[SOLVED] Bsods due to memory, GPU or PSU?

[swamptree]

I get blue screen errors almost daily since i initially built this rig at the end of 2013, I was not really able to trouble shoot things because at the time win7 wouldn't update or producing minidump files for me, Aswell i had some serious health setbacks in the beginning of 2014 with flesh eating bacteria, i've been focused on recovering from the physical and overcoming becoming partially blind .

In January of 2019 i upgraded components, I get new cpu water cooling solution, new gpu (rx 590) and upgrade to SSD (samsung evo 860), I perform clean windows install and update both win7 and all available chipset/drivers and bios + ME from manufacture (msi z87-g45).


The memory bsod i am referencing started when i first updated the gpu, very rarelyoing web browsing/streaming/video play back (would hang for example when closing browsers or attempting to restart) but the majority of issues seemed especially prominent playing newer gpu intensive games which will result in my system hanging and giving some sort bsod of memory error related message and pointing to possibly faulty hardware as cause half the time.

Often I see page fault in non paged area, bad pool header, memory management, irq no more or less, System service exception and most of these are tied either into Direct x/graphic driver, Of the last ten crashes, 4 of them mentioned possible faulty hardware in whocrashed though i doubt it is that transparent of a problem, a proper debugging would really illuminate actual driver or process causing the issues.

So i've tested both sticks together two times with 5 passes no problem, I reseated my ram from dims 2/4 to 1/3 but that did not solve things, I took out a single stick of ram last night and let it run 5 passes with no issue and i even managed to get about six straight hours of gaming when it would crash easily within the first 30 minutes so this is why i say it might be memory related if it runs fine on 8gb but not both sticks of ram.

I will be switching over to the other stick today and will attempt to stress test things and see if the system hangs on the secondary stick, I believe my dim slots are functioning and in good order but im willing to dust them and clean the contacts if necessary. If the other stick pulls up errors on its own then the answer is simple, I need to replace my set of ram in entirety.

What happens though if i end up with no issues on the other stick and system stability is good?

Ive seen it suggested that if running two sticks is unstable but they test clean on their own to try bumping the voltage on the ram a tiny bit, My ram is rated for 1.5v (cmz16gx3m2a1600c10) and i have manually set the voltage now to 1.5v since it was running undervolted at 1.488v via the bios auto settings, timing settings are auto set but they read correct correct to specs of the ram. Ram was also bought in a set!

I'd appreciate any help in trouble shooting this issue please!

 

[geofelt]

If you are testing with memtest86 and can complete 5 passes with NO errors, I would exclude ram as an issue.

What are the rest of your specs?

A new RX590 is relatively power hungry and should probably have a quality 600w psu.
I suspect your problems are either gpu or psu related.

 

[swamptree]

If you are testing with memtest86 and can complete 5 passes with NO errors, I would exclude ram as an issue.

What are the rest of your specs?

A new RX590 is relatively power hungry and should probably have a quality 600w psu.
I suspect your problems are either gpu or psu related.

Specs,

MSI z87-g45 (1.9 bios + me update)
Real-Tek High Def Audio
killer e2200 ethernet adaptor
i5-4670 non-k
2x8gb sticks corsair vengeance ddr3 1600mhz 10-10-10-27 1.5v
1 hdd x Western Digital Blue 1 tb 7400
1 ssd x Samsung evo 860 1 tb (main drive)
gpu x Sapphire Nitro+ Rx 590
Cooling x corsair h60 for cpu cooler and a few 120mms for intake
psu x 750 watt Bronze thermaltake 80+

Peripherals plugged in,
1 x razor black widow 2013
1 x logitech g400s
1 x logitech c920


The memory specific bsod started after install of the rx 590, Though i confess i am not entirely certain the crashes previous were not the same here, I do feel i am seeing new BSOD stop codes i've never encountered prior to install of said gpu, The gpu will run bench marking software seemingly with no issues (i ran super position from unigine for 1.5 hours on 1080p extreme with no issue beyond the terrible fps LOL)

Also as i mentioned i pulled a single stick out last night and got six hours of resident evil 2 remake with no crashes, previous with two sticks in, I would hang within an hour or often within the first 30 minutes, so that is why i suspect hardware.

 

[geofelt]

A low quality psu can cause all sorts of problems.
Thermaltake has units ranging in quality from tier 1 to tier 7(awful)
https://linustechtips.com/main/topic/631048-psu-tier-list-updated/

Can you borrow a known good quality psu in the 600w range to test with?

 

[swamptree]

A low quality psu can cause all sorts of problems.
Thermaltake has units ranging in quality from tier 1 to tier 7(awful)
https://linustechtips.com/main/topic/631048-psu-tier-list-updated/

Can you borrow a known good quality psu in the 600w range to test with?

no i don't have a available psu to swap for a test but i could just buy one, my current psu is a 2013 thermaltake smart m series 750w, looking at your list, it mentioned this psu is rated as tier three, I'm not exactly sure of the significance beyond it is not as good as tier 2 and 1 obviously

you mentioned this is either gpu or psu, so i tried something for both just now, stuck another pci-e power adaptor onto the psu it so the card is running on two rails now instead of daisy chained off of 1, i do understand you can get between 300w to 360w on a single rail for pci-e but ive read some people were having stability issues.with high draw cards apparently.

 

[geofelt]

A pcie slot has a nominal capability to deliver 75w via the slot.
That is why strong graphics cards may require added 6 or 8 pin power connectors.
6 pin can be 75w and 8 pin can be 150w.

Looking at the motherboard, in addition to the standard 24 pin psu connector, the motherboard has a 8 pin aux cpu connector which should have been populated.
The purpose of the aux cpu connectors is to provide extra 12v power for overclocking or extra pcie slot power.
A motherboard that will run without the full 8 pins connected may not do so if you overclock or have a fully loaded pcie configuration.

Using two cables vs. one does not matter.
It will not hurt, but a single cable is better for cable management.
Most all power supplies have only one power generating rail.
The division you see is a artificial UL separation for safety.

As to tier lists, take them with a grain of salt and only a guideline.
There is some opinion involved.
Still such a list is useful if you do not know all about power supplies.
I include myself in the less than knowledgeable list.
Tier 3 is normally fine.
Find a review on a respected psu review site such as jonnyguru for a better opinion.

 

[swamptree]

Correct me if i am wrong, my wattage for this card is rated at 238w to 250w,, using the pci-e slot for 75watt and a 8 pin pci-e power cable we get 150w for a combined total of 225w. this means we are still minus 13 watts at a minimum and to a max 25w, So running a secondary pci-e power cable should be done then correct? or am i mistaken in how this works. I keep seeing guides reference a single pci-e power cable capable of between 75w to 150w dependant of the 6 or 8 pin connector but i also see people reference a total 300watt per power cable aswell, I'm going to assume the 150w figure is correct.

This is my second day with the single stick of ram and i have yet to see any crashes yet of any games or any bsods either, I really hope we start seeing the errors when i stick that other stick in confirming its just failing ram.

A pcie slot has a nominal capability to deliver 75w via the slot.
That is why strong graphics cards may require added 6 or 8 pin power connectors.
6 pin can be 75w and 8 pin can be 150w.

Looking at the motherboard, in addition to the standard 24 pin psu connector, the motherboard has a 8 pin aux cpu connector which should have been populated.
The purpose of the aux cpu connectors is to provide extra 12v power for overclocking or extra pcie slot power.
A motherboard that will run without the full 8 pins connected may not do so if you overclock or have a fully loaded pcie configuration.

Using two cables vs. one does not matter.
It will not hurt, but a single cable is better for cable management.
Most all power supplies have only one power generating rail.
The division you see is a artificial UL separation for safety.

As to tier lists, take them with a grain of salt and only a guideline.
There is some opinion involved.
Still such a list is useful if you do not know all about power supplies.
I include myself in the less than knowledgeable list.
Tier 3 is normally fine.
Find a review on a respected psu review site such as jonnyguru for a better opinion.

 

[swamptree]

Correct me if i am wrong, my wattage for this card is rated at 238w to 250w,, using the pci-e slot for 75watt and a 8 pin pci-e power cable we get 150w+75w for a combined total of 225w. this means we are still minus 13 watts at a minimum and to a max 25w, So running a secondary pci-e power cable should be done then correct to take of the excess wattage the pci-e slot and pci-e 8 pin are covering? or am i mistaken in how this works. I keep seeing guides reference a single pci-e power cable capable of between 75w to 150w dependant of the 6 or 8 pin connector but i also see people reference a total 300watt per power cable aswell, I'm going to assume the 150w figure is correct.

This is my second day with the single stick of ram and i have yet to see any crashes yet of any games or any bsods either, I really hope we start seeing the errors when i stick that other stick in confirming its just failing ram.

I read a few indepth reviews for my psu, it seems to be of a sufficient quality and performs within spec.

A pcie slot has a nominal capability to deliver 75w via the slot.
That is why strong graphics cards may require added 6 or 8 pin power connectors.
6 pin can be 75w and 8 pin can be 150w.

Looking at the motherboard, in addition to the standard 24 pin psu connector, the motherboard has a 8 pin aux cpu connector which should have been populated.
The purpose of the aux cpu connectors is to provide extra 12v power for overclocking or extra pcie slot power.
A motherboard that will run without the full 8 pins connected may not do so if you overclock or have a fully loaded pcie configuration.

Using two cables vs. one does not matter.
It will not hurt, but a single cable is better for cable management.
Most all power supplies have only one power generating rail.
The division you see is a artificial UL separation for safety.

As to tier lists, take them with a grain of salt and only a guideline.
There is some opinion involved.
Still such a list is useful if you do not know all about power supplies.
I include myself in the less than knowledgeable list.
Tier 3 is normally fine.
Find a review on a respected psu review site such as jonnyguru for a better opinion.

 

[geofelt]

Whatever 6 or 8 pin power connector a graphics card has, it is because it needs them to run properly.
Some cards get overclocked and will need right up to the maximum that the power connectors can supply.

A low quality psu may not deliver advertised power.
The ratings sometimes lie and are measured at room temperature, not at the higher temperatures found inside a case.
The power may also be peak power, not continuous. The power delivery may fluctuate and cause instability.
An aged psu may also deteriorate.
A cheap psu will not have the protective circuitry to protect your parts if it should fail under load.

I do not want to be an alarmist, but a bad psu can cause strange symptoms.
Hence the suggestion to see if you can solidly verify that yours is good.

If you can't borrow a good one, perhaps you can buy one from a shop with a good return policy.
Expect to pay a 15% restocking fee if the psu is not the problem and you want to return it.
I like the Seasonic focus gold units. Perhaps the 650w one.

 

[swamptree]

Whatever 6 or 8 pin power connector a graphics card has, it is because it needs them to run properly.
Some cards get overclocked and will need right up to the maximum that the power connectors can supply.

A low quality psu may not deliver advertised power.
The ratings sometimes lie and are measured at room temperature, not at the higher temperatures found inside a case.
The power may also be peak power, not continuous. The power delivery may fluctuate and cause instability.
An aged psu may also deteriorate.
A cheap psu will not have the protective circuitry to protect your parts if it should fail under load.

I do not want to be an alarmist, but a bad psu can cause strange symptoms.
Hence the suggestion to see if you can solidly verify that yours is good.

If you can't borrow a good one, perhaps you can buy one from a shop with a good return policy.
Expect to pay a 15% restocking fee if the psu is not the problem and you want to return it.
I like the Seasonic focus gold units. Perhaps the 650w one.

New PSU is in the works but the closest pc store is about 2 hours away by bus for me (4 hours total back and forth) so i will have to get to that this week since bus service on sunday is limited, I understand you keep pushing me to replace my psu but i do not understand why i should be replacing my psu when the bsod error stop when i switch to single stick of ram instead of two, if you could enlighten why you still feel it is related to psu, i would appreciate to understand your logic here please!

on topic of memory, I switched onto stick #2 of ram last night, Performance was slightly crap, Games would hang for up to 5 to 10 seconds at moments but then things would keep rolling, Whereas with Stick #1, I would not have any of these slight hang issues. Stick# 2 did pass 10 passes last night with Memtest86, I observed the dram voltage on #2 fluctuating consisently between 1.48v and 1.506v, Stick #1 was at a solid 1.48v via bios monitoring.

Age is def a factor on why you suggest to replace psu, one reason i can accept on why i should replace but these issues have been consistent since build, The issues disappear when the ram is pair down to a single stick.

I'm going to rma this gpu when i buy a new psu, the store will just swap it there for me, kill two birds with one stone right?

corsair TX750M psu is tier 1 correct?

 

[geofelt]

I am not pushing for a new psu, only pointing out that a flaky psu can cause all sorts of problems.

You may be on to something with ram.
Are both sticks from a single matched kit?
If not, that could be the problem.
But, since you exercised all the ram together using memtest86, Ram should not be the issue.

Can you take your pc with you when you visit the store?
They should be able to test out by replacing parts.

Corsair has a lifetime warranty on ram, perhaps you could test with a replacement kit.

Since it takes some time for your problem to develop, you may not resolve it at the shop.

TX750M is well reviewed by jonnyguru:
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=528

 

[swamptree]

I am not pushing for a new psu, only pointing out that a flaky psu can cause all sorts of problems.

You may be on to something with ram.
Are both sticks from a single matched kit?
If not, that could be the problem.
But, since you exercised all the ram together using memtest86, Ram should not be the issue.

Can you take your pc with you when you visit the store?
They should be able to test out by replacing parts.

Corsair has a lifetime warranty on ram, perhaps you could test with a replacement kit.

Since it takes some time for your problem to develop, you may not resolve it at the shop.

TX750M is well reviewed by jonnyguru:
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=528

Yessir it was indeed purchased in a set of two sticks, I had my first blue screen last night with stick #2 while gaming after i removed my slight dram voltage boost ( i bumped it to 1.546v from the 1.5v it was auto set for)

infact i got CRITICAL_STRUCTURE_CORRUPTION which was a completely brand new one to me, could be a driver, could be faulty ram via a whocrashed.exe report, gee a mystery one could say,

Now i just need to re-verify stick #1 and see if my results of error free operation were true, I mean it passed 48 hours of uptime, I managed multiple 3ish hour playthroughs without a bsod with good observable performance and this was at stock dram voltage.

With stick #2 dram voltage was set to 1.546v after reading quite a bit that it may help a weak stick on the first day, I managed to get a good day of uptime but somewhat poor game performance, (stutters, straight up freezing for 5-10 seconds but no crashes)

I was told to revert the dram v until we come to that road so i did as instructed, the second day resulted in the game CTD repeatedly with the final exit resulting in the entire system bluescreening.

This was behaviour atypical to when i was running both sticks, I would end out a program and it would then bsod, i feel like i need to test some more of stick #2 before i can confirm that was not a one off situation, but it is very clear there is something wrong with stick #2.

I understand corsair has a life time warranty but i no longer have my proof of purchase nor is the place of business open further to provide me a copy of said invoice, I was told i should still attempt to rma my sticks.

I'm thinking i will just go get some new sticks of another brand, rma my gpu aswell to be safe, even if it is not broken, it can't hurt in theory, Then see if i corsair will rma my sticks, if they do, Cool, i might get some nice sticks and can restock the new bought sticks for a small fee.

I still have yet to run prime95, I will return with my results from that.

 

[geofelt]

Open a dialog with Corsair support.
With a lifetime warranty, you probably do not need proof of purchase.
I think you could be a bit more aggressive than 1.546 voltage.
DDR3 ram in higher speeds will run at 1.65v.

 

[swamptree]

Open a dialog with Corsair support.
With a lifetime warranty, you probably do not need proof of purchase.
I think you could be a bit more aggressive than 1.546 voltage.
DDR3 ram in higher speeds will run at 1.65v.

So, further trouble shooting, I rma'd the GPU on monday and the errors are still present, The sticks are clearly the issue but i had to be certain yesterday that it was not the gpu, 30 day rma period was coming up so i needed to make a decision with it.

being said i am getting the same memory errors on both my older hd 7870 and the replaced rx 590, Had i seen the memory errors with my 7870 before i left to RMA, i would of just grabbed some ram aswell but i am on a fixed budget so i wanted to try a fix that cost me just the bus fair.

with stick testing it is clear stick #2 has issues, i am back to both sticks with new gpu currently to see the stability and since install from last night we have had multiple blue screens, even with just idle, seems to be that not even bumping dram voltage seems to help, I will go back to the good stick, test for a few days and see if i can make the blue screen memory issue appear.

I would hate to spend money on new ram and find these errors are still present, but it is clear it is not gpu related after this rma.

Would you recommend one of three ram sets here,

Hyper fury x 1600mhz 10-10-10-30
Rip Jaws X 1600mhz 10-10-10-30
Vengeance Low Profile 1600mh 10-10-10-27

 

[swamptree]

It was the ram, one module was failing, it would pass for the memtesting i would perform but durring performance cause constant bsods, If you are experiencing memory related bsod, check your memory even if it passes testing, If your ram is facing similar issues, Id recommend you try underclocking your dram and to maybe loosen the timings untill you get better stability if you are unable to afford replacement sticks.

 

[geofelt]

If you tested with memtest86, your experience makes me wonder a bit about the value of memtest86.
Unfortunately, I know of no other better tester.
Glad you found the problem.

 

[Madmilldro1]

This happened to me. I started speedrunning this game very frequently, first week or so was fine but very stuttery after a few days, the menu would hang for 1-2 seconds after picking up an item. Have just installed a new 16gb (8x2) kit of corsair vengeance lpx ddr4 too before the game came out as well as a pny gtx 1080 8gb gddr5x GPU. Ram was running at 2400 mhz because I have only a b250m-hdv motherboard. Yesterday I started getting frequent blue screen after just being idle. Fixed that with a fresh install of windows on an ssd. But anytime I'd try to play resident evil 2 remake I'd get a blue screen that said Memory_managment every time. Tried well over 10 times with same error message. I didnt take any sticks out but I went into bios and set it to 2133 instead of 2400 and got through some gameplay. This has never happened, just with re2 remake.