weeks ago I bought upgrade parts for my PC. I purchased an Intel i7 920, ASUS P6T motherboard, and a 3x2GB G.skill DDR3 1600 8-8-8-21 memory kit. I am using a Thermalright Ultra-120 extreme 1366 cooler and running Vista Ultimate 64bit.

I had my mobo set to defualts except for using the XMP memory profile to get the rated memory clock speeds. When I run OCCT it blue screens with a memory error as soon at the test kicks in. This also happens on any game a minute or two into it. If I set my memory to the stock 1066MHz speed, everything is fine. I tried swapping in my buddies Corsiar Dominator 1866 memory (3 sticks) with the XMP profile and the same thing happened. We also tried to manually enter these settings and run at 1.65v. No luck. So I RMA'ed the motherboard, which I recieved the new one today. The same issue happens. Could this be my CPU? Could the memory controller be bad? I am new to the i7 arena so I am looking for help.

Thanks,

 

[cjl]

What QPI/DRAM voltage are you running?

 

[jay2tall]

Not sure. I am at work so I can't check. But I have everything to auto and let the XMP profile make any changes. I think the dominator memory had more settings it automatically changed, but I do not recall.

If you are refering to the DRAM voltage and not one of the other dozen. The XMP profile says 1.6 for the G.skill and was 1.65 for the corsair. I tried 1.65 with the g.skill as well and had the same issue.

 

[Stiffex]

Maybe you could run the newer memtest that has support for i7 http://www.memtest.org/ and see where that takes you.
Also see below:

Memory
According to Intel, the new Nehalem processors are specified up to a memory speed of DDR3/1066, while the current Core 2 architecture can be operated with DDR3/1600 memory. But according to the benchmark tool Everest 4.60, the internal memory controller supports up to 1333MHz. It could be that the system would not work stably in all situations at that frequency, so Intel opted for the more conservative specification. For optimal performance no more than three memory modules should be used. If four DIMMs are used, memory performance falls because the important memory parameter Command Rate can only handle two wait states.

You may want to do a search to see if you can find a cpu memory controller test before you rma your processor

 

[cjl]

Not sure. I am at work so I can't check. But I have everything to auto and let the XMP profile make any changes. I think the dominator memory had more settings it automatically changed, but I do not recall.

If you are refering to the DRAM voltage and not one of the other dozen. The XMP profile says 1.6 for the G.skill and was 1.65 for the corsair. I tried 1.65 with the g.skill as well and had the same issue.

No, I'm not referring to that one. It's a separate one, and it includes voltage to the memory controller. Mine ran fine at 1600-8-8-8-24 at 1.28V QPI/DRAM, and I currently have it set to 1600-7-8-7-22 (1T) at 1.35V QPI/DRAM - it wouldn't run stable at the tighter timings without a bit more power, although the RAM did fine at stock voltage. It should be in the XMP profile, but just in case that wasn't set properly, it'd be the next thing to check.

 

[jay2tall]

I find it very odd because my buddy can run his Dominator 1866 memory at 1950 at 1.65v and with his CPU at 4.0GHZ without a hitch.

However I am running 1600 memory and I can't even run stock.

 

[mamw93]

If the memory controller didn't work you would be having so operating system to boot to. You most likely have some damaged memory.

 

[jay2tall]

I had tried my buddies memory on the last board with the same issue. And he has been running that memory for 4 months at 1950 with no issues at all. But yet I have the same issue on my board with that memory as the G.Skill I bought.

 

[jay2tall]

I found this blip in the Newegg reviews for my memory. G.Skill left a message for someone:
Dear customer

Becuase the i7 only guarantee DDR3 1066 with their default QPI voltage, please increase QPI voltage to 1.4v and see it can work at DDR3 1600 or not. if you have any issue please contact our support

Thank you
GSKILL SUPPORT

Quality and customer service are our top priorities.

Tech Support Email: ustech@gskill.com
RMA Dept Email: rma@gskill.com
G.SKILL Forum: http://www.gskill.com

 

[cjl]

That's the voltage I'm referring to. 1.4 is a bit high (my board shows it as into the red area past about 1.35), but definitely mess with it a bit and see if it helps.

 

[jay2tall]

Does anyone know what the stock QPI voltage is? Just curious.

I guess the i7's are not as friendly as the Core2's when it comes to memory. What ever happened to plugging in memory and it just taking off?

 

[jay2tall]

I found it 1.1v on the QPI is stock. It says 1.3 is a good starting point for Overclocks and faster RAM. I'll try that out.

 

[cjl]

The QPI voltage was fine with the "stock" settings stored in XMP for me (1.3V IIRC - it just loaded from the XMP profile). You do need to up this voltage some for nonstandard (>1333 9-9-9-24) memory from that 1.1V, but it should usually have it in the XMP profile. Your situation seems unusual.

 

[jay2tall]

I tried setting it to 1.3v and it won't even boot. AT 1.4v it boots but isn't totally stable. It seems better though. I upped the voltage to 1.65 for the DRAM though and that didn't do much.

Me and my buddy were testing some things and if you lower the multiplier to 18 and then set it to 1600 it works fine. He seems to think it's the memory controller to.

 

[jay2tall]

We also tried each DIMM individually and at 1600 they are fine. As soon as you stick 2 in it blue screens.

 

[geofelt]

We also tried each DIMM individually and at 1600 they are fine. As soon as you stick 2 in it blue screens.

Did you install the ram in the Orange slots? It is not the usual closest to the cpu location, but with only three slots it is necessary. The mobo manual does have this info.

Are there any bios updates? Early on, there were several that addressed ram issues.

As long as you are stable,I don't think the speed has much effect. The memory controller seems to be able to keep the cpu filled with data easily.

In my P6T deluxe, I used patriot 1600 without a problem. It runs at 1.5v, freq=480(1440) 4-6-6-18-1T

 

[flyin15sec]

Are those GSkil PI silver?

Have you tried lowering the voltage to 1.5v? Also make sure you are using 8-8-8-21 2T. 1T will fail testing.

 

[jay2tall]

We have tried so many things. We tried the ram voltage on auto and under 1.6 which is what is in the XMP setting. The XMP profile gives it the 8-8-8-21 2T timing, which we have also tried manually. I am using the Orange slots, as the black ones won't even boot. I tried the single DIMM in A1 which is the closest Orange slot to the CPU. Individually they all work. But when you hit dual channel they fail.

I am going to RMA my CPU. My buddy built one for a friend with the same Mobo and CPU and didn't have any issues with it on auto running the XMP profiles. He seems to think it's the memory controller.

 

[cjl]

If the extra memory controller voltage helped, then yes, that seems to be the problem.

 

[dragonsprayer]

So 2 weeks ago I bought upgrade parts for my PC. I purchased an Intel i7 920, ASUS P6T motherboard, and a 3x2GB G.skill DDR3 1600 8-8-8-21 memory kit. I am using a Thermalright Ultra-120 extreme 1366 cooler and running Vista Ultimate 64bit.

I had my mobo set to defualts except for using the XMP memory profile to get the rated memory clock speeds. When I run OCCT it blue screens with a memory error as soon at the test kicks in. This also happens on any game a minute or two into it. If I set my memory to the stock 1066MHz speed, everything is fine. I tried swapping in my buddies Corsiar Dominator 1866 memory (3 sticks) with the XMP profile and the same thing happened. We also tried to manually enter these settings and run at 1.65v. No luck. So I RMA'ed the motherboard, which I recieved the new one today. The same issue happens. Could this be my CPU? Could the memory controller be bad? I am new to the i7 arena so I am looking for help.

Thanks,

NO

WE MADE A HUGE MISTAKE AND RAN A FEW SYSTEMS LIKE THIS, here is the story: when we switch from giga mobos to asus we misread the bios names and crossed the voltage for the memory controller with the dram voltage

key is too keep all under 1.65v and most under 1.45v or even 1.4v (intel spec is 1.38v)

keep in mind we break many of intels rules on every system sold and have done so for over 6 years, We only build oc systems. We have had ZERO FAILURES DUE TO OVERCLOCING IN 6 YEARS! ZERO OR NONE!

we ran the memory controller at 1.65v instead of 1.25-1.4v (i have not posted are i7 settings like i did q6600 in aug of 2007).

system started crashing with time, when the mistake was caught some systems had run the memory controller for over 4 weeks at up to 8 hours a Day at 1.65v. (NOTE WE MODded THE COOLING FOR THE MISTAKE NOT KNOWING WHY, eventually the cpus did not like 1.65v for mem control and air cooled stock heat sink)

we still have zero failures - once the setting where reversed all systems run perfect at 4.1ghz (920 blck plus turbo) and time will tell if we damaged them. - i might make a post on this?????

YOU BE FINE -- I JUST LIKE TO TYPE CRAP WHILE I DO OTHER CRAP!

 

[jay2tall]

^
I'm confused by Dragon's post. The QPI is clearly labeld in the bios as well as the DRAM voltage.

 

[cjl]

I think he's saying you can run QPI/DRAM at 1.65V and not hurt anything. You could try it if you really wanted to, but I think RMA is the better option here.

 

[jay2tall]

My thought has always been, if it can't run what it's supposed to, send it back. I have no issues if something is just not a good overclocker, because you OCing can be hit or miss sometimes. But, if I can't run at stock speeds without pumping alot of voltage to it. Something is wrong.

Newegg sent me a UPS return label.

Here is a hint. Always call newegg for an RMA. They have always given me a return label even though their site says you are responsible for return shipping.

 

[jay2tall]

Ok so today I got my replacement CPU. Popped it in and ran everything like a CHAMP! Actually it warned me to update my BIOS and it wouldn't run my memory at 1600 until I did so. Apparently the replacement i7 I got was the new D0 stepping chip. Oh darn, it OCes to 4.0GHz at 1.325v with everything else on auto. Now that is what I am talking about. I would have been happy with 3.6 or 3.8. I could probably press it more but I'm not sure about temps. It idles around 55-57C and in OCCT with Large Data set, High Priority, and HT as Dual Core, it sits around 86-90C. Is that to hot? I was told it is fine for an OCCT test and will never hit this in reality.

 

[cjl]

86-90 is a bit hotter than I would run, but I run folding on mine (SMP client), so I run at 75-80% load 24/7. Because of that, I like to keep mine at <80C linpack temps, which translates to <70C Prime95 or so (and my current 4GHz OC is pushing that a bit - I actually run at 75C prime or so, but it still isn't that bad, because my folding temps are in the mid 50s). If you don't run yours at that kind of load though, it should be fine.

Oh, and glad to hear the new CPU works fine. Sounds like the memory controller was at fault after all. What is your current memory controller voltage for DDR3-1600 CAS 8 by the way? Mine runs CAS 8 at 1.30V, and CAS 7 at 1.34V.