Help diagnosing RAM problems

[foster1984]

Hi guys, this is my first post here and might be a long post to explain my issue, so please bear with me and thanks in advance for your patience.

Recently I have been experiencing game crashes and random BSoDs while gaming; but only whilst gaming. Usually the game either just freezes or locks up and then exits to desktop, or sometimes it will go to BSoD which reads "bad module" or "Memory management issue".

So I would run the built in windows memory test immediately after the crash and it would detect a hardware fault; but MemTest86 and windows built in test would only detect a fault immediately after a crash. I actually thought it might be heat related and so replaced the thermal paste on my CPU and did a bit of a tidy up on the cable management yesterday. But after 30 odd minutes of gaming the other night, its started crashing again.
So I left MemTest86 running overnight and when I woke up and checked it, there were 23 errors showing over 5 test runs (please see attached picture). So it appears it was a RAM issue after all.

I therefore ran MemTest86 on another extended run again whilst I was at work, with stick 1 in slot 1, and then again with stick 2 in slot 1 while i go to the gym. these both showed errors, so I put stick 2 into slot 2 on it's own and ran the test again overnight. (The results of which can be seen in the pictures below, I unfortunately didn't take a picture of stick 1 in slot 1's results, but they looked very similar). Stick 1 is currently testing in slot 2 whilst I am at work typing this.

Top picture is test results of both sticks, 2nd picture is test results of Stick 2 in slot 2 overnight, 3rd picture is results of Stick 2 in slot 1 whilst I went to the gym for a few hours.

I would just like some help, figuring out the results of these tests, as I'm not 100% clear on what they mean. As I know it might not be the RAM and might be the DIMM slots on the motherboard or even CPU that is faulty, so I have borrowed some RAM from the IT guy at work, and will test that when I get home.

Depending on which or if both are faulty I will look to replace them. But I can't figure out what if faulty, as surely the chances of having 2 faulty RAM sticks is pretty low, but then I'd have thought the same of 2 faulty DIMM slots too.

Below I have included the FAQ requested system information. Sorry again for the long post :-(

• OS - Windows 10
• x64
• What OS was originaly installed on the system? Windows 7
• Is the OS an OEM version? Retail purchased OEM version of Windows 7 installed by me.
• Age of system: Approx 5 years
• Age of OS installation - Few months, as reinstalled a few months ago to clear up SSD space.
• CPU model: i5-4590
• Video Card model: MSI GTX970
• MotherBoard - Asus H81i-Plus
• Power Supply - Corsair CX750m
• System Manufacturer: N/A, self built
• Exact model number: N/A
• Laptop or Desktop? Desktop

[Radium_Angel]

@foster1984

 

Thanks for the detailed post, it's refreshing to get all the data up front as opposed to having to drag it out.

What MT86 is saying is you have bad ram, the software runs a checksum against the ram, and compares the results against known values. Anything different and it throws an error.

 

Now, assuming it isn't the mobo, a simple ram replacement will solve this. Known good ram from your IT guy will confirm this.

It *is* also possible your power supply is going bad, which is why under load (gaming) your OS crashes, as it can't supply enough power to the video card and other components at the same time. This has happened to me before, but let's start with the ram, and get back to us.

[foster1984]

@Radium_Angel

 

Thank you for your prompt reply. This is my first post here, but I have been a long time reader and have seen some of the drawn out troubleshooting caused by people not being upfront with all the information. So tried to supply as much as I could think of upfront.

 

The RAM I have borrowed works in the machine it was taken from, but I can't confirm it is "known good" as such; as my RAM works fine for everything except gaming. I only get any problems after about 30 minutes plus of gaming.

 

I don't think it is the PSU or the GPU, only because the BSoD is VERY rare, in fact it only happens while playing PUBG. All other games just freeze and then exit to desktop, which I wouldn't have thought possible if the GPU or PSU was failing. I am ready to be corrected, but would they not always cause a BSoD or black screen crash?

 

For expense reasons, I am hoping it is just the RAM. On a scale of probabilities, and in your opinion, how likely is it to be the MB or CPU? I assume it is still most likely to be the RAM?

 

I just know RAM faults are rare, and ATM it would appear that both my RAM sticks are faulty. :-( I will keep you updated when I get home tonight with the results of stick 1's extended test in slot 2. Then I'll test the IT guy's RAM in slot 1.

 

With Memtest86, is there any way to speed up the process? As currently I am having to test for 4-5+ hours at a time. It's very time consuming.

[Radium_Angel]

13 minutes ago, foster1984 said:

@Radium_Angel

 

Thank you for your prompt reply. This is my first post here, but I have been a long time reader and have seen some of the drawn out troubleshooting caused by people not being upfront with all the information. So tried to supply as much as I could think of upfront.

 

The RAM I have borrowed works in the machine it was taken from, but I can't confirm it is "known good" as such; as my RAM works fine for everything except gaming. I only get any problems after about 30 minutes plus of gaming.

 

I don't think it is the PSU or the GPU, only because the BSoD is VERY rare, in fact it only happens while playing PUBG. All other games just freeze and then exit to desktop, which I wouldn't have thought possible if the GPU or PSU was failing. I am ready to be corrected, but would they not always cause a BSoD or black screen crash?

 

For expense reasons, I am hoping it is just the RAM. On a scale of probabilities, and in your opinion, how likely is it to be the MB or CPU? I assume it is still most likely to be the RAM?

 

I just know RAM faults are rare, and ATM it would appear that both my RAM sticks are faulty. :-( I will keep you updated when I get home tonight with the results of stick 1's extended test in slot 2. Then I'll test the IT guy's RAM in slot 1.

 

With Memtest86, is there any way to speed up the process? As currently I am having to test for 4-5+ hours at a time. It's very time consuming.

Bad ram is more common than you think. Heat will do them in as well. No way to speed up the process. In fact I recommend running it for 24 hours at least, just to really work over the ram.

 

PUBG is well known here, and on the web in general, for being buggy and poorly coded, so a BSoD from that makes sense to me.

In my decades of being a professional IT guy, I can count on maybe 1 finger the number of bad CPUs I've seen. I've even re-used CPUs from systems that have been destroyed by lightning strikes. Motherboards are a different matter, but again, experience says to me motherboard issues manifest themselves in different ways, not in bad ram.

 

[foster1984]

@Radium_Angel

 

So likely faulty parts are RAM, Motherboard, CPU in that order? That is OK, as at least that is least to most expensive components in terms of fault likelihood.

 

The heat would make sense, as I have an mITX build in a Coolermaster Elite 120 Advanced case. Prior to redoing the thermal paste on the CPU and blowing out everything with compressed air, I had measured the CPU at a consistent 95c. And since that was the first proper clean out the PC had been given since it was built (other than when I opened her up to put a new HDD in), I can imagine the RAM has been cooking in there.

 

I am really crossing my fingers that when I get home and test the IT provided RAM that it tests as OK. As that will conclude that it is the RAM at fault, if not, I am back to troubleshooting.

 

But I will keep you updated.

[Radium_Angel]

1 hour ago, foster1984 said:

@Radium_Angel

 

So likely faulty parts are RAM, Motherboard, CPU in that order? That is OK, as at least that is least to most expensive components in terms of fault likelihood.

 

The heat would make sense, as I have an mITX build in a Coolermaster Elite 120 Advanced case. Prior to redoing the thermal paste on the CPU and blowing out everything with compressed air, I had measured the CPU at a consistent 95c. And since that was the first proper clean out the PC had been given since it was built (other than when I opened her up to put a new HDD in), I can imagine the RAM has been cooking in there.

 

I am really crossing my fingers that when I get home and test the IT provided RAM that it tests as OK. As that will conclude that it is the RAM at fault, if not, I am back to troubleshooting.

 

But I will keep you updated.

Sounds like a plan, and 95c is *waaaaaay* too hot for a CPU. My core i5, running with a stock cooler, (and granted, a mid tower case) ran at 30c under load...

[foster1984]

After the compressed air clean out, re-thermal pasting and tidier cable management, I had a game of Rainbow Six Siege; this caused the CPU temp to go up to 80-85c, so still hot but I was happy with an immediate 15c decrease in max temp. It's always going to be hot in my case, its small so not a lot of room for air flow.

Below are some pictures of before and after in terms of space in the case (the first 2 are before the cleaning, etc and the 3rd is after); as you'll see, its a lot better, but no where near optimal and not much more I can do with the space.

[Ace McPlane]

1 hour ago, foster1984 said:

After the compressed air clean out, re-thermal pasting and tidier cable management, I had a game of Rainbow Six Siege; this caused the CPU temp to go up to 80-85c, so still hot but I was happy with an immediate 15c decrease in max temp. It's always going to be hot in my case, its small so not a lot of room for air flow.

Below are some pictures of before and after in terms of space in the case (the first 2 are before the cleaning, etc and the 3rd is after); as you'll see, its a lot better, but no where near optimal and not much more I can do with the space.

Yea, that ram is definitely bad.

 

What kind of CPU cooler are you using though?

85 is still pretty hot.

If you can fit one, maybe put an evo 212 on there since they're only like 20-30 bucks and are actually very capable coolers.

[foster1984]

1 hour ago, stateofpsychosis said:

Yea, that ram is definitely bad.

 

What kind of CPU cooler are you using though?

85 is still pretty hot.

If you can fit one, maybe put an evo 212 on there since they're only like 20-30 bucks and are actually very capable coolers.

Are you saying the RAM is bad from the test results or from the pictures of it in the case?

 

I'm using the stock Intel cooler, as from the pictures you can see there is barely any clearance between the CPU and the PSU. I could fit a low profile cooler in there, but only select ones are any good for my case.

 

I will upgrade to a full sized case and ATX MB, etc at some point when funds allow. But right now I have to get it working again and all funds will be used towards that.

[foster1984]

OK, so just got home from work and below is the test result screen.

 

I'm guessing that is a definitely bad RAM stick is it? It has posted an error address for the fault, so that part of that stick is bad, is my thinking correct?

 

I'm about to start testing with the stick I was given by my work's IT guy now. Hopefully, this won't return any errors and I know I just have 2 faulty RAM sticks.

[Ace McPlane]

1 hour ago, foster1984 said:

Are you saying the RAM is bad from the test results or from the pictures of it in the case?

 

I'm using the stock Intel cooler, as from the pictures you can see there is barely any clearance between the CPU and the PSU. I could fit a low profile cooler in there, but only select ones are any good for my case.

 

I will upgrade to a full sized case and ATX MB, etc at some point when funds allow. But right now I have to get it working again and all funds will be used towards that.

Yea, you're right.

What am I saying?

That's a small form factor build so yea only something low profile would work unless you could fit maybe an h60 in there.  Hey, plug all of your system's parts into pcpartpicker and see what coolers it says that'll fit your system. You'll get a good idea that way.

 

Yea, I'm saying by the memtest that your ram is bad unless you've been overclocking it and didn't do it right..

it's either one or the other.

A memtest shouldn't come back with even a single error if you ram is good.

[foster1984]

@Radium_Angel and @stateofpsychosis

 

I will probably leave the cooler for now. I am moving in the next couple of weeks and my mate is going to give me a new case as a moving present, so I will get a cooler then. rather than buy a low profile one, only to want to replace it with a full size one in a couple weeks.

 

Nope, never overclock the RAM. Or anything for that matter, I prefer stability. Plus I have a H81 board, so don't even think overclocking is supported.

 

So I have run a test of the borrowed RAM using Memtest86 v7. The results can be seen in the pictures below. It passed, although the new Memtest86 does only run 4 passes, unlike v4.3 which just kept testing constantly.

 

So I'm going to try this stick in DIMM2, just be sure; but it looks like I just have 2 faulty RAM sticks.

 

Thoughts?

[Ace McPlane]

11 minutes ago, foster1984 said:

@Radium_Angel and @stateofpsychosis

 

I will probably leave the cooler for now. I am moving in the next couple of weeks and my mate is going to give me a new case as a moving present, so I will get a cooler then. rather than buy a low profile one, only to want to replace it with a full size one in a couple weeks.

 

Nope, never overclock the RAM. Or anything for that matter, I prefer stability. Plus I have a H81 board, so don't even think overclocking is supported.

 

So I have run a test of the borrowed RAM using Memtest86 v7. The results can be seen in the pictures below. It passed, although the new Memtest86 does only run 4 passes, unlike v4.3 which just kept testing constantly.

 

So I'm going to try this stick in DIMM2, just be sure; but it looks like I just have 2 faulty RAM sticks.

 

Thoughts?

Well, it's not just about whether ram is bad or not either.

It can also be a compatibility thing. See, XMP profiles are only guaranteed to be stable on systems that mark that ram as compatible where it's been tested and confirmed so it's kind of important to read through that list before you buy. You can find this in the same place where you download your motherboard drivers from.

It's probably something like that.

 

Yea, 4 passes really isn't enough.

I'd fire back up the other version and let it run overnight.

About 24 hours.

It can take quite a few passes before an error turns up at times.

 

 

[foster1984]

The RAM that I have been using (the stuff I'm not trying to figure out if it's at fault) has been fine in the system for years and was listed as compatible on PCPartPicker when I was looking for components. So don't think it's a compatibility issue.

 

The RAM that passed the new test was the borrowed stuff from work, so I am hoping that passes. But I will test it using the old v4.3 of Memtest86 rather than v7. I can run it overnight and see if it errors, then switch slots and run it while I'm at work. Fortunately timescale limits don't permit that I can test it in both slots for 24 hours, as they have an inventory check coming up and it has t be returned by Friday morning. So a 7 hour overnight test in one slot and a 7 hour test in the other slot will have to do it.

 

That will likely give it about 8-10 passes, in that timescale. If there are no errors in that time with the loaned RAM, when my RAM gave me 100s of errors in the timeframe, I think it would be safe to assume it's not the MB, CPU or PSU that is causing an issue, and it is my RAM that's faulty, surely?

[Radium_Angel]

48 minutes ago, foster1984 said:

The RAM that I have been using (the stuff I'm not trying to figure out if it's at fault) has been fine in the system for years and was listed as compatible on PCPartPicker when I was looking for components. So don't think it's a compatibility issue.

 

The RAM that passed the new test was the borrowed stuff from work, so I am hoping that passes. But I will test it using the old v4.3 of Memtest86 rather than v7. I can run it overnight and see if it errors, then switch slots and run it while I'm at work. Fortunately timescale limits don't permit that I can test it in both slots for 24 hours, as they have an inventory check coming up and it has t be returned by Friday morning. So a 7 hour overnight test in one slot and a 7 hour test in the other slot will have to do it.

 

That will likely give it about 8-10 passes, in that timescale. If there are no errors in that time with the loaned RAM, when my RAM gave me 100s of errors in the timeframe, I think it would be safe to assume it's not the MB, CPU or PSU that is causing an issue, and it is my RAM that's faulty, surely?

I'd agree here, the RAM is bad.

And you will be cooking your next sticks if you are not careful. Maybe run with the case open?

[foster1984]

25 minutes ago, Radium_Angel said:

I'd agree here, the RAM is bad.

And you will be cooking your next sticks if you are not careful. Maybe run with the case open?

OK, I'll give the borrowed RAM and extended test in DIMM1 overnight and then an extended test in DIMM2 tomorrow while I'm at work. Provided that it returns no errors from either of them, I will know it is my RAM which is faulty.

 

I can't really do that, due to cats and my young daughter; but it took 5 years for the current RAM to go bad, so a maximum of 2 weeks until I get a bigger case shouldn't do the new RAM any harm.

 

I'll keep you guys posted on how things go.

 

If it is the RAM that's faulty, should I replace it with 2 4GB sticks like I had, or go for 1 8GB stick? (maybe upgrade with another 8GB stick in future, funds permitting) I'm only really interested in if dual channel vs single channel will make a noticeable difference in gaming, as everything else I use my PC for is basic and won't come close to maxing the RAM.

[Radium_Angel]

1 hour ago, foster1984 said:

OK, I'll give the borrowed RAM and extended test in DIMM1 overnight and then an extended test in DIMM2 tomorrow while I'm at work. Provided that it returns no errors from either of them, I will know it is my RAM which is faulty.

 

I can't really do that, due to cats and my young daughter; but it took 5 years for the current RAM to go bad, so a maximum of 2 weeks until I get a bigger case shouldn't do the new RAM any harm.

 

I'll keep you guys posted on how things go.

 

If it is the RAM that's faulty, should I replace it with 2 4GB sticks like I had, or go for 1 8GB stick? (maybe upgrade with another 8GB stick in future, funds permitting) I'm only really interested in if dual channel vs single channel will make a noticeable difference in gaming, as everything else I use my PC for is basic and won't come close to maxing the RAM.

Based on your specs, your RAM won't be the limiting factor.

[foster1984]

For your daily update on the actual problem, below is a picture of the results of the borrowed RAM after it's extended test overnight in DIMM 1. As you can see, no errors. So it appears the issue was that my RAM is faulty. The borrowed RAM is currently having an extended test in DIMM 2 while I'm at work, just to confirm this though.

 

As this appears to be the case, and the RAM isn't a factor for me, I will likely just get 1 8GB stick for now then. As I can then upgrade to 16GB at a later date.

 

What would be the limiting factor in my setup? The CPU or the GPU? I currently game on a 1080p 60Hz TV from my sofa, but am about to switch to a 1080p 60Hz monitor at a desk to reduce input lag.

[foster1984]

@Radium_Angel & @stateofpsychosis

 

It is definitely my RAM that is faulty. Below is a picture of the results screen from the borrowed RAM's extended test in slot 2 while I was at work...again no errors.

 

I'll pick up some RAM on Monday and all should be right in the world again.

 

Thanks for all your help guys!

 

P.S How do I mark this thread as solved?

[Ace McPlane]

2 minutes ago, foster1984 said:

@Radium_Angel & @stateofpsychosis

 

It is definitely my RAM that is faulty. Below is a picture of the results screen from the borrowed RAM's extended test in slot 2 while I was at work...again no errors.

 

I'll pick up some RAM on Monday and all should be right in the world again.

 

Thanks for all your help guys!

 

P.S How do I mark this thread as solved?

Nice, I'm glad you figured it out.

Just find the first comment that suggested the correct fix and you can make that comment as the solution with the options below it.