MSI X570 Tomahawk new 1.3 BIOS update no good.... Idle shutdown

Spartacus

2[H]4U
Joined
Apr 29, 2005
Messages
2,130
Tried out the 7C84v13 update today and it caused a spontaneous shutdown at idle.
Nothing in the Event log other than the unplanned shutdown.

Only overclocking items enabled were XMP and PBO Enabled (instead of Auto).
Rock solid again back on the 1.2 BIOS, same settings.

System was fine with last two BIOS version as well.

So something seems fubar with this latest update.

Update description is " Updated AMD AGESA ComboAm4v2PI 1.0.8.1 ".

.
 
Tried out the 7C84v13 update today and it caused a spontaneous shutdown at idle.
Nothing in the Event log other than the unplanned shutdown.

Only overclocking items enabled were XMP and PBO Enabled (instead of Auto).
Rock solid again back on the 1.2 BIOS, same settings.

System was fine with last two BIOS version as well.

So something seems fubar with this latest update.
Actually, it is a problem with virtually all AMD motherboard BIOSes with XMP memory. The biggest problem is that the XMP profiles will apply an incorrect amount of VDDCR SOC voltage - often, less than 1.0V. This will cause stability issues, in the form of frequent program crashes (with older BIOSes) or rantom reboots during idle (with newer BIOSes). It just seems as though AMD really only likes stock JEDEC memory profiles (which means slow speeds and/or very loose timings).

As a result, I learned to live with all of those problems with AMD incorrectly interpreting the XMP profiles, and set the required parameters manually in the BIOS.
 
Actually, it is a problem with virtually all AMD motherboard BIOSes with XMP memory. The biggest problem is that the XMP profiles will apply an incorrect amount of VDDCR SOC voltage - often, less than 1.0V. This will cause stability issues, in the form of frequent program crashes (with older BIOSes) or rantom reboots during idle (with newer BIOSes). It just seems as though AMD really only likes stock JEDEC memory profiles (which means slow speeds and/or very loose timings).

As a result, I learned to live with all of those problems with AMD incorrectly interpreting the XMP profiles, and set the required parameters manually in the BIOS.

Thanks for the info. I know XMP has been an on-going issue on the AMD boards for a while.
I thought less so with the newest round of mobos from what I've read.

You may be correct that the new AMD AGESA broke XMP.

XMP has been running flawlessly on every BIOS version until this one though, so I think for now I'll skip this BIOS.

ETA: You're right that it's better to dial-in the memory timings manually. I used to do that but XMP has made me lazy. :)

.
 
Last edited:
AMD and XMP don't fare well together. Just download Ryzen Dram Calculator and use that to get your values and you'll be good to go.
 
AMD and XMP don't fare well together. Just download Ryzen Dram Calculator and use that to get your values and you'll be good to go.
And even that calculator is no guarantee of stable operation. You see, it currently assumes that all memory IC chips are of 8 Gbit and lower densities. This means DIMM sticks that are no larger than 16 GB total per dual-rank stick. If the OP were using 32 GB sticks like I do, then the calculated recommendation for the tRFC is way off: 16 Gbit ICs used in 32 GB DIMMs have a much higher tRFC latency than 8 Gbit ICs. To compensate, you must multiply the calculated value by 11, then divide that result by 7.

And if you have an advanced calculator function that includes tRFC, tRFC2 and tRFC4 (as Asus and some other brands have), then DO NOT take that calculated tRFC value as the tRFC value to input into the calculator as that would result in a much higher value than is appropriate. Instead, take that value, and multiply by 1,000 and then divide by your true memory clock speed in mHz (for example, 1,800 in the case of DDR4-3600 RAM). That will result in the ns tRFC value to input into the calculator. Alternatively, simply divide the calculated tRFC value by 1.346 to obtain the tRFC2 value, and then divide the calculated tRFC2 result by 1.625 to obtain the tRFC4 value.
 
Thanks, but the more I think about it, I don't think it's an XMP issue.
The system booted up fine on BIOS v1.3 and ran fine for a while.

Then it did a complete shutdown while just sitting idle.
Like pulling the power plug on it.

As I said the system runs fine with XMP enabled on BIOS v1.2 and every previous one.
I even tightened one of the memory timings beyond XMP now and it's still running fine.

Not really worried about it though, I'll just skip BIOS v1.3.

ETA: Memory is GSkill Ripjaws 8GBx2 3600.

.
 
Thanks, but the more I think about it, I don't think it's an XMP issue.
The system booted up fine on BIOS v1.3 and ran fine for a while.

Then it did a complete shutdown while just sitting idle.
Like pulling the power plug on it.

As I said the system runs fine with XMP enabled on BIOS v1.2 and every previous one.
I even tightened one of the memory timings beyond XMP now and it's still running fine.

Not really worried about it though, I'll just skip BIOS v1.3.

ETA: Memory is GSkill Ripjaws 8GBx2 3600.

.

Probably just as well that I cannot try it on my new B550 Gaming Plus then, since downloads are not working on the MSI site at this time.
 
So is it best to just run XMP to get the numbers then input those numbers manually? Is it the XMP timings that are the problem or something about just having XMP running?
 
No, as I noted above I don't think the problem is XMP in the 1.3 BIOS.
I think it has more to do with the "AGESA ComboAm4v2PI 1.0.8.1" update.

I'm still running the 1.2 BIOS with XMP enabled and one of the timings tightened.
So yes, you can still change the timings with XMP enabled.

I'll try the 1.4 BIOS when it's out of beta.

.
 
Actually, it is a problem with virtually all AMD motherboard BIOSes with XMP memory. The biggest problem is that the XMP profiles will apply an incorrect amount of VDDCR SOC voltage - often, less than 1.0V. This will cause stability issues, in the form of frequent program crashes (with older BIOSes) or rantom reboots during idle (with newer BIOSes). It just seems as though AMD really only likes stock JEDEC memory profiles (which means slow speeds and/or very loose timings).

As a result, I learned to live with all of those problems with AMD incorrectly interpreting the XMP profiles, and set the required parameters manually in the BIOS.

This hasn't been my experience and I've run through my share of X370, B350, X399, X470, B450 and X570 motherboards. It really depends on the RAM your using. I've seen issues with XMP profiles, but I've also enabled it on tons of these boards without a problem. If your using Corsair, you need to increase the voltage up to 1.4v, G.Skill is usually good to go though. I've run both B-Die and Hynix modules from them with GREAT success on all of the above motherboards. I've even managed to make 4x16GB DIMMs run with XMP profiles at DDR4 3600MHz speeds on a T-Topology board.
 
My x370 board was corsair and liked 1.38V - or 1.4. Gskill i switched to is happy with XMP. My TR1 system is having fits, but that may be a bad slot too. Still fiddling it.
 
Yep, I've been running GSkill memory for many years, no problems on either Intel or AMD systems with XMP.
 
Yep, I've been running GSkill memory for many years, no problems on either Intel or AMD systems with XMP.

That's actually what AMD sent us as part of the review kit for the Ryzen 3000 series. It's worked well with most test boards I've worked with since.
 
I flashed to v 1.44 beta on the MSI B550m today .. updated all windows as I am sure it's for the Ryzen 5000's support and the Ryzen 3 3100 is set to default but it's really picked up some speed as I use to have to auto overclock to hit 470 in single trend and now it does it stock .

Here is data https://valid.x86.fr/gph7l6
 
Back
Top