New CPU & GPU install, Sporadic POST and BOOT

Gutzy

Caliper Novice
Original poster
Mar 11, 2018
31
14
Hi all,

I figured id post this here as I won an S4M and figured some of you may know how I should tackle this issue.

My Specs...

MoBo: Gigabyte GA-B350N Gaming Wifi
CPU: Ryzen 2600X 95w TDP & 2200G 65w TDP
GPU: Powercolor Vega 56 Nano 1x8pin+1x6pin PCIE connector and RX580 ITX 1x6pin PCIE connector
RAM: G-Skill DDR4 Samsung B-Die 3200MHZ 2x8GB
Drive: Samsung 960 EVO NVME PCIE SSD
PSU: HDPLex 400W DC-ATX + Official Dell 330W AC adapter
Monitor: ASUS MG24QR 144Hz Freesync monitor
Keybaord: Logitech G413 Mechanical LED Keyboard
Mouse: Logitech G502 RGB gaming mouse

Previously was running my RX580 and 2200G with no issues, then decided to upgrade to 2600X and Vega Nano.

  • Utilized DDU to remove Rx580 drivers and shutdown system to install New GPU, and the CPU.
  • Turned on system and it took a while to POST, seemingly surging as it did (fans would spin, slowdown slightly, spin normally again, slowdown etc)
  • System finally booted up and I proceeded to reinstall chip-set drivers and Vega Nano GPU drivers
  • Proceeded to reboot the system to wrap up driver install and the system seemed to be shutting down normally with the screen stating "restarting" and the screen going blank but the power button on the S4M stayed lit.
  • After giving it a few minutes, i decided to hold the power button to manually shutdown, and it would not shutdown!
  • Had to pull the plug to manually shut system down moving forward
  • After plugging system back in and powering on the system, it would either complete POST but nothing would show up on screen or heard from speakers, or not complete POST at all ( system beeps once when POST completes, not at all when it doesn't)
  • Resetting CMOS gave similar results, so proceeded to disconnect all peripherals except for Display port cable and of course the power.
  • After doing this the system would Boot to windows more consistently but ofcourse I was unable to access BIOS without keyboard connected.
  • Once i was in Windows I immediately hooked up mouse and keyboard and everything was fine.
  • No issues in device manager, and VEga Nano and 2600X were seen by system just fine.
  • Proceeded to reduce voltage on Vega Nano by -50% and chose the "Power Saver" profile within windows.
  • Spun up Overwatch and played that expecting the system to shut off due to power demands but it played just fine.
  • Thought maybe this was all a fluke and proceeded to restart system to attempt accessing the BIOS, again power button would not shut off and I had to manually pull cord.
  • And its been a merry go round since, where sometimes it will POST/Boot up if i disconnect all peripherals before Powering on system and then works fine once I'm in windows.
Diagnosing steps i took:
  1. BIOS up to date
  2. Reset CMOS
  3. Removed residual power from board
  4. When able to access, reset BIOS to Defaults
  5. Enabled/disabled memory xmp profile in BIOS
  6. Disabled XFR2 and any CPU boost options within BIOS
  7. Reinstalled BIOS
  8. NO bent pints on CPU

I assumed it was probably the Vega Nano requiring too much power so swapped that out for my RX-580 ITX and the issue persisted.

I swapped in my previous processor, the 2200G and the everything worked fine

I then swapped in my Vega Nano to pair with my 2200G and everything works fine too in that configuration.

Its only when I add the 2600X that it gives me issues.

So I assumed it must be a power issue due to the peripherals requiring power at startup and removing them makes it easier on the PSU to supply required power.

My concern is, usually during power-up the system isnt using a lot of power am I right? So the Dell brick should be able to get it to boot up and I would expect random shutdowns while actually doing something that demands power like gaming.

My concern is that this CPU might be bad ( which i assume to be unlikely as CPU's are pretty robust from my experience)

So is this the PSU or the CPU?
If it is the PSU shouldn't it provide enough power to at least get the system to boot into windows?


I have a Kill-A-Watt monitor and Multi-meter if any of you would like me to test some things out.

Thanks in advance.
 

confusis

John Morrison. Founder and Team Leader of SFF.N
SFF Network
SFF Workshop
SFFn Staff
Jun 19, 2015
4,129
7,057
sff.network
Moved to Hardware from NFC vendor subforum due to nature of post - chances are the case isn't at fault here!
 

TheHig

King of Cable Management
Oct 13, 2016
951
1,171
Ugh. I feel for you. The joys of home building right!?

Few thoughts.

Breadboard that thing. Pull everything out of the case and test on a non conductive surface. I use a piece of cardboard personally. Atx mobo boxes work well too.

Also you could of try booting off of a regular atx /sfx psu of 500w or more if you have access to one. It’s an easy way to see if your 2600x and Vega are under powered by the Hdplex/ dell brick.

Could it be a short somewhere? It’s very possible to have things shift around when swapping parts. I have personally had this happen. Breadboarding is good for eliminating that.

Good luck!
 
  • Like
Reactions: Obelisk

Gutzy

Caliper Novice
Original poster
Mar 11, 2018
31
14
Ugh. I feel for you. The joys of home building right!?

Few thoughts.

Breadboard that thing. Pull everything out of the case and test on a non conductive surface. I use a piece of cardboard personally. Atx mobo boxes work well too.

Also you could of try booting off of a regular atx /sfx psu of 500w or more if you have access to one. It’s an easy way to see if your 2600x and Vega are under powered by the Hdplex/ dell brick.

Could it be a short somewhere? It’s very possible to have things shift around when swapping parts. I have personally had this happen. Breadboarding is good for eliminating that.

Good luck!

Thanks @TheHig will try this tonight. Unfortunately I wont have access to a standard ATX/SFX anytime soon, might just purchase one as a diagnosing tool to be honest.

When I initially put the system together with all the parts I performed it on a wooden table for the same reasons you mention ( non conductive ) but of course the case is conductive itself and I did have everything still mounted to the case during install and eventual power-up.

Thanks and ill follow up.
 
  • Like
Reactions: TheHig

Gutzy

Caliper Novice
Original poster
Mar 11, 2018
31
14
After many hours breadboarding the system, and trying every option under the sun in BIOS (even an attempt at disabling cores lol) The resolution came in the form of an undervolt and underclock.

Undervolting from the stock 1.225 to 1.115 and underclocking to 2.5GHZ resolved my issues.
I guess it was a PSU issue, but for the life of me I don't understand as the wattage from my kill-a-watt never got anywhere near 330W during the many failed attempts to POST or BOOT.

I may attempt to push the clocks higher to match at least an R5 2600 but for now I'm just glad it works and my PC doesn't seem sluggish at the moment with clocks at 2.5ghz

If anyone has any explanation as to why an undervolt and underclosk would resolve my issues despite the system never pulling anything near 330W during POST/BOOT even under the factory specs of the 2600X please chime in.

For some performance specs under current working CPU underclock, DOOM set to Vulkan, and high settings across the board, power from the wall came just under 330W at 175FPS.

Capping frames to 60FPS in Wattman for DOOm specifically brings wattage from the wall to around 160-175W total

Card has no problems staying cool even when I had it at 175FPS in Doom at high settings and I attribute that to @Josh | NFC 's Sky-4M design and a fan he highly recommended (Scythe Slipstream 120MM slim fan), which I have mounted on the sky bracket to the the backside of the Vega Nano.

Big thanks to @TheHig for the help

 

jtd871

SFF Guru
Jun 22, 2015
1,166
851
After many hours breadboarding the system, and trying every option under the sun in BIOS (even an attempt at disabling cores lol) The resolution came in the form of an undervolt and underclock.

If anyone has any explanation as to why an undervolt and underclosk would resolve my issues despite the system never pulling anything near 330W during POST/BOOT even under the factory specs of the 2600X please chime in.

Your power brick is the culprit I would focus on - even official bricks can go bad or have manufacturing issues.
 
  • Like
Reactions: TheHig

TheHig

King of Cable Management
Oct 13, 2016
951
1,171
+1

Could be the brick after all you have been through. If you are in the US I can lend you my 330w brick to test. I’m on a road trip for like 12 days starting Thursday.

PM me to figure something out if you are interested.
 

Gutzy

Caliper Novice
Original poster
Mar 11, 2018
31
14
Your power brick is the culprit I would focus on - even official bricks can go bad or have manufacturing issues.

Thanks @jtd871 , so far DC voltage measured from the DC plug is at 19.7 volts and AC voltage from the power cord leading into the brick is at 120 volts AC. Last thing to check is the amps. Unfortunately my meter is limited to 10amps DC max and this brick states it pumps out 16.9 amps.

Will see if I can test amps soon. If all checks out fine then I will need to test out the HDplex 400 instead to make sure.

+1

Could be the brick after all you have been through. If you are in the US I can lend you my 330w brick to test. I’m on a road trip for like 12 days starting Thursday.

PM me to figure something out if you are interested.

Thanks for the offer @TheHig, I'm currently eyeing some ATX/SFX psu's on sale to use for testing/diagnosing so i may go that route since I've always really wanted one just for these situations.
 
  • Like
Reactions: TheHig