Jump to content
Corsair Community

ASRock X79 Extreme 6/GB freezes at random intervals with 64GB RAM installed


Toni

Recommended Posts

Hello,

 

I’m having frustrating issues with my ASRock MoBo and RAM config since I started to assemble my new rig weeks ago.

I began to experience freezes, lockups and BSODs frequently even inside the UEFI BIOS.

 

I ordered two sets of 32GB LP RAMs because I bought a Noctua NH-D14 SE2011 heatsink which requires low profile heat spreaders in order for the heatsink to be seated properly onto the CPU base.

 

I’m planning to OC my machine once I get it stable, though. I hope the Noctua cooler will handle CPU temps within acceptable range.

 

Now, to troubleshoot the issue I changed my RAM timings from BIOS default of 1333 MHz to XMP 1.3 Profile 1, which sets the timings, voltages and frequency to 1600 Mhz per manufacturer’s spec. Then I ran Memtest 86+ ver. 4.2 (latest version) to see if my RAM is faulty.

 

With 64 GB of RAM it took me over 10 hours per pass to complete. I was told that I should let it run for at least 4-6 passes with no errors before you can determine the RAM is good. In my case I ran the test for a full 13 passes and I got the first error on the 2nd pass. See the pic for reference. (IMG_01)

 

In my 2nd try I reseated ram modules and shuffled them around in random order to see if those few errors were due to bad contact between the slot pins and the ram gold-plated connectors. I ran the test again and I still got 1 error in the 2nd pass. See pic for reference. (IMG_02).

 

What’s interesting is that both tests failed at the same address and at 7937 MB. Since I’m not too technical about computers, I don’t really know exactly what it means or where the problem lies based on these data. It could be a particular RAM that is defective or one of the slots maybe the problem.

 

This time I decided to run the test on each module separately. However I used only slot DDR3_A1, DDR3_B1, DDR3_C1, and DDR3_D1 to test all 8 sticks one at a time looping the slots twice. I could not use any of the secondary slots to test since the computer won’t even boot from any of them.

 

I ran an average of 14 passes for each module with no errors at all on any of them.

So I have to rule out the possibility that any of the sticks and 4 primary slots are defective. Again all the tests so far have been completed with default bios setup except the XMP 1.3 Profile 1 is activated. So, what I’m not sure about is if any of these secondary slots may be the problem???

 

Well, since I’m not able to test them individually I ran the test with only 4 primary slots populated to see if the error maybe due to incompatibility of the ram sticks working together.

 

To my surprise the memtest reported one more error in pass 5. See pic for reference. (IMG_03).

 

I’m really confused now. Knowing that each ram and each of these slots gave no error when tested one by one, I have to suspect that the problem may be due to the RAMs throwing errors when paired together. That’s why I came to this forum to ask for some advice what to do next to make this thing work.

 

In the end if nothing works, I really need to identify which of these hardware MoBo or RAM kits need to be RMA-ed. It will cost me more frustration and lots of wasted time and money to RMA the wrong parts. Help me troubleshot this issue, please!!

 

Thank you

IMG_01.thumb.JPG.1d7b7560f0f503a1e1465f110c83f585.JPG

IMG_02.thumb.JPG.95e3569d55e9468b5bba8b2bd9e7b769.JPG

IMG_03.thumb.jpg.8da96c1f233a6988c2e784a0860076d0.jpg

Link to comment
Share on other sites

Can you match up the sticks back to their respective kits? If an RMA is required, that may be mandatory anyway. Are you using this pc as a server? Trying to understand the need for 64GB RAM. 32GB is even overkill for most normal pc uses.

 

Since each stick has been tested individually in the primary DIMM slot without faults, the answer can only be that the two kits mixed together will not work. Sometimes, you can get that scenario to work by increasing the DRAM V. or relaxing the timings, increasing the SA V. for example.

 

This 64GB kit may be an option. They look to be of low profile configuration; CMX64GX3M8A1333C9

Link to comment
Share on other sites

I ran an average of 14 passes for each module with no errors at all on any of them.

 

Unfortunately you are using two kits 32 gig kits at the same time which is not suggested or supported by Corsair because of the various issues that can come from it. And it sounds as if this is the case here. Having all sticks pass individually would suggest the RAM is fine.

 

Take Hu16E advice and isolate the two kits and instal them one set at a time and see if stability is still an issue.

 

If it is not then you have your answer and might think about using a matched 64gig kit .Or you may have some other underlying issue with the MB itself.

Link to comment
Share on other sites

OK, addressing Hu16E’s questions.

 

In my daily work I do plenty of Archviz projects which sometimes involve animated presentations. I use mainly 3ds Max and for post-production I use eyeon Fusion. Both software require abundance in RAM size to stimulate physics and to be able to load hundreds of megabytes of texture map sizes and proxy objects at render time. Fusion the same, when I use its internal particle system to render mid range to high complexity particle effects it becomes a RAM hog if you don't have enough of it. 64 GB RAM was an attractive option a few months back and some X79 platforms such as mine, offered this possibility.

 

From my experience with 3ds Max (and other pro software) I predict in a couple of years 64 GB RAM would be the minimum required to run the software, :D:. So, yes I wanted to future proof my system, as well.

 

Are you using this pc as a server? Trying to understand the need for 64GB RAM. 32GB is even overkill for most normal pc uses.

Well, I plan to use it as Workstation with a “Renderfarm” config networked with 3 smaller machines in my office. This one will be a primary working environment for modeling, texturing, animation and particle simulations and as a Manager/Server through a Backburner system provided by 3ds max.

 

I know a server machine might have been a better choice for me but with at least double the cost, which I could not afford atm.

 

OK, back to RAM issue.

 

Since each stick has been tested individually in the primary DIMM slot without faults, the answer can only be that the two kits mixed together will not work.

 

Yeah I fear that. My mistake is that when I removed the RAMs out of the box on both kits to assemble my unit I didn’t take note of their individual serial nr, thus I can’t determine now which module belongs to the 1st kit and which one to the 2nd. Even Corsair box doesn’t list their nr on it. If it does I like to know where.

 

It’s really frustrating if you mix modules from 2 or more kits and want to group them back to their respective kit, you will get it right only by chance. Now I have to play with 16 combinations in total in my case and test each time (4x4), which I’m not going to do for now, too much work. If you can suggest me a direct way to pair those sticks back to their original kit please let me know. I’ll test them again each kit at a time, and see if ram mismatch is the culprit. If not, than I have to look at the MoBo.

 

At peanutz94.

 

I looked around online a few months back as I remember to find 64GB kits Corsair RAMs at low to mid range price but was not very successful at the time. The most I could find were kits of 2 and 4 modules with 1600 MHz frequencies. ******** might have had that kit as an option, not sure, but I passed that brand over. I think Corsair is a better option. :dunno:

 

Unfortunately you are using two kits 32 gig kits at the same time which is not suggested or supported by Corsair because of the various issues that can come from it.

 

Well, if Corsair doesn’t support the use of 2 kits in one system, I’ll do my best to use only one kit once I sort out what members belong to each of my individual kits and test the system with on kit installed at a time.

 

If it is not then you have your answer and might think about using a matched 64gig kit.

 

I don’t know if Corsair has that high density module available at these frequencies and timings in a set of 8, even today. If it does, I’ll ask them to exchange it for me if they can do it.

 

Thank you again for your help and support.

Link to comment
Share on other sites

http://i1249.photobucket.com/albums/hh504/ClipperSnipper/RAMStick.jpg

 

Sorry about the low quality of the image, but if your sticks have a serial number, it will be located on the stick's label. Also, it may have a batch number on the label as well.

 

I don’t know if Corsair has that high density module available at these frequencies and timings in a set of 8, even today.

 

Yes, they do. They even have a 64GB 'rated' at 2133!

 

CMZ64GX3M8A1600C9 CMD64GX3M8A2133C9

 

Keep in mind, running RAM at some of those 'rated' speeds is 'very' difficult due to limitations in the CPU IMC's.

Link to comment
Share on other sites

http://i1249.photobucket.com/albums/hh504/ClipperSnipper/RAMStick.jpg

 

Sorry about the low quality of the image, but if your sticks have a serial number, it will be located on the stick's label. Also, it may have a batch number on the label as well.

You can NOT go by serial numbers on the modules because of the way the modules are tested and packaged they are very rearely the same or not sequential . But if they happen to be different version numbers that would be one way.

Link to comment
Share on other sites

You can NOT go by serial numbers on the modules because of the way the modules are tested and packaged they are very rearely the same or not sequential .

 

I have not found that to be the case with the Platinums or many other kits I have, but a good point as some will not always be sequential. In fact, in many kits, there is no serial number on the labels at all. I was only trying to assist the OP with suggestions in the quest to attempt to match up the right sticks with the right kits. :):

 

Added: Another suggestion/recommendation. If you want to get away from that monster CPU cooler problem with DIMM slots getting blocked, there are others on the market that won't do that & are 'very' efficient at cooling. Since Corsair makes an assortment of coolers, it would be inappropriate for me to list my favorite air CPU cooler that does not block DIMM slots.

Link to comment
Share on other sites

  • Corsair Employee

Toni

Go by the lot code that will be on every set of modules we make and will start with the Julian year they were made; in this case most likely 12XXXXXXX the matching sets will all be the same lot. In the event they are both the same you can try to match them up by the SN#'s but as mentioned they are not always sequential.

Link to comment
Share on other sites

At HU16E

 

Yes, they do. They even have a 64GB 'rated' at 2133!

 

They have 64 GB kit Low Profile at 1600 Mhz???? Sorry, I don’t think so. If they have it I can’t see it advertised in any 3rd party online vendors, though, including newegg and others. However, CMZ64GX3M8A1600C9 is a good alternative with tighter timings than the one I have but aren’t low profile.

 

CMD64GX3M8A2133C9

Keep in mind, running RAM at some of those 'rated' speeds is 'very' difficult due to limitations in the CPU IMC's.

 

Pointless to use that kind of RAM, in opinion….For, at most, 2-5 % increase in overall performance, you pay almost triple the price of a 1600MHz model. And as you said it, it’s really hard for the CPU to maintain the RAM frequency at its factory rated speed. And if you OC the CPU with 64GB of 2133 MHz frequency, it will suffer and degrade much faster with that type of RAM, since you have to push the CPU VTT, VCCSA and even DRAM Voltages to the “danger zone” to maintain 24/7 stability, if you are lucky with the best X79 board on the market.

 

Also that model (CMD64GX3M8A2133C9) was not even an option for me back in August when I bought my “CML32GX3M4A1600C10”. It’s a new generation of Dominator series, “Platinum”….OK:headbang:!!!

 

Added: Another suggestion/recommendation. If you want to get away from that monster CPU cooler problem with DIMM slots getting blocked,

 

Well, no I don’t want to get away form that massive cooler because it does the job very well. I’m glad I’m cooling my CPU with a “high-end” line of air coolers Noctua (according to many tests and recommendations of customers and professionals who have used it in the past), which I’m satisfied so far with temps barely at low 60s (highest core temp) under full load with IBT and Prime95.

 

there are others on the market that won't do that & are 'very' efficient at cooling.

 

The ONLY ‘very’ efficient at cooling that DOES NOT block DIMM slots and that competes with consumer air cooler such as Noctua is H100 closed loop water cooling solution from Corsair costing almost 50% more than price of Noctua. Many people have complained about the liquid getting evaporated over time and worst of all some have leaked and have damaged PC components inside the chassis. Sorry, I don’t mean to make bad advertising of this particular Corsair product. You will face the same issue on all liquid cooling solutions, regardless of brand. This is my little argument why the choice I made is “better” then what you are suggesting.

 

And also, H100 has 2 year limited warranty, while Noctua NH-D14 SE2011 has 6 years. However, the only problem I have with this heatsink is that it is ONLY compatible with Vengeance LP and partially compatible with XMS3 Dominator and XMS3 Dominator GT series. So CMD64GX3M8A2133C9 and CMZ64GX3M8A1600C9 is NOT an option for me at all with this cooler. So, not many varieties of RAM’s to choose from with this.

 

OK, now back to my real problem.

 

I have looked at the module’s bar code which has numbers and letters top and bottom. The only numbers that differ from each of the sticks are the 6 digit nr at the bottom of the bar code. Everything else is consistent on each module. To illustrate what I’m telling you here, I have included a pic of 2 of my sticks that have all the info. See IMG_03 below for reference.

 

Now please tell me how to put all 8 of them back to their respective kits, judging on what you see in the image. I’ll give you all (8 total) of the 6 digit nr marked on each stick, if needed, and you help me put them in 2 groups of 4 with some luck:biggrin:. And by the way, those numbers are not sequential at all, as peanutz94 pointed out.

 

And, sorry but I don’t really get the RAM GUY’s suggestion to go by the lot code. What lot code? :thinking:

 

Go by the lot code that will be on every set of modules we make and will start with the Julian year they were made

 

Pffffff. Honestly, it’s quite ridiculous that people can get confused so easily with smth as stupid as unintentionally mixing RAM’s from different kits and spending days to figure out how to arrange them back together at factory settings.:(:

 

Maybe Corsair, for the series I own at least, may want try to color code, mark and/or number all the sticks on the outside from the same kit in a logical way so people can regroup them easily when they are mistakenly mixed with sticks from another same kind of kit.

Also does it avoid the warranty if I return a kit for replacement that contains sticks that are from 2 or more other kits? How would they know I have mixed them?

 

Anyways, hope to resolve this ASAP, so I can move on to the next step of testing them.

IMG_04.thumb.jpg.82e54e7c300868ec17333dadb31589c9.jpg

Link to comment
Share on other sites

Technobeard

 

Your info about the warranty is the official one, indeed. I referred the warranty to newegg’s site and it reads 2 years, but obviously it is not accurate :sigh!:. Check for yourself, in ‘Details’ tab:

http://www.newegg.com/Product/Product.aspx?Item=N82E16835181017

 

But still, to me makes no difference since I already made my decision on CPU cooling, but good to know what it is exactly.

 

Now, I have these numbers to sort out in two groups from my RAM sticks. Someone may help?

 

843221

843223

843230

843240

843242

843243

843244

843245

 

Thank you

Link to comment
Share on other sites

I just love it when people get an attitude when all we try to do is assist & make suggestions & recommendations. I have a brand X CPU cooler that will do everything that monstrosity you have will do while not blocking DIMM slots.

 

Check the spec height of the CMZ's I listed. I think you will find that they may just fit under that cooler you have. Since you have named the CPU cooler you have, I will go ahead & name mine, Zalman CNPS9900Max-B! IMO, it beats the heck out of what you have. In 'extreme' OC sessions on air cooling, I've OC'd 3 Sandy 2700k CPUs to 57X using it! :):

 

You now want help to match up your sticks? The last 4 listed seem to be a fairly safe bet that 843242-843245 are one of the kits. Those 4 sticks look pretty sequential to me.

Link to comment
Share on other sites

Since you have named the CPU cooler you have, I will go ahead & name mine, Zalman CNPS9900Max-B! IMO, it beats the heck out of what you have.

 

Zalman is a good looking cooler with an exotic design, but I don’t think it’s superior over Noctua in cooling overclocked CPUs. Tom’s hardware doesn’t even have that model on the testing benchmark at all, I don’t know why but…look for yourself.

 

http://www.tomshardware.com/reviews/LGA-2011-i7-3960X-Air-Overclocking,3130.html

 

In 'extreme' OC sessions on air cooling, I've OC'd 3 Sandy 2700k CPUs to 57X using it!

 

Hmmmmm impressive!!! Do you think it’s the cooler to be prized here or maybe the combo good chip +good mobo and then the decent cooler in the backstage doing the trick?

 

I have heard some people OC their latest Intel unlocked CPUs 24/7 to about 50X or so with just CM Hyper 212 EVO and with really impressive temps….but really I’ have to see how well this cooler will handle my OC once I get the RAMs stabilize with my mobo. I predict, I can’t expect much OC since that humongous amount of RAM will rise the overall temps quite a bit and I plan 24/7 OC not a benchmarking OC.

 

Check the spec height of the CMZ's I listed. I think you will find that they may just fit under that cooler you have.

 

Nope, unfortunately. It’s in the manufacturer’s website RAM compatibility list, crossed off.

 

http://www.noctua.at/main.php?show=compatibility_ram_gen&products_id=34&lng=en

http://www.noctua.at/main.php?show=faqs_single&faqs_id=44

 

Indeed, If I remove the 120 mm fan of this cooler (or raise it upwards a little) the only DIMM slots blocked on my X79 board are the two slots closest to the CPU. If I have to mount ONLY 4 RAM sticks in the primary slots than this cooler will not block any RAM at all. I can choose freely any type of RAM I want, even CMZ's. Since I want to populate all 8 slots then the RAM choice becomes important.So for now I have to run with 32 GB matched pair of LP modules (if they pass memtest) and wait till Corsair produces 64 GB of LP RAM’s for a possible exchange.

 

OK, struggling to sort my sticks in their own kits, I’m still pretty much clueless after all the fuss.

 

RAM GUY has mentioned to look at the lot code but where to find the lot code I have no clue. Factory boxes that housed the 2 sets of sticks are identical in the content and labeling. Even the bar codes printed on them are identical. Strange!!!

 

RAM GUY, please reveal your secret. Where is the “Code”? lol

 

Technobeard-Corsair Employee: Corrected me of the cooler warranty but told me nothing about matching my RAM sticks, at all. He is the Corsair employee, he should know, I guess…no? But still I appreciate your input. Last of all, I’m to blame for it all.

 

At HU16E:

 

The ones that are shooting errors in my last test (IMG_03) are exactly those seated in primary DIMM slots with SN from 843242 to 843245. I strongly agree with peanutz94, they are not sequential. Errors we see in the last test prove it, unless my MoBo is doing the deed.

 

:sigh!: I may start thinking to test them in different combination, 16 times for at least 4 passes each but ~6 hours a pass x 4 = 24 hours per combo. In the worst case scenario it will take me more than 2 weeks to determine which is which, unless by a great chance I get no errors in the 1st or 2nd combo and call that group a kit. Done!!!

Link to comment
Share on other sites

Ok, let's see if I can help out here. Ram Guy already gave the information to look for in the batch/lot number in Reply #8. In my Reply #5, you see two oval circles? The small left one is the serial number, & the larger oval to the right is the batch/lot number.

 

It appears from what you have said, the 843242-843245 are the troublesome sticks. I would bet money on them to be one of the kits. They 'are' sequential in serial number, or shall I say it another way, they are consecutive. The batch/lot number should start with the numbers 12. The kit I'm looking at right now is 122909879 on all the sticks. Yours should be 12xxxxxxx or similar. If all 8 sticks have the same batch/lot number, which looking at 'all' those serial numbers I am guessing they very likely do, then IMHO, 843242-843245 should be considered as one of the two kits for RMA purposes. :):

 

The best way to find a stick that is faulty, is to test each one individually in the primary DIMM slot. The primary DIMM slot is often listed as DIMM 1 or maybe DIMM 0 depending on the manufacturer of the mainboard.

 

As a side note, right, for everyday rigs, I only have 3 that will do 50X or more on less than 1.400v's on the CPU. Temps with the Zalman run between 68C & 73C at sustained load, ambient 20C. Case air flow is completely optimized. I have no idea why Tom's didn't test & review the Max-B, it's their loss. Just to show I'm not blowing smoke, here are two of the validations. Too lazy to look for the third. :):

 

http://valid.canardpc.com/show_oc.php?id=2148989

http://valid.canardpc.com/show_oc.php?id=2127859

Link to comment
Share on other sites

HU16E,

 

Thank you for your thorough explanation of the lot code. It is one of the numbers on the right side of SN that is made of 9 digits starting with nr 12. I got it.:biggrin:

 

Up to now, I was going in loops to find the lot code which according to RAM GUY will help to solve the problem with the mixed sticks.

 

It appears that these are SN#'s and you should go by the Lot code to match the memory up as one set.

 

However, since the lot code is identical (just as you guessed it) on all 8 sticks then in my case the lot code becomes useless in helping me to sort the sticks out.

 

Why? Because just as peanutz94 and RAM GUY mentioned, respectively:

 

You can NOT go by serial numbers on the modules because of the way the modules are tested and packaged they are very rearely the same or not sequential .

 

In the event they are both the same you can try to match them up by the SN#'s but as mentioned they are not always sequential.

 

In contrary, you are assuring me that the sticks with sequential serial nr from 843242-843245 are indeed the members of one kit, which is highly suspicious to me.

 

If I happened to have 2 kits with same lot code but with non-consecutive or with say at most 2-3 consecutive nr out of 8 modules, then what logic should I use to determine which of those belong together?

 

My case, again, proves that even with 4 consecutive nr there is no guarantee that these sticks are part of one set based on your assumption, because when I memtest all 4 of them they were displaying errors. And these errors I’m assuming for now happen because of mismatched sticks. Refer to IMG_03 to see the error code of these sticks.

 

This error occurs when I test 4 of them together ONLY, in primary DIMM slots; DDR3_A1, DDR3_B1, DDR3_C1, and DDR3_D1. <---Labeled as such from my mobo manufacturer.

 

I have already tested all 8 individually with no errors as such:

 

843221 ----> tested in DIMM slot DDR3_A1 ----> 13 passes ----> no errors

843223 ----> tested in DIMM slot DDR3_B1 ----> 13 passes ----> no errors

843230 ----> tested in DIMM slot DDR3_C1 ----> 15 passes ----> no errors

843240 ----> tested in DIMM slot DDR3_D1 ----> 12 passes ----> no errors

843242 ----> tested in DIMM slot DDR3_A1 ----> 19 passes ----> no errors

843243 ----> tested in DIMM slot DDR3_B1 ----> 10 passes ----> no errors

843244 ----> tested in DIMM slot DDR3_C1 ----> 16 passes ----> no errors

843245 ----> tested in DIMM slot DDR3_D1 ----> 14 passes ----> no errors

 

I was not able to test in the secondary slots DDR3_A2, DDR3_B2, DDR3_C2, and DDR3_D2, because the machine could not boot at all if RAM was seated in any of these slots.

 

So at this point, for certainty, I would like someone (a lab technician possibly) to tell me which of these sticks have been tested in the lab with which and assembled as one kit. It’s important.

 

I would wish to try and test the same sticks in another X79 board and see if they shoot any errors on that board, but, unfortunately, I do not have one on my side available atm. Maybe a local PC shop will make me this favor….I dunno, I have to find out.

 

Update: I ran memtest again after I reseated all the same 4 sticks 843242-843245 in the primary slots and now the first error appeared right on the first pass. :(: I’ll leave it run overnight and see if I get more errors, but maybe pointless.

 

As a side note: It appears the error has the same pattern as the error on the first test (IMG_03 in my first post) and occurs at the same density, 13532.0 MB. Can someone interpret which of the modules and in what slot is giving me the error?

IMG_05.thumb.JPG.f73c8c6fa088da9e455bd1dae5124190.JPG

Link to comment
Share on other sites

This is my personal opinion about this. peanutz94 statement is a bit misleading about not being able to go by serial numbers. If they are 'sequential', 'consecutive', I would almost bet my life on them being a matched & tested together kit. I hope that helps make your decision on how to RMA which 4 sticks. I guess I just need you to trust me on this. I pulled out over a dozen Corsair RAM kits, that's all I own btw, & every kit other than a few low voltage low profile kits I have are sequential/consecutive serial numbered. Some of my Corsair sticks don't have serial numbers at all. In that case, the only thing to go by would be a batch/lot number or maybe revision if they ever got mixed up. Since mixing RAM is 'not' recommended, confusing kits wouldn't be an issue. No insult intended. :):

 

Just a few Corsair kits in the hobby/test room arsenal;

 

http://i1249.photobucket.com/albums/hh504/ClipperSnipper/RAMCollection.jpg

Link to comment
Share on other sites

  • Corsair Employee
As a side note: It appears the error has the same pattern as the error on the first test (IMG_03 in my first post) and occurs at the same density, 13532.0 MB. Can someone interpret which of the modules and in what slot is giving me the error?

A: No that is not possible because of bank interleave it is impossible to tell which physical module is failing.

 

Update: I ran memtest again after I reseated all the same 4 sticks 843242-843245 in the primary slots and now the first error appeared right on the first pass. I’ll leave it run overnight and see if I get more errors, but maybe pointless.

A: That seems reasonable but I would test the other modules and see if you get the same error.

 

So at this point, for certainty, I would like someone (a lab technician possibly) to tell me which of these sticks have been tested in the lab with which and assembled as one kit. It’s important.

A: I am sorry but that is not possible either, I would suggest requesting an RMA on Both sets either from the reseller and if its the reseller purchase a 64 Gig matched set to solve the issue. If the reseller will not or cannot help you then please use the link on the left and request an RMA and we can ask if our customer can do anything to help. However the only 64 Gig set are Domionator Platinum and you may have to use a different CPU cooler. Normally we are not able to change to another part number in an RMA but you are welcome to ask

 

In contrary, you are assuring me that the sticks with sequential serial nr from 843242-843245 are indeed the members of one kit, which is highly suspicious to me.

 

If I happened to have 2 kits with same lot code but with non-consecutive or with say at most 2-3 consecutive nr out of 8 modules, then what logic should I use to determine which of those belong together?

A: As I have stated all along in many different posts here on our forum we do not officially support mixing sets you can do this some times it does not work that is why we offer matched sets at a rated speed. In some cases it will work but they have to be installed in the correct slots so they are not miss matched. We tried to help you match the modules but there is no exact science to this it is up to you to match them when you install the modules.

 

And HU16E

The SN#'s are not always sequential they can be but not always in fact on some parts the SN# is added after the part is packaged and no SN# is on the module it self.

Link to comment
Share on other sites

we do not officially support mixing sets

 

Suppose, I’m an employee in a PC shop. Let say I have 2 PC’s to assemble for 2 different customers. One customer, named John, wants 32 GB LP kit installed in his PC-01. The other customer, named Charlie, wants the same 32 GB LP kit installed in his PC-02. I begin to assemble these 2 PC’s inside the shop in the same desk at the same time, because both customers have given me a short deadline and they want both computers to get ready to ship out by the same time.

 

Now, I open both RAM boxes and place the sticks UNMIXED near each PC and leave them unattended for a while because someone has called me outside. When I come back to the shop to continue my work I find that someone (a colleague of mine) has pushed “my stuff” in one pile to make room in that same desk for doing his own job. As you imagine, all Corsair sticks now mixed. I decide to randomly group them together, because I can’t tell which is which, here. I finish assembling my PC’s and put this for testing before shipping to customers.

 

While testing I see both PC have errors because sticks from one set are mixed from another. I know what is the reason (mixed sets) but I can’t revert the accident anymore because I don’t know which stick form one set is mixed with the other one. The deadline is approaching and my vendor exchanges if the product is proven to be defective, ONLY. So now I’m out of luck and in big trouble. Deadline has passed and my customers have started a legal case against me and now I’m in the brink of bankruptcy just because of a damn accident. How unlucky a person has to be to go through this?

 

I‘m just slightly a little better than this, though. My work has started to pile up and I’m wasting my time troubleshooting in void at this point. :(:

 

Not officially supporting mixed sets sounds acceptable to me from a technical point of view but denying my right to want to group my sticks back together to their respective kits after I have paid BIG money for it is not good business practice, in my opinion.

 

Such as arguing:

We tried to help you match the modules but there is no exact science to this it is up to you to match them.

 

And I don’t agree, it is not exact science. Indeed it is. If you tell me Match.com is not exact science when people are matched, I would believe you on that big time, but telling me RAM match is not exact science?!!!!.:eek::rofl:

 

Sorry about my rant here :rant: , but if you were in my situation, you would feel a little pissed just as I do now for not being able to revert such a stupid and mindless accident of mine back to the day I opened those damn boxes.

 

I would suggest requesting an RMA on Both sets either from the reseller and if its the reseller purchase a 64 Gig matched set to solve the issue. If the reseller will not or cannot help you then please use the link on the left and request an RMA and we can ask if our customer can do anything to help

 

RAM GUY thank you for the suggestion. I wanted it to be my last resort since RMA is costly too, in time and money (shipping & handling and possible custom duty charges), but as I can see I have already hit the wall with the troubleshooting thing :brick:.

 

The only problem is I need to wait “maybe” till Corsair starts to ship kits of 8 LP sticks from their facility. And the Dominator Platinum may or may not be an option for exchange as you mentioned, right?

 

And by the way RAM GUY, why are you writing in bold…..????? :thinking:

 

At HU16E

 

I would almost bet my life on them being a matched & tested together kit.

 

Ohhhh, c’mon HU16E. Don’t bet your life on that. My issues with earthly matters are NEVER as important as one’s life :D: . It's just a toy that is acting funny on me and it is trying to test my patience. I’ll overcome.

 

I guess I just need you to trust me on this.

 

Well, with all due respect I can’t trust you unless you give me a plausible argument why those belong to a kit, besides telling me they ARE sequential/consecutive.:nono: :biggrin:

 

Again, thank you all for trying to help with this. :praise:

Link to comment
Share on other sites

That seems reasonable but I would test the other modules and see if you get the same error.

 

Good point RAM GUY!!! :nodding: I'll test those tomorrow, since I decided to let the test on current ones run overnight.

Link to comment
Share on other sites

  • 3 weeks later...

********UPDATE*******

 

“But I didn't know until this day that it was the mainboard all along!!!” ;):

 

I suspected that the board was the next thing to look at, and lucky me the MB manufacturer has released a NEW BIOS update just at the right time. I flashed the BIOS to 1.9 and now to 2.1 ver. immediately and ran memtest again to see if that fixed my issue and it did, thankfully.

 

I ran memtest again on both groups of RAM separately (I refuse to call them kits for obvious reasons) after BIOS update for a full 8 and 6 passes respectively and both of them passed with no errors. I then populated all DIMM slots with both groups at the same time and ran memtest again for a full 8 passes without errors. So all 8 of them seem to coexist with no conflicts so far and, practically, I have had no freezes or lockups on my machine since.

 

However it still bugs me a little that I can't know for sure which of them make a set (as per manufacturer) for RMA purposes only, in case if smth breaks along the way...Hopefully NOT.

 

Now, as I mentioned I wanted to OC this beast CPU for a 24/7 setup. But I couldn't go more than 4.2 Ghz OC maintaining healthy parameters for longevity such as safe range of voltages and temps as per Intel recommendation. It seems like an average chip. :cool:

 

Safe voltage for this CPU range from 0.600 - 1.350v per Intel spec. Mine never exceeds 1.256v with all Power Savings features ON. TCASE is 66.8°C, and mine never exceeded 65 max core temp under full load stable for 12 hrs in Prime95 and IBT. Meanwhile my RAM settings were left at default in the BIOS:

 

CPUZ displaying:

 

DRAM Frequency 666.7 Mhz x2 = 1333Mhz

 

Latency:

 

9-9-9-24

 

Command Rate: 2T

 

Then I tested again with IBT, at first, after I switched RAM settings to XMP 1.3 Profile 1 in BIOS and noticed that the speed (GFlops) droped by half as much.

 

AIDA64 memory and cache benchmarking read also lower results. XMP1.3 sets my RAM to 1600 Mhz at 10-10-10-27 and 2T per manufacturer and the outcome I suppose should have been better than with BIOS default frequency and timings. Indeed it is the opposite.

 

So, why higher frequency but loose timings makes the benchmarking show poorer results?

 

Should I attempt to set frequency and timings manually at 1600 and 10-10-10-27?

 

How much OC headroom will this type of RAM allow if I want to tighten the timings a bit like 1600 and 9-9-9-24, will it support it?

 

Edit: I tried to set the timings ONLY at 9-9-9-24 manually while the XMP profile was active and tested the new settings with memtest and it was throwing tons of errors. Increased DRAM voltage to 1.545v and tested again but no luck, same thing. So I reverted back to XMP 1.3 factory settings.

 

Tried changing ONLY the timings from BIOS default, to 8-8-8-24 and looks stable so far, with every RAM and IMC voltages to default. Benchmarking now shows better results than with all parameters to BIOS default for RAM settings, but by a very small margin though.

 

<<<<<See IMG_05>>>>>>

 

Are there any other settings I should try to make my RAM run at factory frequency or higher that yields better results than the ones I already tried?

 

Any input is very much appreciated and thank you again for all your help, guys.

IMG_05.thumb.jpg.58208b4e93deb55103963f7d7643d5a6.jpg

Link to comment
Share on other sites

So, why higher frequency but loose timings makes the benchmarking show poorer results?

Looser timings probably. But a benchmark may not reflect real world performance.

Should I attempt to set frequency and timings manually at 1600 and 10-10-10-27?

It's not going to hurt anything. The worst that would happenis it would be unstable. You really cant damage the modules that way

 

How much OC headroom will this type of RAM allow if I want to tighten the timings a bit like 1600 and 9-9-9-24, will it support it?

There is absolutely no way to tell other than try and see what you can get out of them. Every set will OC differently just like a CPU or mainboard.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...