Rampage V Extreme Power Cycling and other Boot Issues

[CaptainPhaser]

When booting my system, it usually power cycles multiple times until one of the following things occur:

• Successful Boot with Compatibility Support Module (CSM) Disabled
  
• No VGA Detected Beep Code (1 Long, 3 Short) with CSM Disabled [Occurs the most]
  
• No VGA Detected Beep Code with CSM Enabled
  
• Successful Boot with CSM reverted to enabled by motherboard automatically
  
• Overclocking Failed screen despite no overclocking
  
• Overclocking Failed screen with 3/4 DIMMs detected with no overclocking [Occurs the least]
  
• Power Cycling typically goes through the following boot codes [CSM setting DISABLED]:
  
  
  
  • 70 PCH DXE initialization started
    
  • 79 CSM initialization
    
  • 91 Driver connecting is started
    
  • 95 PCI Bus Request Resources [Repeats till boot or VGA not detected]
    
  
  

I've tried the following to fix the issue:

• Re-seated GPU [GPU is in suggested x16 slot]
  
• Tested GPU in all PCI Express Slots
  
• Tested 2nd GPU in all PCI Express Slots with CSM enabled and disabled
  
• Updated GPU BIOS
  
• Changed PCI Express Power Cabling
  
• Re-seated all cabling on motherboard and PSU
  
• Tested PSU with multi-meter to verify voltages [All OK]
  
• Re-seated RAM
  
• Tested RAM with memTest [All Tests Passed]
  
• Cleared CMOS
  
• Replaced CMOS Battery
  
• Updated Motherboard BIOS
  
• Tested outlet to verify voltage with multi-meter
  

Specs:

• 
  
• OS: Windows 10 Enterprise 64-bit Version 1709 Build 16299.461
  
• CPU: Intel Core i7 5930K @ 3.5GHz [STOCK]
  
• MB: Asus Rampage V Extreme [Socket 2011-v3] [Intel X99 Chipset] [BIOS 3902]
  
• PSU: Seasonic X-1250 (SS-1250XM2) 1250W 80+ Gold
  
• RAM: G.SKILL Ripjaws 4 16GB (4x4) DDR4 2666MHz C15 1.2V [Quad Channel] [Running at 2133MHz with XMP DISABLED for troubleshooting]
  
• GPU: GTX 1080 Ti [Gigabyte Aorus Xtreme Edition GV-N108TAORUS X-11GD]
  
• Boot SSD: Samsung 850 EVO 1TB
  
• Storage HDD: Seagate Barracuda 3TB 7200RPM [ST3000DM001]
  
• CPU Cooling: Corsair H115i 280mm AIO
  
• Keyboard: Corsair K95 Platinum
  
• Mouse: Logitech G502
  
• Monitor: Acer Predator XB271HU [Connected w/ DP]
  

So far I have always been able to boot eventually, but it can take multiple power cycles and re-seating the GPU or completely powering off and disconnecting the PSU power before I can get a successful boot. Sometimes I get a nice, fast, normal boot, but mostly it's the above issues. I feel like the issue is likely the motherboard, but it is out of warranty and I'm trying to avoid replacing it if possible.

Here's a video of 4 total cycles until a successful boot showing boot codes:


I also get distorted boot screens from time to time right before the windows login. The POST screens are not distorted prior to the Windows loading screen. I've experienced this with both GPUs tested.



I doubt that there's anything I can do on my end to fix this, but I am hopeful. If nothing can be done I am considering an out of warranty repair, but I have no idea of what that might end up costing or if its possible.

[CaptainPhaser]

Arne Saknussemm wrote:
Not sure about the gunk on there...but I'd clean it off with some alcohol...

I'd look at your cooler mounting...and test your RAM stick by stick....corruption on fresh OS install means something is definitely not working somewhere

I'd be tempted to strip things down clean it and set up on the MB box a minimal install. Make sure the OS image you are installing is freshly downloaded...

I've already remounted my cooler and applied new thermal paste, made sure to tighten as evenly as possible. Also, prior to my clean install I was on Windows 10 Build 1709, I then downloaded a new iso direct from microsoft and even used a different flash drive with build 1803. Same corruption issue. HOWEVER, something interesting has happened.

The CSM, when it would be enabled, the text that displayed "Please press DEL or F2 to enter UEFI BIOS setting" appeared larger and lower resolution due to legacy BIOS restrictions, but after testing between my 1080Ti and an old 9800GT today now even if I enable CSM and set to LEGACY ONLY (Except Boot Drive), it is detecting my 1080Ti as UEFI and I get native resolution text during POST. Prior to this I would always get the legacy sized text with my 1080Ti whenever CSM was enabled, or turned back on automatically.

At first I thought perhaps the CSM was corrupted somehow, but it works fine between my 1080Ti and my 9800GT, displaying native text for the 1080Ti, and larger, lower resolution text for the 9800GT both while the CSM is ENABLED and all settings except the boot drive set to LEGACY ONLY. Is this normal behavior? I've cleared the CMOS multiple times over the past week or so so and have kept my settings consistent, so nothing has really changed there. Since this happened I haven't had a single boot issue (yet).

Also, both Windows installations were done the same way with the same UEFI settings (CSM disabled, Fast Boot Enabled, Secure Boot Enabled) so nothing changed there either.

Essentially all I did was swap my 1080Ti for a 9800GT, enable the CSM, and then I switched back to my 1080Ti and booted both with CSM enabled and disabled and get the same native resolution text either way and somehow no boot issues as of right now.

EDIT: Just got a 4 power cycle boot with my 1080Ti, but the CSM enabled in Legacy Only mode is still showing native text. I'm just lost as to what is actually going on. 😞

[neyla12]

Thanks for posting.

[Arne_Saknussemm]

I would suspect the RAM personally (RAM IMC on CPU)...the corruption, the multiple boots... probably RAM training...all the CSM stuff I would regard as a red herring...but I may be wrong...

[CaptainPhaser]

Arne Saknussemm wrote:
I would suspect the RAM personally (RAM IMC on CPU)...the corruption, the multiple boots... probably RAM training...all the CSM stuff I would regard as a red herring...but I may be wrong...

Just went through another battery of tests

• 1 Pass memTest86 with all DIMMs: No Errors
  
• 1 Pass memTest86 per DIMM each in a different slot: No Errors
  
• 8 Hour Asus RealBench Stress Test: No Errors
  

All of my components seem to be functioning perfectly fine after boot. I have yet to have any sort of test fail on my current configuration. I haven't touched any voltages apart from RAM going to 1.35v as I'm trying to stay at stock. I could start making tiny adjustments to System Agent which sits at around 0.845v and VCCIO which is around 1.05v. I think it's either the motherboard, or I have a DIMM that is just on the edge of stable that I would have to test quite a bit before I can actually pick up an error.

Any recommendations on testing methodologies would be appreciated.

[Rob_W_]

I would agree with Arne, *I had all the above issues trying to get my ram to run manually ( different ram I know) but multi boots until setting vccio and vccsa 1.21v / 1.175v withram at 1.35v*
Worth trying to get boot stable, ram can cause errors on boot that make you think it’s cpu, gpu, mobo or all of them and just cause you to tear your hair out,
You seem to have tried most other things.
Some of the errors I got was ( mem not detected/ test nvram/ test cpu/ gpu not detected/ hdd detect/ load gpu bios ) that was just some.*

[Arne_Saknussemm]

It's my understanding that memTest86 is not particularly useful on that or even later platforms. People tend to rely more on HCI memtest (free just make sure you launch enough instances to cover memory) RamTest (paid but much faster testing)

I'd try to raise SA voltage a bit...say 1.0v or a touch more...see if that stops boot loops...

Something is not right if you are corrupting fresh OS install so obviously passing some tests is to be taken with a pinch of salt for the moment...if something is borderline it will take quite a lot of testing and patience....

[CaptainPhaser]

Arne Saknussemm wrote:
It's my understanding that memTest86 is not particularly useful on that or even later platforms. People tend to rely more on HCI memtest (free just make sure you launch enough instances to cover memory) RamTest (paid but much faster testing)

I'd try to raise SA voltage a bit...say 1.0v or a touch more...see if that stops boot loops...

Something is not right if you are corrupting fresh OS install so obviously passing some tests is to be taken with a pinch of salt for the moment...if something is borderline it will take quite a lot of testing and patience....

Ran to 500% coverage with HCI memtest. 0 errors. Upped SA to 1.0V, still had the same 4 power cycles to boot. Will try at 1.05 and see if anything changes.

[Arne_Saknussemm]

Just so we are on the same page....how are you running HCI...one instance per thread and 90% memory?

[CaptainPhaser]

Arne Saknussemm wrote:
Just so we are on the same page....how are you running HCI...one instance per thread and 90% memory?

The first time I ran it with 6 instances (woops) 90% memory, just reran with 12 instances, 90% memory, 500% coverage. 0 Errors.