AMD Ryzen: From Buggy to Bountiful

Requested information for SR #{ticketno:[8200766680]} Service Request: {ticketno:[8200799002]}

--------------- Introduction ---------------
Below is a video showing a Ryzen 7 1700X PC with Debian GNU/Linux running the ryzen-test program from https://github.com/Oxalin/ryzen-test/ -- complete with video feed showing the CPU and case cooling at work. Ryzen-test is a very mild stress test consisting of compiling GCC, which any CPU should be able to do with no problems. Certain early Ryzen chips seem to fail at this test every time. The presentation running at top left helps relieve some of the boredom with some system config information and problem description; otherwise it's mostly just the temperature very slowly climbing into working range. Spoiler: The video gets more interesting around 12m:36s, which is about when the test fails. Any good CPU should pass this test without any segmentation faults. I made this video to help AMD understand my issues as part of the process of getting an RMA on this CPU. Other information within this post was also listed to help in that regard. Things with personally identifying information, such as the original invoices, were sent separately; this is just the good stuff for both AMD's private and everyone's public consumption.
--------------- --------------- ---------------

Motherboard: Micro-Star International MSI B350 TOMAHAWK (MS-7A34) (per BIOS screenshots)

Motherboard BIOS Version: E7A34AMS.120 03/16/2017 (per BIOS screenshots)

Make and Model of RAM:
- DIMM Bank A: 2x GeIL CL15-15-15 D4-2133 (per BIOS screenshots), GeIL EVO POTENZA Desktop Memory Model GPR416GB2133C15DC (per Newegg Invoice)
- DIMM Bank B: 2x Team Group TEAMGROUP-UD4-2400 (per BIOS screenshots), Team Elite Plus Desktop Memory Model TPRD44G2400HC1601 (per Newegg Invoice)
- GeIL RAM Info: https://www.newegg.com/Product/Product.aspx?Item=N82E16820144924
- Team Group RAM Info: https://www.newegg.com/Product/Product.aspx?Item=N82E16820313741
- Total RAM GeIL 8 + GeIL 8 + Team Group 4 + Team Group 4 = 24GB
- DDR Speed: 2133MHz (per BIOS screenshots)

Make and Model of Power Supply Unit (per Newegg Invoice):
- Make: APEVIA
- Model: ATX-JP1000W 1000W ATX12V SLI CrossFire 80 PLUS BRONZE Certified Power Supply
- Further information: https://www.newegg.com/Product/Product.aspx?Item=N82E16817148060

Voltages (per BIOS screenshots):
- CPU Core: 1.208 [Auto]
- CPU NB: 0.904 [Auto]
- DRAM: 1.200 [Auto]
- System 3.3v/5v/12v: 3.360/5.160/12.288

Cooling strategy:
- CPU Cooling: Air
- Cooler Model: DEEPCOOL ICE BLADE 200M
- Cooler Rating: "For 130W CPUs" per mfr.
- Cooler Features: Dual 8mm direct-contact heatpipes, Dual 92mm Fan Push/Pull.
- Heat Sink Grease: Arctic Silver (Not the Deepcool OEM grease)
- Cooler Info: https://www.newegg.com/Product/Product.aspx?Item=N82E16835856001

Temperatures:
- Ambient (room) temp about 23 degrees C.
- No appreciable load, around 30 - 35 degrees C according to k10temp
- Under load, peak at around 52 degrees C per k10temp
- Temperature at failure point, 51.9 degrees C according to k10temp

Information video:
- Window: Impress presentation (Good Info)
- Window: https://github.com/Oxalin/ryzen-test (very mild stress test)
- Window: watch -n .3 sensors \|tail -9\|cut -d "\(" -f 1 (monitor AMD-related temperatures)
- Window: Camera monitoring inside of PC (showing cooling live as it happens)
- Window: tail -f /var/log/kern.log |grep segfault (watch for specific problems)
- Window: top (see top CPU processes)
- At failure time (12m:37s), evidence appears in ryzen-test and tail kern.log windows.
- Temperature at failure observed to be 51.9 degrees C.
- Video online: https://www.youtube.com/watch?v=Pc10dtRaXso.
- Video download: The_AMD_Ryzen_Linux_Bug_-_What_it_looks_like.mkv [140MB]

Test system:
- Debian GNU/Linux "10" Buster/sid
- MSI Radeon RX 460 using "amdgpu" kernel driver
- Kernel uname -a: Linux quiz 4.15.2-amd64-1 #1 SMP Sat Feb 10 20:41:29 EST 2018 x86_64 GNU/Linux (from Kernel.org; compiled on this system)

Bios Screenshots:

BIOS Screen Shot: OC Page with motherboard model, BIOS version and date and voltages

BIOS Screen Shot: Settings Page showing configuration information

BIOS Screen Shot: RAM Info for DIMM A1

BIOS Screen Shot: RAM Info for DIMM A2

BIOS Screen Shot: RAM Info for DIMM B1

BIOS Screen Shot: RAM Info for DIMM B2

Leave a Reply

Your email address will not be published. Required fields are marked *