[Solved] Hackintosh Killing RAM

So this is a bit strange,

I purchased four sticks of 4GB PC3 10600R memory for use in my Dell t3600-based hackintosh (I used a pulled motherboard) used off of eBay from a reputable seller. The system runs fine usually for a little while, then starts to bog down, a LOT. The graphics get choppy, it responds very slowly, and eventually freezes. Upon reboot after an episode like this, the machine usually tells me it has found errors with DIMMs, and has corrected them, usually it recommends DIMM replacement. I thought it was just one bad RAM stick, but then the next one failed, then the next. I have one working DIMM that I'm running off of now, and sometimes the system still locks up, and reports errors on reboot. I would have thought it was bad RAM, but since it's eating these like candy bars, is this indicative that there is a problem with my motherboard?

Are there any pop-corning or mushrooming capacitors?

Did you set the correct settings in the BIOS? (Does the Dell BIOS allow RAM settings?) Did you update the BIOS to the last latest version? Did you run MemTest86 on the memory for at least 2 passes, but preferably 24 hours?

Thanks for your reply.

I checked visually for blown capacitors, there is nothing obvious.

Dell doesn't allow any customization of RAM in the bios, I will have to check if the bios is the latest version, I tried running memtest but just from within the OS, so I don't think it is getting all the RAM. I ordered some memory specifically for my machine, if that fails too then I'll know it was my MOBO or CPU

You can download MemTest86 and burn it to USB or CD. http://www.memtest86.com

I believe that your PC can run Registered (which you have) and unbuffered, iirc.

http://www.crucial.com/usa/en/compatible-upgrade-for/Dell/precision-workstation-t3600

By running MemTest86 your are isolating the hardware as a problem from the software (OSX).

Did you mix and match RAM, different manufacturers, different speeds, different chips? I suggest they be bought as a single kit so that they are all identical.

When looking at the RAM slots do you see any crud on the pins? Blow them out with canned air, but don't blow ice cold air on the pins. You have to be very patient so that frost doesn't form on the pins, so go very slow and intermittent.

When you removed the old RAM was there any green crud on the fingers?

Did you install the RAM with the AC cord plugged in? You always want to install components with the AC cord removed. Never touch the pins, etches or chips when handling and installing RAM.

Pay particular attention to the capacitors in the RAM area. You may have to use a strong flashlight and a magnifying glass to check for leakage at the bottom of the caps. Typically the tops of the caps mushroom, instead of being flat, but sometimes they leak from the underside and don't mushroom or get fat.

But before running you may want to re-seat the processors and lave on some fresh thermal paste, say some Shin-Etsu G751. When you remove the processor look at the motherboard pins and make sure that old specks of thermal paste aren't on the pins and check for bent pins. Or did you use Artic Silver? Some of it have have dribbled down into the pins and shorted out a circuit. Hopefully you're using stock heatsinks and haven't crushed any pins.

Lastly try a different PSU, if possible, since RAM runs off the 3.3V rail. If you have an under powered PSU then you may not be supplying enough voltage to the RAM. You may even have to take apart the PSU to see if there is any pop-corning or leaking caps.

Lastly, if you are running the PC off a voltage UPS and the UPS doesn't support the PSU draw you should hear a high pitched whine as it gets over driven. You wouldn't want a 350W UPS driving a 650W PSU, for example.

That's very helpful, thank you.

I examined the mobo but not the underside of the caps, and I could probably have a better look at the slots. I may also check the CPU for bent pins, the cooler is the one recommended by my local computer shop and came with thermal paste on it, a little.

I didn't unplug the AC every time, so maybe that's it, I will definitely be more careful.

I'll try to report back what worked, if anything.

My similar story: 2 days ago my Hackintosh froze up completely while I was working, I later found out it was due to an Xcode memory leak. Since then my PC that has been working fine for the past 5 years, cannot boot with my RAM set to its stock 1866MHz frequency. (only boots if 1333 or 1600MHz are selected)

Most probably this is a problem with the motherboard, but the fact that I've been using OSX all day for the past month made me suspicious that the actual OS may be affecting my RAM (after the freeze, the PC would not shut down, neither under windows nor under OSX, until I cleared my CMOS and reset the BIOS). My first though was that my hackintosh killed my ram, as you suggest here. I know this makes no sense, but can someone verify that the software cannot mess with the actual RAM?

P.S.: I've ran memtest for 10 hours and no errors, last resort is to try my RAMs on a different PC. Just wanted to share my concerns with you here. Most probably these are just very bad coincidences.

I had a similar experience in that I ran the onboard memory diagnostic (the DELL hardware test) and it didn't find anything, but at boot-up it would warn me there were errors in memory. Is it the OS? Maybe? I wish I knew how to tell. I can't change the RAM settings with this board so we will see. I'm beginning to wish I just bought another actual Mac!

Wish you the best of luck, keep me posted on anything you find. I'll be testing my RAMs on another board in about 2 weeks once a friend gets back, to see if my mobo is actually causing the problems. I'm pretty sure OSX has nothing to do with our problems, but who knows.

On your original problem, have you tried running activity monitor who the OS becomes extremely slow? Is the ram full, or it's slow with normal ram usage? (when Xcode f***d me up, it was "using" 48GB of swapped RAM and I could not even move the mouse for 10 minutes)

Did you eventually run the memtest?

Yes, I definitely was using Activity Monitor and sometimes the command-line tool "top" to see what was causing the extreme slowdown, since that's what I would have checked on my real Mac. In my case the memory pressure was low, nothing unusual was happening, and CPU usage was at at around 100% for a photoindexd (or something) indexing my photo library, but that happens on my real mac without causing issues, and even my macbook can go all the way up to 400% CPU usage due to the four cores, my hackintosh is a 6-core Xeon e-2640 and all temperatures seem normal using HWmonitor. The Console didn't give me any clues either, just normal messages as far as I can tell. I would still get random reboots and sudden slowdowns, which leads me to believe it really is a hardware problem, unless MacOS is somehow messing with the memory during operation, but I don't know how it could be doing that. It could be a disconnect between my CPU and the motherboard, so I will check that when I have some time in a couple days. I'm really not looking forward to unscrewing everything and reapplying thermal paste but oh well. For now I have to get back to work.

I didn't boot with memtest, just ran from inside the OS, but that will be the next step, after I put the graphics card back in, etc. Since I was examining the board.

moosefuel said:

... the cooler is the one recommended by my local computer shop and came with thermal paste on it, a little.

Click to expand...

Unless it was a Noctua, Cryorig, etc., or any heat sink that won't allow the crushing of pins, like Dell stock, HP stock, etc., then it is still suspect.

Same goes for SSDs since some Samsung EVO TLC drives have been known to slow down while accessing old files or an SSD being 80% full.

but at boot-up it would warn me there were errors in memory.

Click to expand...

Heck, for all we know it could be a virus or malware; in the case of Windows it could be that it went into hibernation and then we shut it down through a power switch. In Windows 10 you would open an Admin terminal and issue "powercfg.exe -h off" to disable Hibernation. That probably won't affect most users but since Linux detects the drive as having problems when gparted runs it could affect those running OSX and Windows on the same PC.

What RAM did you get, Registered or Unbuffered? As I said you probably want to get a single kit of four identical sticks as there is less chance of problems. Mix and match, even from the same manufacturer with the same part numbers, could have problems (which may not show up until the RAM is really taxed.)

Is there a chance that it's your mobo? Slight chance. Some HP mobos have their secondary channels go bad, so two slots tend to die at a time. Idk about Dells, though. Do you have access to another PSU that you can temporarily hook up? It should be greater than 635W because that is what the stock PSU rating is.

ftp://ftp.dell.com/manuals/all-products/esuprt_desktop/esuprt_dell_precision_workstation/precision-t3600_owner%27s%20manual_en-us.pdf page 21 shows what your heat sink should look like.

Now that I've seen what your caps look like, page 35, that is probably not your problem since these are high quality caps not subject to pop-corning. If it isn't a crushed CPU pin, or a bad CPU, then it could be a PCI-E board, graphics or otherwise. If you need to get into the BIOS make sure that the USB keyboard is in the USB2.0 slot only. PS2 keyboards obviously go to their respective round connector. I've actually had bad mice and bad keyboards prevent a PC from powering up and one or two that could only use a PS2 kb to be able to get into the BIOS. Probably that's not your problem.

http://www.playtool.com/pages/psurailhistory/rails.html
http://www.jonnyguru.com/forums/archive/index.php/t-12921.html
etc. As the last link said, it could be a VRM. In which you would probably need a new motherboard. No, you don't really have a VRM, Voltage Regulator Module; what you do have is the usual MOSFets / voltage regulator chips around the CPU. If one were bad I would expect the unit not to even power up.

So, it could be that when you installed the RAM chips with the AC cord connected that it either fried a RAM stick and/or a mem controller circuit inside the CPU; "fried" being a relative terms because it could have weakened it, with the effects usually showing up months later.

Don't even think of removing your CPU heatsink without having a tube of thermal paste on hand.

Is that a stock CPU you're running?