TRX40 Creator - "Code 54" After Reboot |
Post Reply | Page 12> |
Author | |
Moose
Newbie Joined: 17 Apr 2018 Location: UK Status: Offline Points: 221 |
Post Options
Thanks(0)
Posted: 24 Apr 2021 at 9:27pm |
ASRock TRX40 Creator + AMD Ryzen Threadripper 3970X Processor
BIOS 1.73 64GB (4x Kingston Server Premier KSM32ES8/16ME Memory 16GB 3200MHz DDR4 ECC CL22 DIMM 1Rx8 Micron - Part Number: Kingston 9965684-032.A00G (Micron) PC always boots up fine from a cold start. After a software initiated reboot (windows restart for example), the system will stay on a black screen and the red debug LED cycles through and stops on Code 54 which seems to indicate a memory issue. After a manual reboot using the hardware reset button, the PC boots up everytime. I know this RAM is not on the official Memory QVL yet new but a very similar Kingston ECC is listed as supported (KSM26ED8/16ME). This issue might be a related to something else like a BIOS setting and I've tried various things. PC has been stress tested (CPU, RAM, GFX) and no issues over the past 2 months. Just raising awareness of this to see if anyone else has the issue? |
|
ket
Senior Member Joined: 13 Jul 2017 Location: UK Status: Offline Points: 1676 |
Post Options
Thanks(0)
|
You can try;
1. Running HCI memtest to see if any errors are detected, let it run to at least 300%. 2. Increase memory voltage to 1.4v 3. Take 2 sticks out and see if you get the same problem 4. Did you get this issue with firmware 1.7? (non-beta release) 5. Look at the serial numbers on your memory, the final digits will be in sequence. Eg; 0314, 0315, 0316, 0317. Install the memory modules so they are all in order. 6. Try increasing tCWL and tRDWR to 16 or 18. |
|
Moose
Newbie Joined: 17 Apr 2018 Location: UK Status: Offline Points: 221 |
Post Options
Thanks(0)
|
Thanks for these ideas to try.
1. Already ran various memtest86 tests which took a long time and passed, unless there is something special about HCI memtest. 2. Tried voltage at 1.4v on all banks, no change. 3. Tried with 2 sticks, no change. 4. Yes, happened in 1.7 non-beta. 5. Moved them to be in order by serial number (assuming bank A2 is the first slot, B2 second, C2 third and D2 last for quad channel) - no change. 6. tCWL was already set to 18 by default, so tested at 22 and no change. Also increased tRDWR to 18 and no change. |
|
ket
Senior Member Joined: 13 Jul 2017 Location: UK Status: Offline Points: 1676 |
Post Options
Thanks(0)
|
Have you tried disabling Gear Down mode and manually setting Command Rate to 2T? TRFC should already be set loose enough but you could try adding 50 or 100 to that and see if that helps as well.
|
|
Moose
Newbie Joined: 17 Apr 2018 Location: UK Status: Offline Points: 221 |
Post Options
Thanks(0)
|
Tried all that but didn't fix the issue. |
|
ket
Senior Member Joined: 13 Jul 2017 Location: UK Status: Offline Points: 1676 |
Post Options
Thanks(0)
|
What are your full system specs? POST failure on restart is probably a RAM training issue but depending on things like what CPU you have a voltage bump to vsoc might be effective, as might increasing ProcODT. Try downloading ZenTimings and posting a screenshot of it something might stand out.
Edited by ket - 29 May 2021 at 9:13am |
|
Moose
Newbie Joined: 17 Apr 2018 Location: UK Status: Offline Points: 221 |
Post Options
Thanks(0)
|
CPU: AMD 3970X
MB: TRX40 Creator RAM: 4x Kingston 16GB 3200 ECC (Matching) PSU: Seasonic 1300W Platinum GPU: RTX 3090 HDD: 2x WD SN850 Black 1TB, 1x Samsung 970 EVO 500GB |
|
ket
Senior Member Joined: 13 Jul 2017 Location: UK Status: Offline Points: 1676 |
Post Options
Thanks(0)
|
Do you know what ICs your memory is using? (You can get that infor with Thaiphoon Burner) Quite a lot stands out there;
1. Disable GDM (Gear Down Mode) 2. ProcODT should be ok at 60 Ohms but you can try 68 Ohms as you're running 4x sticks. 3. Try reducing VDDG voltage I've found on the 3000 series at least if that voltage is set too high it can actually negatively impact stability, on my 3700X with 32GB 3600MHz Klevv BoltX I have it set to 0.85v, it could probably go even lower than this. Try 0.9v and slowly work your way down to see if that makes a difference. 4. You have a lot of oddly aggressively set timings considering you are running 4 sticks across all DIMM banks. Try changing TRRDS to 6 or 7, TRRDL to 9 or 10, TFAW try 45, TWTRS try 6 or 8, TWRWRSD & TWRWRDD try 7 or 8, try a TCKE of 9. Lastly TRDRDSD & TRDRDDD might be a little tight as well try a value of 5 or 6. |
|
Moose
Newbie Joined: 17 Apr 2018 Location: UK Status: Offline Points: 221 |
Post Options
Thanks(0)
|
Tried all that and still the same.
The BIOS wouldn't allow a lower value than 0.9v for VDDG. This is the datasheet for the memory, it is the Micron variant: https://www.kingston.com/dataSheets/KSM32ES8_16ME.pdf This is what ZenTimings looks like after all these changes: |
|
ket
Senior Member Joined: 13 Jul 2017 Location: UK Status: Offline Points: 1676 |
Post Options
Thanks(0)
|
Micron is a real pain on Ryzen, the only CPUs I have that (kind of) run Micron memory happily are based on Zen+. Hynix based memory works very well on Zen, Zen+, Zen2, etc though. As does the now old and ever growing in expense B-Die. You could try switching CMD2T to 2T to see if that works, disabling PowerDown might help too. Outside of that though I think you're out of luck unless you're happy to reduce the memory frequency a bit.
|
|
Post Reply | Page 12> |
Tweet
|
Forum Jump | Forum Permissions You cannot post new topics in this forum You cannot reply to topics in this forum You cannot delete your posts in this forum You cannot edit your posts in this forum You cannot create polls in this forum You cannot vote in polls in this forum |