Can RAM be incompatible but still function?

So I'm having random BSoD (related mainly to ntoskrnl.exe) and random restarts. I've eliminated everything but the motherboard (which up until a few weeks ago worked fine - before getting my new RAM [might be answering my own question there]) and power supply.

Either way, I've ran Memtest86 on the RAM at XMP speeds (2133mhz) both as a pair and separately overnight (8-9 passes) with ZERO errors. So I can assume my RAM isn't the problem.

This goes to my question though. Even though my motherboard (Asrock Z77E-ITX) says it supports 2133 and Windows for the most part works fine (minus the BSoD and restarts), is it possible that although the RAM tests fine that my computer can still function despite the RAM not being fully compatible?

Basically, can the RAM not be fully compatible but still work but due to it's non-100% compatibility still cause issues?

Why not check to see what the BSoD states is the issue? Or even look in the Event viewer at the problems listed there first?

I have. 99% of the time the BSoD is caused by ntoskrnl.exe failing. It is NOT Windows or my hard drives as I've reinstalled Windows multiple times on different drives over the past couple weeks and the problem persists.

It is something hardware related.

Seeing as how the motherboard functions fine with no issues (nor is there any physical damage such as bulging caps or burn marks, etc) and considering that the power supply, if it were messed up, I'm sure would be causing more than what is happening the only thing I can see causing the problem would be the RAM.

So that is why I asked that even if the RAM tests fine with no errors, is it possible that there could still be some compatibility issue that's causing these problems?

JoeUser said:

Either way, I've ran Memtest86 on the RAM at XMP speeds (2133mhz) both as a pair and separately overnight (8-9 passes) with ZERO errors. So I can assume my RAM isn't the problem.

Basically, can the RAM not be fully compatible but still work but due to it's non-100% compatibility still cause issues?

Click to expand...

Certainly it can. That "2133 MHz" RAM (of unknown brand) is in effect being overclocked when you run it at that XMP profile (at an unknown voltage). ASRocks aren't the highest quality boards; overclocked RAM and/or stuff running at non-standard voltages can cause issues. Just the other day someone here had an issue running low voltage RAM on an ASRock board.

Change the memory timings to one of the JEDEC defaults--preferably 1600MHz if it offers that. If not, manually change the speed to 1600 MHz. Run it for a while with the OS loaded at that speed and see if anything changes.

And then try a different RAM testing program with both sticks installed at the (overclocked) XMP profiles. Such as Vista or 7's included built-in pre-boot Memory Diagnostic Tool and Memtest86+.

BTW, were you running the latest Memtest86 (different than "+") that was released just last week or the prior two year old version?

After that, the power supply could definitely be an issue; its also unknown as per your profile.

Alright. The RAM is Team Group. The power supply is the stock 600 watt (80 Plus Bronze) that comes with the SilverStone SG08 case.

The RAM is rated at 2133 @ 1.5v. I was running it at 2133 @ 1.55v.

Currently have the RAM running at 1333mhz (stock if you will) with the voltage and timings set to "Auto".

Going to try it at stock for a couple days and see how it does. IF it runs fine I will try it at again at 2133 with possibly a higher voltage (say 1.6v or so) as the BSoD and restarts happen pretty infrequently (about once maybe twice a day).

I was using Memtest86+ to test the RAM.

Also, if it means anything, I've noticed lately that when the PC does mess up it tends to do so twice in a row (within 10 minutes or so) then it'll run fine for a while. This isn't always the case however.

JoeUser said:

Alright. The RAM is Team Group. The power supply is the stock 600 watt (80 Plus Bronze) that comes with the SilverStone SG08 case.

Click to expand...

Team Group what model and size? You can verify the model # through the free CPU-Z if you don't want to open the case.

Before that though...yikes. Normally RAM with high XMP overclocks of 2133MHz (or more) has at least one default JEDEC timing of 1600MHz. If the top JEDEC timing of yours is in fact 1333MHz (can see this in CPU-Z too), that memory is being pushed even more over-spec to run at 2133. Which can increase the chance of problems and/or compatibility issues.

Hopefully the default timings resolve the issues.

As to the power supply, Silverstone sells both good and not so good ones. Need the model number off the side of it to have any idea if the one you have is any good.

Yah, frequency patterns of BSODs don't mean a lot. Unless they are related to a specific thing you do (only happen running a high power game, etc.).

RAM:
http://www.newegg.com/Product/Product.aspx?Item=N82E16820313234&IsVirtualParent=1

CPU-Z info:


JEDEC seems...off...? 1066 (normal), 1218 (wtf), and 1370 (again, wtf), then 2133. Those two in the middle I've never seen before with any RAM.

As far as the model number on the PSU I have no idea. It is the one that is specially made and sold only with their SG07 and SG08 case.

Also the BSoD happen randomly...can be while gaming, surfing the web, watching a movie, or doing absolutely nothing at all. Truly random, no pattern. Well, I shouldn't say that, it's happened more than once when opening a new browser tab or clicking on a link.

You guys are making me think on this. Doesn't seem like a power supply issue, seems more RAM related than anything.

As I mentioned though, for the time being I'm using stock 1333mhz (it's what the BIOS chose!) to see if I BSoD. Have my cousin mailing me his 2x2GB DDR3-2000 Corsair kit Monday to test should something happen with my PC in the mean time while testing it at stock speeds.

JoeUser said:

JEDEC seems...off...? 1066 (normal), 1218 (wtf), and 1370 (again, wtf), then 2133.

Click to expand...

Yes, a definite wtf. But naw, not anything atypical really. Very often one or more of the JEDEC timings returned by CPU-Z for various memory are a bit "off" of what you would expect to align with common memory timings. You can consider that 1370 to equal 1333 and don't worry about the 1218 .

But yah, that memory now for sure looks like the generic junk it most likely is. For it to even call itself "PC3 17000" when its actually PC3 10700 is deceptive. Also, "Team" isn't exactly the most well known vendor among either enthusiasts or normal people, at least in North America.

Its being pushed way beyond spec to run at 2133. And to imagine it will run fine at that speed at 1.5 volts is more likely wishful thinking than reality.

For example, as a comparison, I have here (in a parts box) five year old Corsair "XMS3 1600MHz" DDR3 RAM. The chips on it are rated...exactly the same as yours: PC3-10700 (1333/667MHz). And your highest JEDEC profile timing (at 1333/685MHZ) is rated exactly the same as that Corsair's highest JEDEC profile timings (at 1333/666MHz): 9, 9, 9, 24, 34. All the JEDEC profile voltages for the Corsair (at 1333/666 and below) are 1.5V, same as yours.

But the Corsair's XMP profile is 1600/800MHz, 9, 9, 9, 24, 41, 2T...and at 1.65V. I can tell you from experience that it ran stable at that XMP profile (in an Asus P6T X58 board) for a few years. Which is a lot more honest than having a XMP profile of 2133MHz with the same speed of RAM as yours has.

So it really looks like you have RAM that's trying to pass itself off as something it's not. Second possibility--that's less likely, but still a possibility as its an ASRock board--is that the memory simply won't function properly at that XMP speed & voltage on that particular board. But at this point, it looks more likely that its the memory alone that's the problem.

Advice if you don't want to exchange that memory for a more well known brand that's on ASRock's compatibility list for that motherboard (which yours is not): if really you want to run at 2133MHz, slowly increase the voltage up to no further than 1.65V and see if you can run Windows without BSODs.

Well I don't want to speak TOO soon...but so far, after all day today and all night last night I haven't had a single issue whatsoever running the RAM at 1333mhz. Going to test it for a couple more days then go back to 2133mhz and give it some more voltage like 1.6v-1.65v or so and see how that goes.

I only say this as well because the last RAM I had (Gskill 1866) would run at 1066mhz stock and 1866mhz XMP but I had to up the voltage to 1.65v to get it rock solid stable.

I have to admit that I found it odd that this RAM is rated 2133mhz at 1.5v...that doesn't even seem possible all things considered. Who knows though.

It might not be the ram.

It might be the IMC in your CPU, Sandy Bridge default ram speed is 1333, anything higher is overclocking. Plus you said you were pushing 1.65v through it, that is also overvolting the IMC inside the CPU (and also the ram of course)

I would set it at fast as it can go at 1.55v stable and call it a day, the difference between 1800mhz and even 2400mhz is negligible anyways.

JoeUser said:

Well I don't want to speak TOO soon...but so far, after all day today and all night last night I haven't had a single issue whatsoever running the RAM at 1333mhz. Going to test it for a couple more days then go back to 2133mhz and give it some more voltage like 1.6v-1.65v or so and see how that goes.

I only say this as well because the last RAM I had (Gskill 1866) would run at 1066mhz stock and 1866mhz XMP but I had to up the voltage to 1.65v to get it rock solid stable.

I have to admit that I found it odd that this RAM is rated 2133mhz at 1.5v...that doesn't even seem possible all things considered. Who knows though.

Click to expand...

Yah, it all depends on the quality & "overhead" of the RAM chips used by the assembler, the flexibility of your motherboard (in terms of electrical stability, not how "flexy" it is ), the circuit board used by the RAM assembler (in your case, Team) and so on. Some memory on some boards can meet its advertised (XMP) overclocking with lower speed chips a lot more reliably than other RAM can. As evidenced by your Gskill.

And no, the IMC ("integrated memory controller") on the CPU has little to do with it. Volting memory beyond 1.65V, of course, is not recommended. But up to 1.65V is just fine. And there is massive evidence out there to support the fact that high quality memory used on high quality motherboards can overclock reliably to speeds of 2800MHz and higher.

Anyway, good luck with it, glad to hear the default 1333MHz speed works without issues and hope you are able to get more by upping the voltage.

Well I stepped the RAM voltage up to 1.65v (running 2133 XMP) the other day and (knock on wood) I haven't had any problems. Been running the PC 24/7 for 3 days now...a length I haven't been able to accomplish in weeks. I don't want to jynx myself, but if my PC were to blue screen I'd like to say that it would have by now.

1.65v makes sense though. I know with my last RAM that 1.65v was the MINIMUM voltage that the RAM would run at stable with XMP. Should have tried this earlier, but even Team Groups website says that 1.5v is enough for XMP 2133...which in my case is far from the truth.

Either way, all I can do is just continue on and hope that my PC stays stable.

GOD FUCK SHIT! BLUE SCREENED AGAIN!!!

WHAT THE FUCK?!?!?!?!?!

Same ole ntoskrnl.exe failing again...goddamn...this shit...I'm about to throw my fucking PC out in the garbage and just start the hell over again...I swear...

BSOD when I was just trying to load a webpage. Makes no sense people...the randomness of these things...I've had my PC on for like 3 days straight...browsing the web, YouTube, gaming (and on the IGP at that so even MORE RAM load), sitting idle, downloading stuff, etc...ZERO problems. Now I just opened a new tab, did a Google search in said tab, results pop up then BAM BSOD.

SIGH!!!

I'm at my wits end here...I don't know if it's the RAM or what. Like I said, I'm running at 1.65v...so I'm convinced the voltage isn't the problem. 1.65v should be enough for any XMP profile regardless of the speed.

At this point I'm thinking it has to be my motherboard...bad RAM slot or something. Even when I had my RAM at 2133 @ 1.525v I got no errors in Memtest86. That was running the RAM together and separately.

I've tried different OS installs on different drives, RAM test, running everything at stock, with and without video card, etc...only thing I haven't tried switching out is the motherboard (haven't swapped RAM but all things I've done so far don't point to the RAM). Power supply still isn't out of the question, but that seems unlikely in my eyes.

I got no errors in Memtest86.

Click to expand...

How many days did you test with memtest86 before calling the ram good?

At this point I'm thinking it has to be my motherboard...bad RAM slot or something.

Click to expand...

I would just back off the IMC overclock and run the ram at a DDR3 1600 to DDR3 1866.

You do know that 1.65v can cause damage to your processor.

Skillz said:

You do know that 1.65v can cause damage to your processor.

Click to expand...

RAM voltage can cause processor damage?

drescherjm said:

How many days did you test with memtest86 before calling the ram good?

Click to expand...

I tested the RAM as a pair and individually for about 8 hours each. 7-8 passes in Memtest.

I would just back off the IMC overclock and run the ram at a DDR3 1600 to DDR3 1866.

Click to expand...

That's what I'm going to do now. My motherboard is said to run RAM up to like 2800mhz though...

My motherboard is said to run RAM up to like 2800mhz though...

Click to expand...

And your processor has multipliers up to 57. That does not mean you will or even can achieve that type of overclock.

I tested the RAM as a pair and individually for about 8 hours each. 7-8 passes in Memtest.

Click to expand...

That may have not been enough.

JoeUser said:

RAM voltage can cause processor damage?

Click to expand...

Its an anti-overclocking sentiment. Intel's statement: "Intel recommends using memory that adheres to the Jedec memory specification for DDR3 memory which is 1.5 volts, plus or minus 5%. Anything more than this voltage can damage the processor or significantly reduce the processor life span." "Can" in no way means necessarily that it "will".

Reliable memory vendors sell OCed memory at high claimed speeds and XMP voltages above 1.5v. Without reports of thousands of CPUs spontaneously 'sploding. That doesn't mean your setup will be able to reach those speeds reliably, though. General consensus over the years since the Core "i" series CPUs were first released is that 1.65v is the "safe" maximum for memory voltage with Intel chips.

Sorry to hear your memory + that motherboard is not stable at claimed speeds (so far). Hope you are able to find something stable above stock speed. As noted by others, run Memtest longer than you have so far with the mem at 1333MHz/1.5v before considering your sticks "good".

Many people have ran 1.65v or higher on their Intel setups for 24/7 and have had no problems. however, i personally dont want to risk killing my 3930K.

OP, maybe there are setting you can change in the RAM timing to get them stable at your desired speeds. I remember when I had an old DFI Ultra-D, there was a lot of settings and one of them, RAM Drive Strength, had to be changed to get all 4 slots working at DDR3200.

Most of the time it seems that if you have multiple sticks, changing to command rate 2T will help a lot and there is very little performance plenty.

JoeUser said:

RAM voltage can cause processor damage?

Click to expand...

http://hardforum.com/search.php?searchid=23172549

I had this problem with one of the workstations I put together here at work. It would just randomly restart or BSOD. Ran memtest for 8 hours and it passed. Intel Burn Test would fail though, even if I raise the voltages. So I really thought it was the motherboard. While looking for a link to the RMA on the motherboard manufacturer's website, I saw another link to a list of compatible memory. I checked the modules and they're not on the list.

So I pulled out the modules on another computer (same brand, different model) and everything's stable. Installed the other pair on the other computer (basically just switched modules) and no errors either.

Check if the modules are on the list of compatible/tested memory for the motherboard.

Try using prime95 blend test and check "round off error checking". It might give you errors much faster then memtest86.

I had memory that passed every synthetic test i could throw at it, except prime95 blend test (tried intel burn test, hci memtest, memtest86+, occt, etc)

Skillz said:

http://hardforum.com/search.php?searchid=23172549

Click to expand...

Link is broken.

xref said:

I had this problem with one of the workstations I put together here at work. It would just randomly restart or BSOD. Ran memtest for 8 hours and it passed. Intel Burn Test would fail though, even if I raise the voltages. So I really thought it was the motherboard. While looking for a link to the RMA on the motherboard manufacturer's website, I saw another link to a list of compatible memory. I checked the modules and they're not on the list.

So I pulled out the modules on another computer (same brand, different model) and everything's stable. Installed the other pair on the other computer (basically just switched modules) and no errors either.

Check if the modules are on the list of compatible/tested memory for the motherboard.

Click to expand...

My RAM isn't on their list but neither is 99.9% of DDR3 RAM that's out there. I really can't put a lot of weight on this one. The RAM I had before that worked 100% isn't on the list either so...

serpretetsky said:

Try using prime95 blend test and check "round off error checking". It might give you errors much faster then memtest86.

I had memory that passed every synthetic test i could throw at it, except prime95 blend test (tried intel burn test, hci memtest, memtest86+, occt, etc)

Click to expand...

Will do!

Just do a search on here for user: IntelEnthusiast keywords: 1.65v and sort them by posts.

You'll see what he has to say about 1.65v.

Hmmm...what IntelEnthusiast is saying seems to be in conflict with what Intel's data PDF's claim.

http://www.intel.com/content/www/us.../core-i7-memory-suppliers-1111-datasheet.html

There it is. Officially from Intel regarding compatible RAM for first and second generation i7's (and consequently second generation i5's I'd imagine). There are a BUNCH of 1.65v RAM modules on that list.

Not to mention the fact that this is the first time I've EVER heard of this. As well as the fact that there are a ASS LOAD of people running 2nd gen i5/i7's with RAM at 1.65v without any issues after years of running them.

I know IntelEnthusiast is from Intel, but I don't know...just doesn't seem right...

Not only that but my last kit ran at 1.65v all day long without an issue.

Edit:

Better yet, here is a list from Intel specifically for 1st and 2nd gen i5 processors:

http://www.intel.com/content/www/us...asheet.html?wapkw=intel+core+i5+system+memory

Plenty of 1.65...even a few 2133mhz @ 1.65v. SHIT! THERE'S EVEN 1333MHZ RAM AT 1.65V! WTF?!

So 1.65v is potentially dangerous yet Intel themselves have no problem recommending and certifying 1.65v RAM? What game is Intel playing here?!

I swear, if my CPU is messed up and I can't get it RMA'ed I'm going to be one pissed off sumbitch!

Stick with 1333 or 1600 at 1.5V, if both/either are stable. You will not be able to discern any difference between 1333, 1600, or even higher RAM speeds with an LGA1155 platform outside of synthetic benchmarks.

I know it sucks to have to underclock the 2133 modules you paid extra for, but if you have to choose between un unstable system with higher speed memory and a stable system with lower clocked memory, then choose the latter.

JoeUser said:

Plenty of 1.65...even a few 2133mhz @ 1.65v. SHIT! THERE'S EVEN 1333MHZ RAM AT 1.65V! WTF?!:

Click to expand...

FWIW, all memory modules are designed to JEDEC standards. For DDR3, that's 1.5V. When MFG's package these modules on dimms for retail sales, it's up to them how they want to market their product. For some customers the advertised 1.65V support is a good thing. (I have come to see at it as snake oil) At the end of the day, all the memory components in your system were designed for 1.5V, no matter what the dimm manufacturer specified.

IME, for 24/7 reliable overclocking, memory modules are not something worth pushing as hard as a CPU or GPU. Benching and other temporary situations, by all means, let the flood gates open.

So many times I've read of people who say their overclock has been stable long-term, when it's only been 6-months since they've built their system. I call a good 24/7 overclock one that lasts 5+ years without issues.

OK...I will jinx myself again I'm sure...but I've been running 1600mhz at ~1.5v (1.525 I think) and so far I haven't had any issues.

We'll see what happens (fingers crossed!).

You know, all those forms say "supplier self certified". I take that to mean that the supplier tested the ram on that platform, and it functioned... not that Intel is saying that it's "officially" supported.

Rock&Roll said:

So many times I've read of people who say their overclock has been stable long-term, when it's only been 6-months since they've built their system. I call a good 24/7 overclock one that lasts 5+ years without issues.

Click to expand...

Many people don't use their overclocked parts for more than a couple years, myself included.

OK so with the RAM clocked down to 1600mhz I haven't had a single BSOD...it's been well over a week. Hmmm...so I guess running at 2133 just has compatibility issues I'm guess? Not even 1866 is stable though my last RAM ran at that just fine. Hmmm...either way, I guess 1600 is fast enough! What ever, I'll make do...

Might be worth the effort to stick everything back to stock, and flash your bios to latest version. Then try overclocking again. Hopefully they updated ram compatibility.


Posted from Hardforum.com App for Android

myself had several problems do make my SB system Prime stable at 2133Mhz... I manage to set the maximum ram oc at 1866Mhz, which is more than enough.

remember that every IMC is different and differs from chip to chip, 2133Mhz is the maximum allowed on SB, and no one garantees that it's 100% stable.