Stock 3950x crashes with prime95 small FFTs - Page 6 - Overclock.net - An Overclocking Community
Forum Jump: 

Stock 3950x crashes with prime95 small FFTs

Reply
 
Thread Tools
post #51 of 66 (permalink) Old 05-23-2020, 01:04 AM
New to Overclock.net
 
smonkie's Avatar
 
Join Date: Sep 2013
Posts: 177
Rep: 4 (Unique: 4)
Quote: Originally Posted by LetterKilled View Post
Update on my situation: RAM optimized for Ryzen (G.Skill trident Z neo) came yesterday and was actually able to post with xmp. Great sign. Took off LLC medium and ran tests. Failed. Took off xmp, still failed. Set LLC low. Failed.
So I now know the ram wasn’t a factor of causing the tests to fail. I am still within my replacement time so I called up Newegg and RMA’d the CPU. They were kind enough to send me the chip first, because I work remotely and can’t be without a computer. So in a few days I should see how that one turns out. I still don’t know if it could be the board causing this and I can’t test it because everything is pretty much sold out from Newegg. So I can’t even get a board replacement.
Really looking forward to your tests. I doubt the issue will be the CPU, but would be satisfying to point that out finally.

By the way, I tried the latest BIOS for x570 Master (F12f) and the issue still persists. My ticket has been forwared to another team.
smonkie is offline  
Sponsored Links
Advertisement
 
post #52 of 66 (permalink) Old 05-24-2020, 11:02 AM
New to Overclock.net
 
Join Date: May 2020
Posts: 1
Rep: 0
Hello,

I just want to add myself to the list of people with the same issue.

Motherboard: x570 Aorus Pro
CPU: 3900x

Cinebench r20 instant blue screen and Prime 95 errors at stock settings. LLC at Medium seems to solve it.
threesix is offline  
post #53 of 66 (permalink) Old 05-24-2020, 11:20 AM
New to Overclock.net
 
The Stilt's Avatar
 
Join Date: May 2013
Posts: 2,638
Rep: 868 (Unique: 331)
So is this somehow related to Gigabyte boards?

During stock operation the load-line of the CPU should NOT be touched in any way.
This is because the CPU autonomously adjusts the load-line based on the actual requirements, through the SVI2 interface.

The load-line controls available in the bios in the vast majority of cases are overrides, which will prevent the CPU from doing this (since ultimately its the VRM controller that sets the load-line).
So for instance if the the CPU requires and sets the load-line to -40% setting and the user has enabled e.g. ±0% "Standard" setting technically represents the stock load-line characteristics,
this might prevent the correct load-line set by the CPU itself coming to effect. At least IR based controllers have a specific load-line override-enabled bit for this purpose, but obviously it is still possible
to have the faulty behavior even on the good controllers, if there are issues e.g. in the bios code.

If increasing the load-line fixes the issue and it only happens on specific model of board, then most likely there is some sort of an issue with either the bios (as described above) or with the hardware design.
The Stilt is offline  
Sponsored Links
Advertisement
 
post #54 of 66 (permalink) Old 05-25-2020, 02:18 AM
New to Overclock.net
 
LetterKilled's Avatar
 
Join Date: Mar 2012
Location: Sacramento
Posts: 73
Rep: 1 (Unique: 1)
Quote: Originally Posted by The Stilt View Post
So is this somehow related to Gigabyte boards?

During stock operation the load-line of the CPU should NOT be touched in any way.
This is because the CPU autonomously adjusts the load-line based on the actual requirements, through the SVI2 interface.

The load-line controls available in the bios in the vast majority of cases are overrides, which will prevent the CPU from doing this (since ultimately its the VRM controller that sets the load-line).
So for instance if the the CPU requires and sets the load-line to -40% setting and the user has enabled e.g. ±0% "Standard" setting technically represents the stock load-line characteristics,
this might prevent the correct load-line set by the CPU itself coming to effect. At least IR based controllers have a specific load-line override-enabled bit for this purpose, but obviously it is still possible
to have the faulty behavior even on the good controllers, if there are issues e.g. in the bios code.

If increasing the load-line fixes the issue and it only happens on specific model of board, then most likely there is some sort of an issue with either the bios (as described above) or with the hardware design.
I feel like it’s the board causing the problem too. Though if there’s a flaw with them, why aren’t all boards having this problem or is that just because not everyone tests them?
LetterKilled is offline  
post #55 of 66 (permalink) Old 05-25-2020, 02:32 AM
New to Overclock.net
 
Join Date: Feb 2017
Location: Sweden
Posts: 1,138
Rep: 44 (Unique: 37)
I think it might be something set wrong in the bios code in regard to the LLC & voltage. The thing seems to effect several Gigabyte boards on AM4. They as I recall develop the bios for one board specific in the line-up and then transfer the code to the other ones with needed adjustments if necessary. They use varied components for power delivery, maybe the code doesn't work correct for each and every layout?
Or a setting has been set that is wrong but missed in QC & bug control?

Gigabyte X570 AORUS XTREME, Ryzen 7 3800X PBO boost, 3800/1900 MEM/FCLK, 13.23.17.30.66.tRFC 600 1T 1.700V
4x8Gb Kingston HX434C19FB2K2/16 3466C19 1.2V (Micron Rev.E 16nm [D9VPP])
Nighthog is offline  
post #56 of 66 (permalink) Old 05-26-2020, 11:03 PM
New to Overclock.net
 
dansi's Avatar
 
Join Date: Mar 2009
Posts: 995
Rep: 32 (Unique: 22)
using pbo + undervolt with most/all voltages at 'normal', including LLC. gigabyte x570 here.
Ran p95 v298 small FFT for 5 mins, seems fine, but did not proceed longer because temps hit 80c!
dansi is offline  
post #57 of 66 (permalink) Old 05-27-2020, 03:57 PM
New to Overclock.net
 
Join Date: Feb 2017
Location: Sweden
Posts: 1,138
Rep: 44 (Unique: 37)
Quote: Originally Posted by dansi View Post
using pbo + undervolt with most/all voltages at 'normal', including LLC. gigabyte x570 here.
Ran p95 v298 small FFT for 5 mins, seems fine, but did not proceed longer because temps hit 80c!
You can have it run 90C as much as you like. 95C might give you errors but usually it's fine 99% of the time. You don't really want it higher than that for stability reasons.

Gigabyte X570 AORUS XTREME, Ryzen 7 3800X PBO boost, 3800/1900 MEM/FCLK, 13.23.17.30.66.tRFC 600 1T 1.700V
4x8Gb Kingston HX434C19FB2K2/16 3466C19 1.2V (Micron Rev.E 16nm [D9VPP])
Nighthog is offline  
post #58 of 66 (permalink) Old 05-28-2020, 05:07 AM
New to Overclock.net
 
smonkie's Avatar
 
Join Date: Sep 2013
Posts: 177
Rep: 4 (Unique: 4)
Quick update: I've tried the same 3950X with a x570 Taichi, three different BIOS versions. UEFI defaults. Prime95 seems somehow more stable and it's using more voltage, but crashes anyway within few minutes.

Does this mean it's the CPU which is faulty? Or is just an expected Prime95 behaviour with 3950X? Haven't had a faulty CPU in... never.

Last edited by smonkie; 05-28-2020 at 05:16 AM.
smonkie is offline  
post #59 of 66 (permalink) Old 05-28-2020, 07:09 AM
New to Overclock.net
 
The Stilt's Avatar
 
Join Date: May 2013
Posts: 2,638
Rep: 868 (Unique: 331)
Quote: Originally Posted by smonkie View Post
Quick update: I've tried the same 3950X with a x570 Taichi, three different BIOS versions. UEFI defaults. Prime95 seems somehow more stable and it's using more voltage, but crashes anyway within few minutes.

Does this mean it's the CPU which is faulty? Or is just an expected Prime95 behaviour with 3950X? Haven't had a faulty CPU in... never.
Are the temperatures staying within the spec, i.e. =< 95°C tCTL during the stress?

If yes, then the CPU is defective.
The Stilt is offline  
post #60 of 66 (permalink) Old 05-28-2020, 08:24 AM
New to Overclock.net
 
DemonAk's Avatar
 
Join Date: Mar 2014
Posts: 26
Rep: 0
Quote: Originally Posted by DemonAk View Post
the same problem guys. mb x370 taichi. First i thinked it's defective cpu, yesterday i replaced to new one and nothing changed (not stable at stock). only help set: LLC1, PBO ON or cpu offset +0.05v. Asrock suport can't help. using latest bios 6.20b agesa 1.0.0.4b but trying 1.0.0.3abba same problem, not stable prime95 small fft's avx and occt power test with avx2.
So, i tested more and +0.05 not fully stable because in settings i have set cpu LLC auto (level 5). Only if i set +0.05v and LLC2(3) rock solid stable prime95 small fft and OCCT small data set with AVX/AVX2 pass. I want to replace my board to another to check this problem.
Spoiler!

Ryzen 3950x (batch 2008), ASRock X370 Taichi, 4x32gb [email protected], Palit GTX 1080ti JetStream, Windows 10 x64 Pro

Last edited by DemonAk; 05-28-2020 at 08:38 AM.
DemonAk 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