New Build - Refuses to run stable!

Gary Morris · 27 Nov 2017 at 14:15

Hi All,

I've recently built my own rig with a custom WC loop. Specs are:

Intel Core i7-8700K 3.7GHz (Coffee Lake) Socket LGA1151 Processor
Asus ROG Maximus X Hero Intel Z370 (Socket 1151) DDR4 ATX Motherboard
G.Skill Trident Z RGB 32GB (4x8GB) DDR4 PC4-30900C18 3866MHz Quad Channel Kit
Asus GTX 1080 Ti ROG Strix OC Water Cooled Graphics Card
Samsung 960 EVO Polaris 1TB M.2 2280 PCI-e 3.0 x4 NVMe Solid State Drive
Corsair HX850i 850W '80 Plus Platinum Digital Modular Power Supply

I have been plagued with stability issues!!!

Firstly, it refuses to run the XMP profile at 1.35v. In fact, the closest I have got is 3200MHz at 1.5v. I have tested this with a separate set of RAM as well, to the same effect.

Secondly, I cannot get it to sit stable. After around 10 minutes of gaming or Prime95, it will freeze and require a hard reset. Once I hard reset it, it gives a Q Code of 55 (memory not installed) or CC (not a listed Q Code) and refuses to restart. It has to be left for around 30 minutes unplugged for it to start, and it will then only start with a clear CMOS.

I know the RAM is fine as it works fine and tests fine
I have also tested one stick/two sticks/four sticks and I still get Q Code 55 until I leave it for half hour
I don't think its heat related, as SpeedFan shows the CPU no higher than 50c before it crashes

I have read that CPU/Mobo issues can cause these sorts of problems. I have tried loosening/tightening the CPU water block when I had Q Code 55 once, and it did then allow it to start, however it then crashed again 10 minutes later and we're back in the same boat. I have tried reseating the CPU water block as well and loosening the standoffs, to no avail.

My question is this...... How can I get it to run stable for more than 10 minutes?! Do I have a faulty CPU or Mobo? Do I need to RMA them? Has anyone else had similar issues?

From what I've read online, I suspect it's the Mobo but I wanted to get other opinions before I go through the hassle of pulling all the WC stuff out and the Mobo!

Thanks in advance

southernorth · 27 Nov 2017 at 14:26

Gary Morris said:
Hi All,

I've recently built my own rig with a custom WC loop. Specs are:

Intel Core i7-8700K 3.7GHz (Coffee Lake) Socket LGA1151 Processor
Asus ROG Maximus X Hero Intel Z370 (Socket 1151) DDR4 ATX Motherboard
G.Skill Trident Z RGB 32GB (4x8GB) DDR4 PC4-30900C18 3866MHz Quad Channel Kit
Asus GTX 1080 Ti ROG Strix OC Water Cooled Graphics Card
Samsung 960 EVO Polaris 1TB M.2 2280 PCI-e 3.0 x4 NVMe Solid State Drive

I have been plagued with stability issues!!!

Firstly, it refuses to run the XMP profile at 1.35v. In fact, the closest I have got is 3200MHz at 1.5v. I have tested this with a separate set of RAM as well, to the same effect.

Secondly, I cannot get it to sit stable. After around 10 minutes of gaming or Prime95, it will freeze and require a hard reset. Once I hard reset it, it gives a Q Code of 55 (memory not installed) or CC (not a listed Q Code) and refuses to restart. It has to be left for around 30 minutes unplugged for it to start, and it will then only start with a clear CMOS.

I know the RAM is fine as it works fine and tests fine
I have also tested one stick/two sticks/four sticks and I still get Q Code 55 until I leave it for half hour
I don't think its heat related, as SpeedFan shows the CPU no higher than 50c before it crashes

I have read that CPU/Mobo issues can cause these sorts of problems. I have tried loosening/tightening the CPU water block when I had Q Code 55 once, and it did then allow it to start, however it then crashed again 10 minutes later and we're back in the same boat. I have tried reseating the CPU water block as well and loosening the standoffs, to no avail.

My question is this...... How can I get it to run stable for more than 10 minutes?! Do I have a faulty CPU or Mobo? Do I need to RMA them? Has anyone else had similar issues?

From what I've read online, I suspect it's the Mobo but I wanted to get other opinions before I go through the hassle of pulling all the WC stuff out and the Mobo!

Thanks in advance

What's your PSU?

Gary Morris · 27 Nov 2017 at 14:36

I knew I'd missed something!

Corsair HX850i 850W '80 Plus Platinum Digital Modular Power Supply

Liamz51 · 28 Nov 2017 at 21:48

I was having issues with my RAM's XMP profile creating instability. My build is an 8700k on an ASUS Strix Gaming E with 16 GB G Skill Trident 3600 RAM. I updated the motherboard BIOS to the newest non beta version. Worked for me. System's been stable since.

Gary Morris · 30 Nov 2017 at 18:48

Liamz51 said:
I was having issues with my RAM's XMP profile creating instability. My build is an 8700k on an ASUS Strix Gaming E with 16 GB G Skill Trident 3600 RAM. I updated the motherboard BIOS to the newest non beta version. Worked for me. System's been stable since.

Sadly, I've already tried a BIOS and Chipset update

You're not the first to say that though. I'm wondering if it's because this is a brand new processor/chipset/RAM combination, and it'll just take time for BIOS updates to cure this?! I might try getting onto Asus, for what it's worth.

Random thought, and probably a stupid one.... but it's the G SKILL RAM with RGB lighting. Could it be the lighting drawing too much power for it to run stable when I stress it?!

facboy · 30 Nov 2017 at 19:04

Does it work at stock?

Gary Morris · 30 Nov 2017 at 19:34

facboy said:
Does it work at stock?

No. At stock frequency and voltage (1.2v) it does take longer to crash though and is the best it has been. I'm going to try stock values and frequency at a higher voltage shortly to see if that helps.

Interestingly, using the Asus AISuite and looking at the timings page, it shows JEDEC timings and XMP timings. The XMP timing shows the advertised XMP frequency of 3866 and voltage of 1.35v but timings of 18-20-20-39, instead of the advertised 18-19-19-39 at 1.35v. I've tried it as advertised, and it refuses to POST, giving me a Q Code of 55 (memory not installed). I'll try the adjusted timings too, for what it's worth.

It's bizarre because once it crashes, it refuses to re-boot, even using the safe boot and Clear CMOS buttons. I have to leave it for half hour with no power before it'll boot, which almost leads me to think its either something which builds up or it's an overheat, but everything is running cool! I'm completely stumped by the whole thing and it's doing my head in now. I expected a £3500 build to work.........

facboy · 30 Nov 2017 at 21:00

that's stock RAM settings right? what are VCCIO and VCCSA at? do you have MCE enabled?

Gary Morris · 30 Nov 2017 at 21:40

facboy said:
that's stock RAM settings right? what are VCCIO and VCCSA at? do you have MCE enabled?

I've tried it on stock RAM settings as well as various XMP settings. All with VCCIO and VCCSA set to auto.

I just had a play around with the timings, It hates 18-20-20-39, crashes instantly and has to be turned off by holding the power button. It then refuses to POST giving a Q code of "CC" and after a minute "55". I've got it unplugged again for half hour again so I can get back into BIOS.

I forgot to look for MCE so I'll look next time.

facboy · 30 Nov 2017 at 22:21

reset to CMOS defaults and try, leave everything at default (ie no XMP). AFAIK MCE shouldn't be on if it is at defaults but make sure it is disabled, not Auto.

Gary Morris · 30 Nov 2017 at 22:56

facboy said:
reset to CMOS defaults and try, leave everything at default (ie no XMP). AFAIK MCE shouldn't be on if it is at defaults but make sure it is disabled, not Auto.

Will do. I tried but it crashed while I was searching BIOS for the MCE. Now waiting half hour again.

I did notice however that the Mobo is being slightly bent/pressured by one of my rigid WC pipes on the CPU block. I'm going to drain it and re-do that pipe while I'm waiting, just in the vain hope that the pipe is putting pressure on the CPU/Mobo and causing issues there

Mysterae_ · 30 Nov 2017 at 23:17

Rather than waiting half an hour can you not instead switch off the PSU at it's switch (or wall socket if it doesn't have one) then turn the computer on which will very briefly power up and down again. This discharges the motherboard a lot quicker than waiting for it to dissipate.

I used to have a motherboard that refused to warm boot, had to be cold booted from a completely discharged system. Bad caps if I remember rightly.

Gary Morris · 1 Dec 2017 at 00:36

Mysterae_ said:
Rather than waiting half an hour can you not instead switch off the PSU at it's switch (or wall socket if it doesn't have one) then turn the computer on which will very briefly power up and down again. This discharges the motherboard a lot quicker than waiting for it to dissipate.

I used to have a motherboard that refused to warm boot, had to be cold booted from a completely discharged system. Bad caps if I remember rightly.

Brilliant idea, however that doesn't seem to work in this case. It still boots and gives a Q Code of "4C", then "CC", then "55" and halts there. No display whatsoever. This leads me to think it may be a Mobo issue again.

I have now reseated the Mobo, CPU, water block, back plate and RAM. All seated properly, all connectors clean. Tried to start it but it gave me the same Q code error again, possibly where I had been half powering the PSU to fill the WC loop. Now leaving it to discharge again and will see what happens.

Smffy · 1 Dec 2017 at 01:52

I can tell you the processor is capable of higher, mine will run up to 4,366 on an Apex even with the CPU @5.3. Seems like the memory is the issue, take out one stick and try but also:

1. VCCIO and VCCSA at 1.15v
2. Whats the CPU clocked at? If overclocking try cache min/max at 44, LLC 6.
3. Try 1.4v on RAM if needed
4. Try one module at a time in case ones died

Reset BIOS maybe and start from stock everything, then XMP and boot, test, fiddle with more, test. Should load the right profile for the RAM all things going well...

Gary Morris · 4 Dec 2017 at 17:24

Smffy said:
I can tell you the processor is capable of higher, mine will run up to 4,366 on an Apex even with the CPU @5.3. Seems like the memory is the issue, take out one stick and try but also:

1. VCCIO and VCCSA at 1.15v
2. Whats the CPU clocked at? If overclocking try cache min/max at 44, LLC 6.
3. Try 1.4v on RAM if needed
4. Try one module at a time in case ones died

Reset BIOS maybe and start from stock everything, then XMP and boot, test, fiddle with more, test. Should load the right profile for the RAM all things going well...

So we have some progress...

It started being a complete pain in the rear end, refusing to start no matter how much I cleared the CMOS or how long I left it to discharge. I eventually got it working again with just one stick of RAM.

With one stick of RAM and VCCIO and VCCSA at 1.15v, I have got it to POST and sit on the desktop with the posted XMP profile of 3866MHz at 1.35v! The rest of the voltages have been left on auto, and are currently showing as:

DRAM Voltage: 1.35v - Set Manually
PLL Termination Voltage: 1.00v
CPU Standby Voltage: 1.00v
DRAM VTT Voltage: 0.675v
CPU System Agent Voltage: 1.15v - Set Manually
DMI Voltage: 1.1625v
VPPDDR Voltage: 2.5v
Core PLL Voltage: 1.25v
CPU VCCIO Voltage: 1.15v - Set Manually
PCH Core Voltage: 1.00v

Is there anything on there that should be adjusted? The CPU is currently running as standard, with auto core frequency and Turbo mode on.

I'm about to try each module on this XMP profile to make sure all four handle in individually, then I'll try a pair, then the quad. I'm also going to stress test each module individually, as a pair and as a quad.

One thing I have noticed though, the higher I pushed the XMP profile towards its advertised speed of 3866, the longer it was taking to get through the DRAM self testing. The Q codes are noticeably slower before it moves onto GPU testing. It's only an extra couple of seconds so I'm not overly bothered, but not sure if it gives you a clue to anything?

Gary Morris · 4 Dec 2017 at 18:44

Okay, so I have tested all four modules on their own on Prime95 with the XMP profile on those settings and they all test fine.

As soon as I try and run a pair or quad, it fails to POST and returns to the Q Code of CC. I can only get it to boot on a single module on those voltages on that XMP profile.

marco.da.man · 4 Dec 2017 at 21:13

im sure one of the intel guys can come up with a solution for you soon.

just out of interest have you tried single ram in different slots to to eliminate the process ? was it booting up before you tried overclocking ? it could be possible you have a faulty mobo with it not playing ball.

i hope you get sorted as i had similar problem with my board in my ryzen started fine, then started have issues where not booting up etc

Gary Morris · 4 Dec 2017 at 21:25

marco.da.man said:
im sure one of the intel guys can come up with a solution for you soon.

just out of interest have you tried single ram in different slots to to eliminate the process ? was it booting up before you tried overclocking ? it could be possible you have a faulty mobo with it not playing ball.

i hope you get sorted as i had similar problem with my board in my ryzen started fine, then started have issues where not booting up etc

If I'm using a single stick, it will only boot in the B2 slot. Any of the other three and it doesn't detect the RAM. It does say in the manual though that when running a single module you have to use B2, so I'm not too surprised at that.

It was originally loading four slots fine on all standard settings, and detecting the full 32GB RAM, so unless it's decided to barbeque something when I've been testing it, all four slots should work fine. It was just damn unstable, and refused to XMP whatsoever on stock settings.

marco.da.man · 4 Dec 2017 at 21:33

Gary Morris said:
If I'm using a single stick, it will only boot in the B2 slot. Any of the other three and it doesn't detect the RAM. It does say in the manual though that when running a single module you have to use B2, so I'm not too surprised at that.

It was originally loading four slots fine on all standard settings, and detecting the full 32GB RAM, so unless it's decided to barbeque something when I've been testing it, all four slots should work fine. It was just damn unstable, and refused to XMP whatsoever on stock settings.

i would contemplate the fact that the mobo could be faulty and hope you havent cooked something. hope you get sorted and all the best with it

Gary Morris · 5 Dec 2017 at 00:54

I've been doing more testing and discovered I can run the XMP profile fine on single channel RAM. I've now got an 8GB stick in B1 and an 8GB in B2 running perfectly well. As soon as I add a stick to A1 or A2 to make the dual channel, it refuses to boot and gives the Q code CC again. Any ideas?