My 3950X system just died - Page 14 - Overclock.net - An Overclocking Community
Forum Jump: 

My 3950X system just died

Reply
 
Thread Tools
post #131 of 141 (permalink) Old 05-19-2020, 05:49 PM
New to Overclock.net
 
Join Date: Aug 2019
Posts: 309
Rep: 6 (Unique: 5)
The new 3950x should be faster too.
2600ryzen is offline  
Sponsored Links
Advertisement
 
post #132 of 141 (permalink) Old 05-19-2020, 06:17 PM
New to Overclock.net
 
tcclaviger's Avatar
 
Join Date: Mar 2012
Location: Washington State
Posts: 375
Rep: 23 (Unique: 20)
I suspect you're a victim of the awesome software voltage measurements we're all subject to these days.

I assume by "normal" vcore you mean allowing it to use VID as the system determines it should be. Normal + .1v is actually fairly high potentially when using "standard" LLC. The higher LLC is more damaging than the voltage setting over time on a [email protected] task, as it loads and unloads each core it's going to feed it a spike not reflected on anything besides an O-scope. On my 3900x VID + .100 offset + mid level LLC is around actual 1.225vcore measured, and something around 1.165 reported via SVI2 TFN. The LLC spikes can be 0.125v higher or lower than target, according to DerBaur's Oscope testing, so if you're board was actually providing 1.225 as mine does, you're chip would've been eating repeated 1.35 spikes while hot, repeatedly.

Not trying to make an excuse for AMD, it's unacceptable to have a chip die that fast at your settings. I've beat on Xeons FAR FAR harder than that for long duration and never had an issue.

Gigabyte is pretty well recognized as providing more voltage than reported, check this out, I know this is a single case, but it IS representative of what I've observed measuring boards using a DMM instead of software. MSI in my experience routinely undershoots the target, Gigabyte and Asus regularly overshoot the target:

As an example, on my C6E Asus board, the SVI2 TFN is NEVER correct, the ASUS WMI reported voltage is closer, always being about 0.02 lower in software than shown on the DMM.
Attached Thumbnails
Click image for larger version

Name:	Voltage.jpg
Views:	25
Size:	444.6 KB
ID:	347584  


Last edited by tcclaviger; 05-19-2020 at 06:33 PM.
tcclaviger is offline  
post #133 of 141 (permalink) Old 05-19-2020, 07:31 PM - Thread Starter
Overclocking since 1998
 
briank's Avatar
 
Join Date: Jan 2005
Location: USA
Posts: 926
Rep: 48 (Unique: 41)
Quote: Originally Posted by tcclaviger View Post
I suspect you're a victim of the awesome software voltage measurements we're all subject to these days.

I assume by "normal" vcore you mean allowing it to use VID as the system determines it should be. Normal + .1v is actually fairly high potentially when using "standard" LLC. The higher LLC is more damaging than the voltage setting over time on a [email protected] task, as it loads and unloads each core it's going to feed it a spike not reflected on anything besides an O-scope. On my 3900x VID + .100 offset + mid level LLC is around actual 1.225vcore measured, and something around 1.165 reported via SVI2 TFN. The LLC spikes can be 0.125v higher or lower than target, according to DerBaur's Oscope testing, so if you're board was actually providing 1.225 as mine does, you're chip would've been eating repeated 1.35 spikes while hot, repeatedly.

Not trying to make an excuse for AMD, it's unacceptable to have a chip die that fast at your settings. I've beat on Xeons FAR FAR harder than that for long duration and never had an issue.

Gigabyte is pretty well recognized as providing more voltage than reported, check this out, I know this is a single case, but it IS representative of what I've observed measuring boards using a DMM instead of software. MSI in my experience routinely undershoots the target, Gigabyte and Asus regularly overshoot the target:
https://www.youtube.com/watch?v=iQB7ELwBbVI

As an example, on my C6E Asus board, the SVI2 TFN is NEVER correct, the ASUS WMI reported voltage is closer, always being about 0.02 lower in software than shown on the DMM.
You're right, the only voltage you can trust is measured with an oscilloscope. Since most don't have access to one though, the generally accepted practice around here is to run stock settings, and observe the SVI2 TFN Vcore under load. That gives you some idea what the factory determined to be safe for this chip during binning. As long as SVI2 TFN is consistent, it doesn't matter that its reading at an offset to reality. Using this observe and then OC method, my Vcore average voltage was never higher than stock because my SVI2 TFN measurement was lower than stock. I can't speak to transients without the O-scope though.

And the fact that I was running a 3.65GHz underclock >90% of the time at a Vcore much lower than stock.... I really struggle to accept that I abused this chip in any way.

CPU
Ryzen 9 3950X
Motherboard
Gigabyte X570 Aorus Ultra
GPU
Powercolor 5700XT Red Devil
RAM
TridentZ B-die @ 3733CL16
Hard Drive
Samsung 970 Pro
Hard Drive
Adata XPG SX8200 Pro
Optical Drive
BD-RE
Power Supply
Seasonic Prime Titanium 750W
Cooling
Noctua NH-D15
Case
Fractal Define R5
Operating System
Windows 10 Home
Monitor
LG 27GL83A-B 27" Nano-IPS
Keyboard
Corsair K68 Cherry Reds
Mouse
Razer Deathadder
CPU
7700K Delid @5.1GHz @1.4V
Motherboard
Asus Z270F Strix
GPU
Asus RX480 Strix @ 1400MHz
RAM
Corsair Vengance PC3200 2x8GB
Hard Drive
Sandisk Ultra II 960GB
Hard Drive
Samsung 970 Pro
Hard Drive
2 x 3TB HDD
Optical Drive
LG BD-R
Power Supply
Seasonic Prime 750W Titanium
Cooling
Noctua NH-D15
Case
Fractal Define R5
Operating System
Windows 10 Home
Monitor
Dell 27" 1080p
Mouse
Razer DeathAdder
▲ hide details ▲


briank is offline  
Sponsored Links
Advertisement
 
post #134 of 141 (permalink) Old 05-19-2020, 07:32 PM - Thread Starter
Overclocking since 1998
 
briank's Avatar
 
Join Date: Jan 2005
Location: USA
Posts: 926
Rep: 48 (Unique: 41)
Quote: Originally Posted by 2600ryzen View Post
The new 3950x should be faster too.
That's my hope. I know before the 3950X I had a 3600X, and the 3600 I'm using now to fill the gap, is faster than that early 3600X. The 7nm process is maturing as one would expect and clocks are getting higher.

CPU
Ryzen 9 3950X
Motherboard
Gigabyte X570 Aorus Ultra
GPU
Powercolor 5700XT Red Devil
RAM
TridentZ B-die @ 3733CL16
Hard Drive
Samsung 970 Pro
Hard Drive
Adata XPG SX8200 Pro
Optical Drive
BD-RE
Power Supply
Seasonic Prime Titanium 750W
Cooling
Noctua NH-D15
Case
Fractal Define R5
Operating System
Windows 10 Home
Monitor
LG 27GL83A-B 27" Nano-IPS
Keyboard
Corsair K68 Cherry Reds
Mouse
Razer Deathadder
CPU
7700K Delid @5.1GHz @1.4V
Motherboard
Asus Z270F Strix
GPU
Asus RX480 Strix @ 1400MHz
RAM
Corsair Vengance PC3200 2x8GB
Hard Drive
Sandisk Ultra II 960GB
Hard Drive
Samsung 970 Pro
Hard Drive
2 x 3TB HDD
Optical Drive
LG BD-R
Power Supply
Seasonic Prime 750W Titanium
Cooling
Noctua NH-D15
Case
Fractal Define R5
Operating System
Windows 10 Home
Monitor
Dell 27" 1080p
Mouse
Razer DeathAdder
▲ hide details ▲


briank is offline  
post #135 of 141 (permalink) Old 05-19-2020, 07:35 PM
New to Overclock.net
 
tcclaviger's Avatar
 
Join Date: Mar 2012
Location: Washington State
Posts: 375
Rep: 23 (Unique: 20)
I totally agree briank, what you were doing wasn't abusive at all. Chip shouldn't have died.
tcclaviger is offline  
post #136 of 141 (permalink) Old 05-19-2020, 09:07 PM
Tech Enthusiast
 
oreonutz's Avatar
 
Join Date: Jun 2017
Location: Las Vegas, NV
Posts: 1,325
Rep: 99 (Unique: 65)
Quote: Originally Posted by tcclaviger View Post
I totally agree briank, what you were doing wasn't abusive at all. Chip shouldn't have died.
Definitely it should not have. There will always be an acceptable amount of failures, but it is typically unheard of. We should keep our eye on this, if we start to see this happening, even with just 2 to 3% of our user base here, then we definitely have an issue on our hand. Hopefully @briank is that one case out of a million that this happens to, and we won't see it again in cases where people weren't shoving 250amps down the chip.

-MattTheTech

Offical Ryzen 9 3900x Owner!
Offical Ryzen 9 3950x Owner!
DodekaZen3000
(34 items)
CPU
Ryzen 9 3950x
Motherboard
ASUS Crosshair VII Hero
GPU
EVGA 1080ti FTW3 Hybrid
RAM
G.SKILL Flare X Series 32GB (4 x 8GB) 3200Mhz (PC4 25600) - F4-3200C14D-16GFX (x2 Kits)
Hard Drive
Samsung 970 Pro 512GB
Hard Drive
Samsung 970 Evo 1TB
Hard Drive
Samsung 850/860 Evo (x3 Raid 0 Game Drive)
Hard Drive
WD Black 4TB (x2 Raid 0 Storage Drive)
Hard Drive
860 Evo
Optical Drive
LG Blu-Ray Burner (Model BH16NS40)
Optical Drive
5.25" to One 3.5" and One 2.25" Bay (Model Syba SY-MRA55006)
Power Supply
Corsair RM1000i
Cooling
EKWB Phoenix 360mm Modular Cooler
Cooling
280mm Alphacool UT60
Cooling
280mm EKWB Coolstream CE
Cooling
Optimus Foundation AM4 CPU Block
Cooling
x2 Corsair Pump/Res Combo
Cooling
EKWB Acrylic 1080ti FTW Block
Cooling
Aquaero 6xt
Case
Old Jonesbo Case
Operating System
Windows 10 Enterprise
Monitor
Samsung NU8000 55Inch TV (Its a 4k TV But I run in it 1440p 120hz)
Monitor
LG 1080p 34 Inch Diplay
Keyboard
Corsair k70 RGB Lux Mk2
Mouse
Logitech G700s
Mousepad
Razer Vespula V2
Audio
Custom Built 7.1 Surround System
Audio
HT-Omega eClaro 7.1 Sound Card
Audio
Focusrite Scarlett 18i20
Other
Aquantia 10Gb NIC
Other
LSI 9211-8i
Other
Oculus Rift HMD
Other
10 ML120 + 6 ML140 Corsair Maglev Fans
Other
Phanteks Halo aRGB LUX x5
▲ hide details ▲
oreonutz is online now  
post #137 of 141 (permalink) Old 05-25-2020, 03:14 AM
New to Overclock.net
 
Join Date: Feb 2017
Location: Sweden
Posts: 1,298
Rep: 54 (Unique: 42)
@tcclaviger

AM4 Gigabyte boards are pretty accurate on reported voltages. I tested mine with DMM and they were the same as the software reports (SVI2 TFN) if measured behind the socket. The other Vcore is more close to the VRM capacitors.

I know "faults" can happen and did so in the past but current generation Gigabyte AM4 is good and can be trusted on what they report.

EDIT: After watching the video once more since 2018 when it came out?

If you watch closely the motherboard is reporting the voltage for CPU VCORE SOC correctly.... They are looking at the wrong place, wrong voltage reading to what they are measuring on the back on the board.

Gigabyte X570 AORUS XTREME, Ryzen 7 3800X PBO boost, 2x8GB 4866/1900 18.27.24.48.90.tRFC 680 1T 1.650V [Micron Rev.J]
4x8GB 3800/1900 13.23.17.30.66.tRFC 600 1T 1.700V Kingston HX434C19FB2K2/16 3466C19 1.2V (Micron Rev.E 19nm [D9VPP]),
2x8GB Kingston HX436C17FB3K2/16 3600C17 1.35V (Micron Rev.J 17nm),
2x8GB 3866/1933 15.21.20.32.60.tRFC 448 1T 1.590V Kingston HX436C17FB3K2/16 3600C17 1.35V (Hynix DJR 17nm)

Last edited by Nighthog; 05-25-2020 at 03:32 AM.
Nighthog is offline  
post #138 of 141 (permalink) Old 06-24-2020, 08:19 AM
New to Overclock.net
 
Join Date: Jun 2020
Posts: 2
Rep: 0
Quote: Originally Posted by briank View Post
Final update:
Replacement 3950X received under warranty. Back up and running! There were no hassles at all with AMD. Their warranty support is very good.

Updated situation:
Confirmed dead 3950X. I've tried an alternate power supply, DRAM kit, GPU and motherboard. The 3950X won't boot with any of them. Ordered a Ryzen 3600 to try out. It works in both the original system and the 2nd one I built up.




Btw, the 3.65GHz underclock was at 1.1V Vcore under load. While the folding workload was stressful, the settings were not.

Original post:

Background:
My hardware is in my sig. (3950X, Gigabyte X570 Ultra, B-die 2x16GB, 5700XT, Seasonic Prime 750W)

I've been running this system since January. Various PBO overclocks before settling on all-core OCs. While I found a 4.35/4.25GHz stable OC, I usually run a 3.65GHz all core underclock because I've got the CPU running folding about 60 hours a week. I mention this because that is pretty much the most stress the CPU has seen. Even my PBO days, I made sure the CPU never saw 1.3V under load.

So tonight, just sitting idle, the machine went unresponsive, black screen.
Turn off the PSU and Reboot: nothing
Reset CMOS: nothing. Looking at the status LEDs: "CPU" comes on briefly. It progresses to "DRAM" and stays mostly there, but occasionally starts over and CPU lights briefly before settling back on DRAM.
I swapped out DDR4 for another set I had. Tried both sets 1 stick at a time, one channel at a time. Cleared CMOS each time. Same result. Stuck at the DRAM light.


Tomorrow, I'm going to borrow another PSU and GPU to swap out, but I'm not too hopeful. I went ahead an ordered a Ryzen 3600. If the PSU or GPU solve the problem I can return it. If they don't, I'll use the 3600 temporarily while I going through the warranty claim attempt.

Hey, this is my first post.


I had a very similar issue. My 3950X died a couple of weeks ago. Replacement 3950X is in the mail (RMA).

I got the CPU just after launch, in November 2019. I use my system for rendering in 3ds Max, not every day, but quite regularly. No OC. Stock settings on an Asus B450 motherboard.

The computer suddenly shut down and never booted again. Stuck in a boot loop when powered on (looping LED's as well). I changed everything in the system and finally bought a Ryzen 3400G to test it. Everything works fine.

I sent my diagnostics procedure to AMD and they accepted the RMA. As mentioned above, my replacement 3950X is in the mail.


Bad luck, I suppose.
phil5 is offline  
post #139 of 141 (permalink) Old 06-25-2020, 05:49 AM - Thread Starter
Overclocking since 1998
 
briank's Avatar
 
Join Date: Jan 2005
Location: USA
Posts: 926
Rep: 48 (Unique: 41)
Quote: Originally Posted by phil5 View Post
Hey, this is my first post.


I had a very similar issue. My 3950X died a couple of weeks ago. Replacement 3950X is in the mail (RMA).

I got the CPU just after launch, in November 2019. I use my system for rendering in 3ds Max, not every day, but quite regularly. No OC. Stock settings on an Asus B450 motherboard.

The computer suddenly shut down and never booted again. Stuck in a boot loop when powered on (looping LED's as well). I changed everything in the system and finally bought a Ryzen 3400G to test it. Everything works fine.

I sent my diagnostics procedure to AMD and they accepted the RMA. As mentioned above, my replacement 3950X is in the mail.


Bad luck, I suppose.
Sorry to hear about your loss Phil. Welcome to the club (although let's hope this remains a very exclusive club)!

CPU
Ryzen 9 3950X
Motherboard
Gigabyte X570 Aorus Ultra
GPU
Powercolor 5700XT Red Devil
RAM
TridentZ B-die @ 3733CL16
Hard Drive
Samsung 970 Pro
Hard Drive
Adata XPG SX8200 Pro
Optical Drive
BD-RE
Power Supply
Seasonic Prime Titanium 750W
Cooling
Noctua NH-D15
Case
Fractal Define R5
Operating System
Windows 10 Home
Monitor
LG 27GL83A-B 27" Nano-IPS
Keyboard
Corsair K68 Cherry Reds
Mouse
Razer Deathadder
CPU
7700K Delid @5.1GHz @1.4V
Motherboard
Asus Z270F Strix
GPU
Asus RX480 Strix @ 1400MHz
RAM
Corsair Vengance PC3200 2x8GB
Hard Drive
Sandisk Ultra II 960GB
Hard Drive
Samsung 970 Pro
Hard Drive
2 x 3TB HDD
Optical Drive
LG BD-R
Power Supply
Seasonic Prime 750W Titanium
Cooling
Noctua NH-D15
Case
Fractal Define R5
Operating System
Windows 10 Home
Monitor
Dell 27" 1080p
Mouse
Razer DeathAdder
▲ hide details ▲


briank is offline  
post #140 of 141 (permalink) Old 06-25-2020, 08:47 AM
New to Overclock.net
 
Join Date: Jun 2020
Posts: 2
Rep: 0
Quote: Originally Posted by briank View Post
Sorry to hear about your loss Phil. Welcome to the club (although let's hope this remains a very exclusive club)!

Thank you

Let's hope!
phil5 is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off