[Need help] How likely to get a bad CPU?

[Parn]

I've been battling with random BSOD/Freezes on my new build for the last 8 days. Wondering if anyone here could share some of your inputs.

New build spec:

• Core i5 6600K
• GA-Z170X-Gaming5-EU (now changed to Maximus VIII Hero)
• 16GB (2x8GB) Corsair Vengeance LPX 2666 C16
• Palit GTX980 SuperJetstream (from the last build, bought in Jan 2015)
• 256GB Samsung 850 Pro SSD (from the last build, secure erased, bought in Jan 2015)
• 1TB WD Black (from the last build, for temp/download data only, 2 years old)
• LG BluRay drive (from the last build, bought in Jan 2015)
  
• SB ZxR (from the last build, only 3 months old, was to replace my aging Audigy2)
• SeaSonic X-750 PSU (from the last build, about 2 years old)

I initially assembled the system using the Gigabyte motherboard listed above. Upon the first power-up, I updated the BIOS to version F4 (factory F2, but version F3 apparently fixes a serious error, so I decided to update past F3 without second thought) with no error. Then I loaded optimised defaults, disabled the onboard igfx and sound, enabled XMP and enabled secure boot while leaving the rest to their default values before installing Windows 10 Pro from my USB drive (Sandisk Extreme 64GB). I always install everything with CPU at stock settings and make sure the system is stable before attempting overclocking. After the installation had been done, I got a BSOD (Page fault in non paged area) on the first boot up of W10. This immediately alerted me of memory error, so I switched off XMP and started to run memtest86 (parallel cpu mode, all tests). I let it run overnight and there was no error found the next morning.

I thought maybe my Windows 10 Pro USB media was somehow corrupted, so I re-formatted it and made another one. This time I installed W10 with memory at default 2133 speed. I got BSOD again after two restarts (IRQL Not less or equal). This made me think my W10 ISO could be the culprit so I decided to try my W8.1 DVD. Again BSOD on first boot up. I thought my W8.1 DVD might be bad, so I fetched my original W7 DVD which I had just used to reinstall my work PC a couple of days ago with no problem. I got BSOD during installation (ntfs.sys error). My first reaction to this was my BluRay drive could be broken. I borrowed the working DVD-RW drive from my work PC to try. But still got BSOD on first boot up of W7. This pretty much ruled out installation media being the problem.

The next thing I tried was to rule out SSD/HDD problem. So I unplugged my WD Black drive and did another W10 installation. BSOD after about two reboots. I thought maybe the 850 Pro was the culprit. Fortunately I have a spare Crucial MX100 in hand, so I swapped the SSD. Same symptoms again. I then used a Seagate 500GB HDD to attempt another installation. The result was no different from SSD. I even swapped SATA cables with some spares I had, but those made no difference.

I did some search online and read memtest86 is not infallible. This made me think the Corsair RAM might indeed be defective. So I bought a new kit (the same model). I ran memtest86 on the new kit for about 4 hours with no error before another installation of W10. Guess what? No difference. I then stripped the system to bare minimum (thought it could be the graphics card or sound card even though they were working perfectly well on the last build) and tried again. The system froze on first boot up (not BSOD this time) and I had to hit the reset button. Then the system was seemingly fine for a few restarts until another BSOD hit (ntfs.sys error).

At this point I thought it might be the PSU (although I highly doubted the SeaSonic X-750 would fail with such light load after being rock solid for the last 2 years). So I swapped it with the SeaSonic Platinum 660 from my work PC. The result was just as I believed, NOT the fault of the PSU.

Now this left the motherboard or the CPU being responsible for the problem. I read on various forums that BIOSes on Gigabyte Z170 boards are quite flaky and a lot of users have been experiencing compatibility issues with different brands of RAM (mainly Corsair & Kingston). I did more search and found out that Asus Maximus VIII Hero seemed to come with more stable BIOS and not suffer from this RAM compatibility issue. So I RMA'ed the Gigabyte and replaced it with the Asus board.

Initially I did not get any freezes/bsod on the first few boot ups. So I thought the problem was down to Gigabyte poor BIOS and had been solved. I finished installing all the required drivers and to my surprise random freezes started again. I checked the system log and only found one error related to event ID 56 (which is a common error seen by a lot of other Skylake users online but most of those users are not getting bsod because of it). I thought the freezes could be caused by a broken driver (a few people on geforce.com claim that 358.50 is unstable). I downloaded a fresh copy of 355.98 on my work PC and decided to start over from scratch. However as soon as I re-installed W10 fresh, BSOD came back (this time the error was KMODE Exception). This was before any driver had been installed. I updated to BIOS to 0902 (which was the most stable version claimed by most users online) but that made no difference.

So far the only part which had not been replaced or tested was the CPU. I did another installation of W10 and managed to get into Windows on second boot up. I then ran Prime95 in Blend mode for about 6 hours to test out the CPU. No error was found.

Now I'm a little puzzled. Is it possible for a defective CPU to pass Prime95? From all the CPUs I've bought since Pentium MMX 200, I've only had one bad unit which was an Athlon XP 3200+ (it had a chipped corner. I could install WinXP with it fine but would get system crash during gaming). These days CPUs are covered by heat spreaders so I can't physically inspect the die (I'm not an extreme overclocker, so I don't do things like de-lidding). How likely have I got another bad CPU?

By the way all the parts carried over from my last build were working perfectly well before. My last build was based on Z97 which is now my work PC. My previous work PC was a 1st gen Core i5 and has been given to my parents as a Internet/Email box.

Please share your inputs to help me to resolve my problem. Many thanks.

 

[BarbaricSoul]

the only thing I can think of you missed(or at least didn't mention doing) has to do with your RAM. Did you try running each stick of RAM separately? I know you completely replaced the RAM, but getting a bad stick or two of RAM is much more common than a bad CPU.


Guy is this thread ( http://www.techpowerup.com/forums/t...-ddr4-issues-with-msi-z107a-gaming-m7.215192/ ), although using a MSI motherboard, seems to had been experiencing similar issues as you. His issue turned out to be a slightly bent pin in the CPU socket. Get a magnifying glass and check your socket pins.

 

[Parn]

the only thing I can think of you missed(or at least didn't mention doing) has to do with your RAM. Did you try running each stick of RAM separately? I know you completely replaced the RAM, but getting a bad stick or two of RAM is much more common than a bad CPU.


Guy is this thread ( http://www.techpowerup.com/forums/t...-ddr4-issues-with-msi-z107a-gaming-m7.215192/ ), although using a MSI motherboard, seems to had been experiencing similar issues as you. His issue turned out to be a slightly bent pin in the CPU socket. Get a magnifying glass and check your socket pins.

Thanks for your advice.

Sorry I didn't mention the RAM testing clearly. I did test each stick of RAM individually on the Gigabyte board, but not the Asus. I will do that on the Asus just to be sure.

There was no bent pin in the cpu socket. I had to take a picture of the cpu socket and email it to the retailer before RMAing my Gigabyte board (it's part of the return procedure). If there were any bent pins, my return wouldn't have been accepted.

 

[alucasa]

In my whole PC life which is about 20ish years, I've never gotten a bad CPU.

So, in any issues with computers or even servers, CPU being bad is almost nil. It has always been something else.

 

[Yukikaze]

Bad CPUs are basically unheard of. They live very well through (moderate) abuse and are very reliably sorted at manufacturing. Motherboards, memory and storage are a lot more likely to fail.

 

[xorbe]

Bad CPUs happen, but it's pretty rare. Check socket, check PSU age (that it can handle low CPU power mode when on), check another ram kit.

 

[cadaveca]

Bad CPUs are basically unheard of. They live very well through (moderate) abuse and are very reliably sorted at manufacturing. Motherboards, memory and storage are a lot more likely to fail.

Welcome to Intel's 14nm. Seems things at this process node aren't as easy to keep perfect. Don't forget the "shortage" of CPUs at launch.

I wouldn't think CPU is the culprit either, but based on the reported testing, DRAM, CPU, board or PSU are at fault. Likely this is just a driver issue, or maybe the sound card. The only thing to do is try single stick of ram, and new installation media. Limit the number of parts used as much as possible.

 

[ne6togadno]

Welcome to Intel's 14nm. Seems things at this process node aren't as easy to keep perfect. Don't forget the "shortage" of CPUs at launch.

I wouldn't think CPU is the culprit either, but based on the reported testing, DRAM, CPU, board or PSU are at fault. Likely this is just a driver issue, or maybe the sound card. The only thing to do is try single stick of ram, and new installation media. Limit the number of parts used as much as possible.

and then add them one by one (2nd stick ram, vga, hdd, ssd etc.) to see which one is causing the problem

 

[jsfitz54]

You might try the Intel Diagnostic Stress Test: https://downloadcenter.intel.com/download/19792

 

[Parn]

You might try the Intel Diagnostic Stress Test: https://downloadcenter.intel.com/download/19792

Thanks for reminding me about this tool. I completely forgot about it.

Downloaded it on my work PC and transferred to the problematic one with a USB key. I managed to get into Windows after a few BSODs. I just ran the test and it failed CPU Stress Test.

Looks like it is a bad CPU although I'm surprised that it managed to pass Prime95 Blend for 6 hours. I'm going to RMA it and will be back with an update once I receive the replacement.

 

[OneMoar]

post the frigging BSOD code ffs
and the minidump ///
you don't have a bad cpu ... that doesn't happen ever ...

 

[Parn]

Update:

Got my replacement CPU today. The retailer do not have 6600K in stock. In fact a lot of the online retailers in the UK don't have any 6600K in stock at the moment. So I got a vanilla 6600 instead (£40 cheaper).

Just finished installing everything and W10 Pro installed nicely along with the rest of the drivers and programs. 0 crash/freeze/BSOD so far

Conclusion: That 6600K was indeed BAD. And Prime95 alone cannot be relied on for CPU testing.

I guess with Intel having so much manufacturing difficulties using 14nm QA for CPUs is no longer as bullet-proof as it used to be. Usually CPU is one of those components you would least expect to fail when it's brand new. That's why I wasted days testing/replacing other components. I suppose I will have to pay more attention to CPU testing in any future builds rather than assuming they are all good.

 

[xorbe]

Yeah, we've known for a long time that Prime95 alone is not sufficient any longer. The dynamic cpu (and gpu) clocking is a pain for stability testing. Congrats on getting it wrangled out!

Also, buy a lottery ticket. Bad cpus are very, very rare. It's been a long time since the last bad cpu thread.

 

[cadaveca]

Also, buy a lottery ticket. Bad cpus are very, very rare.

I would not say very very rare. Uncommon, sure, but not rare, especially in recent years.

 

[Rockhill79]

Owner of faulty i5 6600K reporting from Finland.

New build:
Asus Z170 Pro Gaming
i5 6600k
Kingston Fury HyperX 2 x 4 GB
Seasonic 650W PSU
Samsung 250GB SSD-drive
Asus Geforce GTX-970

I make this short: almost the same story/did the same tests as the OP except i got BSOD's "machine check exception" and "clock watchdog timeout". Finally i bought new i5 6600k processor, installed it and voilá; no more BSOD's.
So the faulty one is going back to retailer tomorrow.

I didn't overclock the old one, it actually worked for one day after build (i played games, surfed in the internet etc.) and the next day bluescreens started to happen.

I was extremely careful when installing cpu in the socket, so i don't think i would have damaged it somehow.

 

[Pill Monster]

Watchdog Timeout is generated by the CPU scheduler, often the result of hung core. Similar to a TDR.

 

[Parn]

Owner of faulty i5 6600K reporting from Finland.

New build:
Asus Z170 Pro Gaming
i5 6600k
Kingston Fury HyperX 2 x 4 GB
Seasonic 650W PSU
Samsung 250GB SSD-drive
Asus Geforce GTX-970

I make this short: almost the same story/did the same tests as the OP except i got BSOD's "machine check exception" and "clock watchdog timeout". Finally i bought new i5 6600k processor, installed it and voilá; no more BSOD's.
So the faulty one is going back to retailer tomorrow.

I didn't overclock the old one, it actually worked for one day after build (i played games, surfed in the internet etc.) and the next day bluescreens started to happen.

I was extremely careful when installing cpu in the socket, so i don't think i would have damaged it somehow.

Wow, another instance of faulty 6600K. Sorry to hear that. I guess Intel is really having a difficult time with 14nm fab.

I wish you luck with your RMA. I'm still in discussion with my retailer about the RMA (I got my temporary vanilla 6600 and now another 6600K from a different retailer). They received the processor more than 3 weeks ago and never got back to me until I chased it up 3 days ago. They then did a very basic test by popping it into a test box with Windows already installed and fired it up to check if the system could get into Windows successfully. As I said in my 1st post the BSOD/Freeze was random and sometimes the system could boot into Windows and remain on the desktop as long as I didn't launch any applications. If I did start to use the PC, constant crash of explorer.exe would kick in after a few minutes which would eventually lead to a complete lock-up/freeze. Their engineer managed to get it boot into Windows, so they came back to me saying it's not faulty with no further testing. I just emailed them back explaining the same story again yesterday evening and am now waiting for their reply.

 

[Rockhill79]

I wish you luck with your RMA. I'm still in discussion with my retailer about the RMA (I got my temporary vanilla 6600 and now another 6600K from a different retailer). They received the processor more than 3 weeks ago and never got back to me until I chased it up 3 days ago. They then did a very basic test by popping it into a test box with Windows already installed and fired it up to check if the system could get into Windows successfully. As I said in my 1st post the BSOD/Freeze was random and sometimes the system could boot into Windows and remain on the desktop as long as I didn't launch any applications. If I did start to use the PC, constant crash of explorer.exe would kick in after a few minutes which would eventually lead to a complete lock-up/freeze. Their engineer managed to get it boot into Windows, so they came back to me saying it's not faulty with no further testing. I just emailed them back explaining the same story again yesterday evening and am now waiting for their reply.

Hi Parn,

sorry to hear your retailer is an a**hole and i have a feeling that so is my retailer too, we'll see. There's a saying: ""If it looks like a duck, swims like a duck, and quacks like a duck, then it's probably a duck".
So excluding all the other components remains the cpu. If all works with the new cpu then the old one must be faulty.

Maybe the best option is to send cpu straight back to Intel because retailers don't have a time or interest.

I got freezes on desktop although i didn't execute any program, sometimes computer froze right after Windows startup or even at the startup.

 

[Parn]

Hi Parn,

sorry to hear your retailer is an a**hole and i have a feeling that so is my retailer too, we'll see. There's a saying: ""If it looks like a duck, swims like a duck, and quacks like a duck, then it's probably a duck".
So excluding all the other components remains the cpu. If all works with the new cpu then the old one must be faulty.

Maybe the best option is to send cpu straight back to Intel because retailers don't have a time or interest.

I got freezes on desktop although i didn't execute any program, sometimes computer froze right after Windows startup or even at the startup.

You're absolutely right.

Now I wish I had sent the faulty CPU back to Intel directly instead of this rubbish retailer. I chose to deal with the retailer because I wanted a full refund after I had got a replacement from a different supplier. If I sent it straight back to Intel I would not get a refund but a replacement weeks later.

However the retailer is going to charge me a test fee if the engineer deem the processor not faulty after some basic tests. In this case I'd rather wait for weeks to get a replacement and sell it on Ebay (probably lower than the price I paid).

 

[GreiverBlade]

interesting, well i am lucky with my 6600K (or unlucky) because my previous issue was a 4x4 kit that was not compatible in 4x4 but only in 2x4 (now i have a 4x4 2800 C14)
and we have almost the same setup, except that you did a sidegrade from a GA-Z170X-Gaming 5 EU to a Maximus VIII Hero, i like both brand i hope you paid less for the MVIIIR than the Gaming 5 ... because for me the MVIIIR was more expensive and added nothing to tip the scale in her favor and make the Gaming 5 looks bad (i had a Maximus VII Ranger and the BIOS of my actual Gaming 5 is totally fine )

the ram compatibility is not technically gigabyte's fault ... there is a compatibility list (QVL) it's not for nothing (actually my retailer did choose the wrong RAM, hence the "he did swap the 2 kits without question and cost added" ) and iirc some X99 DDR4 compatible kit are not Z170 compatible and vice versa (which imho is completely strange and limit stupid but, who know ... maybe the reason behind that situation is something more ... deep")


not answering the thread only for that (nor it is a issue)... but rather ... i have a AMD Duron 700 with a chipped die (well ... chipped, more "one side of the die ripped off" tho it was my fault, heatsink too tight ) ... still working tho and no gaming issue ... (chipped die is the user fault ... not a faulty unit, unless it has been damaged during the manufacture or the packaging, which would be unlikely since i don't think anything with a damaged die would pass the control )


i hope your retailer will not cause to much problems ... i had several a**hole too

 

[Rockhill79]

Well, i just received an email from my retailer saying that they will replace the processor with the new one

 

[Parn]

Well, i just received an email from my retailer saying that they will replace the processor with the new one

Lucky you.

My retailer refused to do any further testing and their after sale manager told me that their test system successfully powered up with the CPU so it must be good. Sigh...The only thing they could do was to waive off the testing fee. Anyway they (box.co.uk) won't get anymore business from me.

I've been in touch with Intel UK support and they have acknowledged my RMA request after I explained everything to them. A collection for the faulty CPU has been arranged and the new replacement is on its way now.

 

[francis511]

Wow a Confirmed Bad cpu. Literally don`t remember the last time I heard that one ! Just as well you figured it out by by yourself. I would have totally Proved there`s no such thing !!!!

 

[CAPSLOCKSTUCK]

post the frigging BSOD code ffs
and the minidump ///
you don't have a bad cpu ... that doesn't happen ever ...

Need i say Moar

 

[Mathias Enggård]

There it is.. Another faulty 6600k owner.
Got the "machine check exception" repeatedly swapping everything but my Hero VIII and the CPU as i had no skylake MB or CPU to swap.
Flashed bios to the latest, which resulted in differing between "machine check exception" and "Clock watchdog timeout" and pulsating rgb lighting on the MB, yay.
I tried disabling quadcore, ie. running only a single core. And then, success! First complete boot for 4 days of searching. This effectively ruled the CPU the culprit and my retailer agreed at first, but obviously has some testing to do to determine if I'm at fault.