MAG X670E Tomahawk WIFI — weird issue with EXPO (amber light stays on when attempting to reboot)

[darkpetre156302de]

Here are my specs:
CPU: Ryzen 7 7800X3D
GPU: Asus ROG STRIX 4090
Motherboard: MAG X670E Tomahawk WIFI
BIOS: 1.80
SSD: 2xSamsung 990 PRO (1x4 TB, 1x2 TB) (most recent firmware)
RAM: Corsair Vengeance 64 GB CL30 (CMH64GX5M2B6000Z30) (2x32 GB sticks) in A2/B2
AiO: NZXT Kraken Elite 360 RGB
Fans: 7xLian Li Infinity 120
OS: Windows 11 Pro
PSU: Seasonic Vertex GX-1000
Everything is up-to-date firmware- and driver-wise.

My issue is that anything above JEDEC default (4800 MHz) will cause the system to refuse to go through the full reboot cycle regardless of where it is initiated (Ctrl+Alt+Del in BIOS, reset button, restart option in Windows etc.). The amber/yellow LED will stay on and not proceed to POST, however even after waiting for 30 minutes nothing further will happen so that I am forced to do a cold boot instead. The system runs absolutely stable, and I've never run into a BSOD or freezes, it's only that I can't effectively reboot the system. If I go back to the default JEDEC setting, then the problem is gone.

Here are the things I've already attempted:
- RAM-related:
* switching RAM stick positions
* starting with one RAM stick only
* doing a memory health check (although the kit has on-die ECC)
* using a different kit of RAM sticks (G.Skill Ripjaws S5 [F5-6000J3040G32GX2-RS5K])
* changing voltage
* trying different timings
* experimenting with MCR/PDE

- system-related:
* unplugging everything, externally and internally, including the graphics card
* changing the battery
* doing a BIOS flashback
* re-cabling and re-plugging everything
* trying all available BIOS versions, including doing consecutive CMOS resets after re-flashing
* reseating the CPU

For now, I am merely considering that I've been unlucky with my choice of RAM sticks, but it strikes me as odd that I can't complain otherwise about the performance so far, aside from the reboot-hang issue. Am I overlooking something?

 

[citay]

Sounds like your typical MSI AM5 problems. I'd open a ticket with MSI and hope they can get on top of it.

* doing a memory health check (although the kit has on-die ECC)

Don't have a false sense of security. Nothing about having to verify RAM stability externally, by running stress tests for example, has changed. On-die ECC is not module-wide ECC. As the name suggests, it works per-die (per individual memory chip on the module), because the higher-density DDR5 dice (dies?) are more prone to bit flips. On-die ECC can correct single-bit errors / bit flips on the die, so they don't have to throw away a whole die if there's only a few single-bit errors. It is essentially a tool that has become necessary for DDR5 production, to make enough chips pass the validation with the higher-density and smaller manufacturing processes, which automatically makes them more prone to errors. You can think of it like this: As the structures on the die become smaller and smaller over time, they more and more need these technologies in order to behave like for example DDR4 did before. It's not something that adds a lot of extra stability or anything, it just something that can preserve the status quo.

On-die ECC provides protection for data on the die, but not for data in transit or data that is being moved to/from the memory module and the CPU/GPU. It does not provide protection against errors happening outside the chip, such as those that occur on the memory module and integrated memory controller that is inside the CPU. So you can easily have RAM errors even on DDR5. Just set your speed to high and/or the timings too aggressive and/or the voltage too low, and you will get tons of errors and bit flips in tests like Memtest86.

But as i said, this looks like a typical AM5 problem here, things like this can happen, you can read it all over the forum. Intel are still about a year ahead in their DDR5 memory support maturity and in their CPU's IMC quality. That's not to say everything is perfect there, but a thing like you're experiencing for example would be highly unusual, while at AM5 it just comes with the territory almost.

 

[darkpetre156302de]

Agreed, on-die ECC is no guarantee, hence why I still did the check. Better safe than sorry, but anyway, the behavior was identical with a different kit.

So, what I am gathering from this is: it could be just my luck and my CPU's IMC is a bit on the weak side?

I have been in contact with the German MSI support, but their suggestion was to simply replace the board. However, meanwhile I thought I had it fixed and messaged them the news about it, only to realize later that I was wrong. I enabled EXPO while in JEDEC configuration, and of course, that would work (hindsight) because the entire system was in a different state! I am not yet convinced that the board is defect because then as a general rule it would not cold boot with full EXPO settings either. Either that, or the system would suffer from instability under load if there was a defective component at play. I could try with a different CPU and try again, or I could RMA the board, but end up with the same problem still...

To your point about Intel's IMC being ahead of this: Yes, this is very much true, they have always been better at this, but this is a bit of a conundrum because the trade-offs between AMD and Intel are too obvious at the moment. In fact, since I am playing at 4k anyway and DDR5 being a lot faster than DDR4, I could live with JEDEC settings still.

 

[citay]

I don't think an RMA is the answer here, especially not of the board, that is a big hassle, and if the problem is more BIOS-/AGESA- or CPU-/IMC-related, then it will all be for nothing.

You can try an intermediate setting: Enable EXPO, but instead of pressing F10 to save&exit, also set DRAM Frequency to DDR5-5600 for example. See if it behaves properly with that. I hope and believe that MSI and AMD will get issues such as yours under control eventually, but the first AM5 board and CPU generations will have some quirks at the beginning. It was even worse when AM4 and the Ryzen 1000-series first came out.

 

[sherbrovideo15a102de]

A few weeks ago I read that this guy fixed it like this:
https://forum-en.msi.com/index.php?threads/problem-with-reboots-when-expo-is-enabled.394791/
Myself not so lucky:
https://forum-en.msi.com/index.php?threads/b650-tomahawk-cant-restart-pc-with-expo-on.388561/
Mainboard, ram, cpu, bios, whatever is wonky is making this problem for us an no one really has an answer.

 

[darkpetre156302de]

I don't think an RMA is the answer here, especially not of the board, that is a big hassle, and if the problem is more BIOS-/AGESA- or CPU-/IMC-related, then it will all be for nothing.

You can try an intermediate setting: Enable EXPO, but instead of pressing F10 to save&exit, also set DRAM Frequency to DDR5-5600 for example. See if it behaves properly with that. I hope and believe that MSI and AMD will get issues such as yours under control eventually, but the first AM5 board and CPU generations will have some quirks at the beginning. It was even worse when AM4 and the Ryzen 1000-series first came out.

I tried lowering the frequencies before — even though I forgot to mention it, and there seemed to be general issues with EXPO, but not because of EXPO as a concept but because of what it technically does (load a manufacturer-tuned OC preset). However, I finally got it running, mainly because of:

A few weeks ago I read that this guy fixed it like this:
https://forum-en.msi.com/index.php?threads/problem-with-reboots-when-expo-is-enabled.394791/
Myself not so lucky:
https://forum-en.msi.com/index.php?threads/b650-tomahawk-cant-restart-pc-with-expo-on.388561/
Mainboard, ram, cpu, bios, whatever is wonky is making this problem for us an no one really has an answer.

This actually worked for me with a few changes of my own. There are two important things to consider here, the first being that you must use the readouts of your kit from when you had EXPO enabled. The second one is that you probably also must set the DRAM voltage to 1.4 volts. The latter seemed to be crucial for my configuration because keeping it on Auto will cause the system to not boot anymore, forcing me to perform a CMOS reset. I suspect that when the DRAM voltage is set on Auto it will especially introduce another point of failure because then it would always boot with the lowest voltage setting first. That when mismatched with present timings would simply cause the system to hang. It could simply be the order of internal readout which causes the issue here when combined with EXPO, and gets even worse without.

So going without EXPO, manual setting of timings and voltages instead seems to be the solution for my problem scenario which I described in my original posting.
I also lowered the SOC to 1.15 V because apparently there is really no need to have it 1.3 V even with 6000 MT/s.

The system runs stable now with Curve Optimizer settings to -20 on all cores and PPT limit set to 75 W.

 

[sherbrovideo15a102de]

This actually worked for me with a few changes of my own. There are two important things to consider here, the first being that you must use the readouts of your kit from when you had EXPO enabled. The second one is that you probably also must set the DRAM voltage to 1.4 volts. The latter seemed to be crucial for my configuration because keeping it on Auto will cause the system to not boot anymore, forcing me to perform a CMOS reset. I suspect that when the DRAM voltage is set on Auto it will especially introduce another point of failure because then it would always boot with the lowest voltage setting first. That when mismatched with present timings would simply cause the system to hang. It could simply be the order of internal readout which causes the issue here when combined with EXPO, and gets even worse without.

So going without EXPO, manual setting of timings and voltages instead seems to be the solution for my problem scenario which I described in my original posting.
I also lowered the SOC to 1.15 V because apparently there is really no need to have it 1.3 V even with 6000 MT/s.

The system runs stable now with Curve Optimizer settings to -20 on all cores and PPT limit set to 75 W.

Glad it worked out for you.

If you find the time maybe update the thread with your entire settings, voltages etc and ram sub-timings etc like the last member did in the topic I linked you, who knows maybe it will save someone else down the line. Cheers.

 

[darkpetre156302de]

Glad it worked out for you.

If you find the time maybe update the thread with your entire settings, voltages etc and ram sub-timings etc like the last member did in the topic I linked you, who knows maybe it will save someone else down the line. Cheers.

Alright. Keep in mind, the RAM timings may only apply to this kit CMH64GX5M2B6000Z30, i.e., Corsair Vengeance 2×32 GB, and not work for any other.

1. 1. Enter BIOS and load optimized defaults (F6)
2. 2. Save and reboot.
3. 3. Enter BIOS again.
4. 4. Set memory OC mode to EXPO and note down all timing values (values marked on the left, may differ on your side)
5. 5. Load defaults again.
6. 6. Start filling all timings with those noted down (replacing the unmarked values on the right)
7. 7. Set DRAM voltage to 1.4 V (important!)
8. 8. Enable Memory Context Restore and Power Down Enable.
9. 9. Reboot (DRAM LED may light up again for about 30–45 seconds, but it should boot normally after that)

Spoiler: For me, the following values are working now

Main Timing Configuration​
tCL	30
tRCD	36
tRP	36
tRAS	76
tRC	112
tWR	90
tRFC1	884
tRFC2	480
tRFCSB	390
tREFI	11677
Sub Timing Configuration​
tRTP	23
tRRDL	15
tRRDS	8
tFAW	32
tWTRL	30
tWTRS	8
Turn Around Timing Configuration​
tRDRDSCL	8
tRDRDSC	1
tRDRDSD	9
tRDRDDD	9
tWRWRSCL	23
tWRWRSC	1
tWRWRSD	9
tWRWRDD	9
tWRRD	8
tRDWR	20

I would also suggest lowering the SoC Voltage to 1.15 V as marked in the 3rd image.

---

For further tuning, I would also try setting PPT to 75 W and tweaking the curve via PBO Curve Optimizer. But I'd start with -5 or -10 and go down from there until it runs stable, since that part is down to silicon lottery as well. For me, -20 all-core works. Sometimes you'll have to increase SoC voltage again, but I'd try to be lower than 1.25 V.

 

[floepie0154002d9]

Alright. Keep in mind, the RAM timings may only apply to this kit CMH64GX5M2B6000Z30, i.e., Corsair Vengeance 2×32 GB, and not work for any other.

I suspect my difficulties with resume from standby may be related to your initial issue. Do you have any issues resuming from standby to report?
It's hard to troubleshoot, as sometimes, resuming just works, and at other times, I must hard reset due to a system hang.
Last question - what is your default EXPO DRAM voltage? I'm wary of frying something if I move it upward.

 

[vyankateshlahane2158002dd]

Here are my specs:
CPU: Ryzen 7 7800X3D
GPU: Asus ROG STRIX 4090
Motherboard: MAG X670E Tomahawk WIFI
BIOS: 1.80
SSD: 2xSamsung 990 PRO (1x4 TB, 1x2 TB) (most recent firmware)
RAM: Corsair Vengeance 64 GB CL30 (CMH64GX5M2B6000Z30) (2x32 GB sticks) in A2/B2
AiO: NZXT Kraken Elite 360 RGB
Fans: 7xLian Li Infinity 120
OS: Windows 11 Pro
PSU: Seasonic Vertex GX-1000
Everything is up-to-date firmware- and driver-wise.

My issue is that anything above JEDEC default (4800 MHz) will cause the system to refuse to go through the full reboot cycle regardless of where it is initiated (Ctrl+Alt+Del in BIOS, reset button, restart option in Windows etc.). The amber/yellow LED will stay on and not proceed to POST, however even after waiting for 30 minutes nothing further will happen so that I am forced to do a cold boot instead. The system runs absolutely stable, and I've never run into a BSOD or freezes, it's only that I can't effectively reboot the system. If I go back to the default JEDEC setting, then the problem is gone.

Here are the things I've already attempted:
- RAM-related:
* switching RAM stick positions
* starting with one RAM stick only
* doing a memory health check (although the kit has on-die ECC)
* using a different kit of RAM sticks (G.Skill Ripjaws S5 [F5-6000J3040G32GX2-RS5K])
* changing voltage
* trying different timings
* experimenting with MCR/PDE

- system-related:
* unplugging everything, externally and internally, including the graphics card
* changing the battery
* doing a BIOS flashback
* re-cabling and re-plugging everything
* trying all available BIOS versions, including doing consecutive CMOS resets after re-flashing
* reseating the CPU

For now, I am merely considering that I've been unlucky with my choice of RAM sticks, but it strikes me as odd that I can't complain otherwise about the performance so far, aside from the reboot-hang issue. Am I overlooking something?

Any solution on this???

 

[sherbrovideo15a102de]

Any solution on this???

Yes he fixed it, answer is a few posts above at post 8, here will copy paste it:

Some users managed to fix the restart issue like this doing a complete a to z manual ram overclock, both links are the same fix only difference is the ram kits used so they had different values to copy down and use for their own settings:
https://forum-en.msi.com/index.php?...ots-when-expo-is-enabled.394791/#post-2248738
https://forum-en.msi.com/index.php?threads/mag-x670e-tomahawk-wifi-%E2%80%94-weird-issue-with-expo-amber-light-stays-on-when-attempting-to-reboot.395642/#post-2252176
PS: always use F10 key to save bios settings that way it always registers the save.
If you ever screw it up and pc won't post then just reset bios (short the clear CMOS jumper or CMOS battery removal method) then you start again from beginning.

 

[gandolfo20315b302e9]

Alright. Keep in mind, the RAM timings may only apply to this kit CMH64GX5M2B6000Z30, i.e., Corsair Vengeance 2×32 GB, and not work for any other.

1. 1. Enter BIOS and load optimized defaults (F6)
2. 2. Save and reboot.
3. 3. Enter BIOS again.
4. 4. Set memory OC mode to EXPO and note down all timing values (values marked on the left, may differ on your side)
5. 5. Load defaults again.
6. 6. Start filling all timings with those noted down (replacing the unmarked values on the right)
7. 7. Set DRAM voltage to 1.4 V (important!)
8. 8. Enable Memory Context Restore and Power Down Enable.
9. 9. Reboot (DRAM LED may light up again for about 30–45 seconds, but it should boot normally after that)

Spoiler: For me, the following values are working now

Main Timing Configuration​
tCL	30
tRCD	36
tRP	36
tRAS	76
tRC	112
tWR	90
tRFC1	884
tRFC2	480
tRFCSB	390
tREFI	11677
Sub Timing Configuration​
tRTP	23
tRRDL	15
tRRDS	8
tFAW	32
tWTRL	30
tWTRS	8
Turn Around Timing Configuration​
tRDRDSCL	8
tRDRDSC	1
tRDRDSD	9
tRDRDDD	9
tWRWRSCL	23
tWRWRSC	1
tWRWRSD	9
tWRWRDD	9
tWRRD	8
tRDWR	20

I would also suggest lowering the SoC Voltage to 1.15 V as marked in the 3rd image.

---

For further tuning, I would also try setting PPT to 75 W and tweaking the curve via PBO Curve Optimizer. But I'd start with -5 or -10 and go down from there until it runs stable, since that part is down to silicon lottery as well. For me, -20 all-core works. Sometimes you'll have to increase SoC voltage again, but I'd try to be lower than 1.25 V.

Hi. I know this is an old post but I’m having this issue on new build and was hoping you could shed some light on this for me.

You say here to enable an EXPO profile and note all the values but when I turn on a profile the values don’t change in the bios at all, and once it is enabled and I reset I just get yellow light and stuck on boot, so how do I get those values if I cannot get back to the bios once EXPO is enabled. Am I misunderstanding something? Are you only noting the first 4 speed values that appear with the EXPO profiles or all the sub values as well?

Specs: b650 tomahawk wifi
Ryzen 5 7600x
T.Create DDR5 6000 Cl30 (2x16gb)

Runs perfectly fine at stock speeds (4800) but yellow light no boot when EXPO is enabled

Crossing my fingers someone sees this and can help lol I have such a headache trying to sort this out and MSI suggested swapping the CPU despite agreeing that it doesn’t appear to be a CPU issue. Already tried new RAM and that isn’t it either.

I know there’s a lot of info in this thread already but what I am confused about is how to get all those sub values for the EXPO profile. Only thing I can see is the first 4 values that show with each profile, all other values remain default.

 

[floepie0154002d9]

I know there’s a lot of info in this thread already but what I am confused about is how to get all those sub values for the EXPO profile. Only thing I can see is the first 4 values that show with each profile, all other values remain default.

If you enable EXPO from the landing page (button turns red), then the settings will be adjusted in MSI's OC section. There is a different area, where you can adjust the timings in the Settings area, and this is the so-called AEGIS section. I have the same board and chip you do, and I have found enabling EXPO in the AEGIS section much more stable. For example, I was not able to 'wake from standby' often if EXPO was enabled in the default location. This all may have been fixed in the most recent 2 or 3 BIOS revisions, or it could be that my memory controller on the CPU is not as good as others?

 

[gandolfo20315b302e9]

If you enable EXPO from the landing page (button turns red), then the settings will be adjusted in MSI's OC section. There is a different area, where you can adjust the timings in the Settings area, and this is the so-called AEGIS section. I have the same board and chip you do, and I have found enabling EXPO in the AEGIS section much more stable. For example, I was not able to 'wake from standby' often if EXPO was enabled in the default location. This all may have been fixed in the most recent 2 or 3 BIOS revisions, or it could be that my memory controller on the CPU is not as good as others?

Thanks for the reply!

This is where I get confused because when I turn on the EXPO profile and the button turns red all the values in the OC menu remain at default. I’ve taken some pictures to show this.

The values remain the same with the red button expo turn on. Am I looking in the right spot?

 

[citay]

First you have to press F10 to save and exit, then it will apply EXPO and do a process called memory training (which takes a bit longer than a normal POST), and if it passes memory training, the correct values should be listed in the BIOS. If it rejects EXPO, then you will find the same values as before (including the DDR speed).

 

[gandolfo20315b302e9]

Ahh I see, I guess this is the stage I have not gotten past. When I turn on the EXPO and F10 save and boot it doesn’t boot, yellow light is on and black screen. I’ve waited up to 10min and nothing shows and I have to short the CMOS to boot again. Maybe I’m just not waiting long enough

 

[citay]

Well, no, after about five minutes, i wouldn't really expect anything different to happen anymore, so ten minutes is already longer than necessary. Memory training can take a few minutes in some cases on AM5, but never ten. So it's safe to say that EXPO doesn't pass memory training, but in such a way that gets the BIOS completely stuck in place. This can happen sometimes. And on AMD-based MSI boards, the recovery is worse than on their Intel-based boards, where a simple press of the reset button usually makes the board boot to BIOS again.

What you can try here: Enable A-XMP instead of EXPO. For some reason, on MSI AM5, a few people had more success with the XMP (even though that is more meant for Intel) than with the EXPO profile. If that doesn't work either, then go back to EXPO, but before pressing F10, also set "DRAM Frequency" to something a bit lower, like DDR5-5600. It will then try to run that speed, but with the proper timings from the EXPO profile. From -6000 to -5600 wouldn't be that big of a difference, but much better than -4800.

 

[gandolfo20315b302e9]

Thanks for the back and forth, I appreciate it.

I tried the XMP and it still doesn’t boot. I also tried the EXPO profile with adjusted speed down to 5600 and same issue as well.

Whats odd is that in both these scenarios after hitting f10 the Pc would shut down completely then boot back up showing both red and yellow error lights. When before it wouldn’t shut down fully and would only show yellow light. Not sure what has changed there.

I’m still within return policy as it’s a new build so I’m considering just having the CPU replaced. Seems like such an odd thing though, no matter the adjust to ram speeds it just won’t boot anything over 4800

Thanks for trying to help!

 

[citay]

Don't replace the CPU over this, unless you have proven without a shadow of a doubt that it is responsible. Try to borrow some other DDR5 RAM kit for a test.

 

[gandolfo20315b302e9]

Ya I feel the same way, MSI suggested swapping the CPU but it seems so unlikely that it’s the problem since the system runs perfectly fine at stock settings.

I have tried other RAM and still had the same issue, albeit it was same brand and model that I tried. But the model I have is listed on the QVL. I want to try the full manual OC that people have suggested here but I don’t know what to set all those sub values to for my kit.

I have also tried turning on EXPO through the AMD overlocking menu that is found in the settings/advanced section will the same failure.