Issue with Asus Prime x670-E-Pro Wi-Fi boot during cold start.
Issue with Asus Prime x670-E-Pro Wi-Fi boot during cold start.
Here’s a revised version of your text:
I really appreciate the thorough explanation you provided—it’s very helpful!
Currently, I have the EXPO configuration on my DRAM, and the specific RAM in question is this one:
https://www.gskill.com/product/165/390/1...6GX2-TZ5NR
As far as I remember, I was very cautious when selecting the motherboard to DRAM compatibility. From what I know, this memory is listed as compatible with that particular board. Regarding the CPU, I don’t think the pairing is crucial, since marketing suggests the MB is meant for the 7000 series with DDR5 support. The x3d version was released about a year ago, so it likely had the necessary support at the time.
I believe my decision to let Asus handle the EXPO settings was too hands-off. Looking around, I think I’d need to use Ai Tweaker and pick the profile there. Even G.Skill has screenshots in their FAQ on how to configure it.
Just a quick reminder, the cold boot problem only showed up after a major BIOS update. Before that, I didn’t encounter any issues with the simple setup process. Maybe the system wasn’t as sensitive then?
In summary, based on our conversation, the cold boot issue seems DRAM-related. If it persists, I plan to perform a BIOS reset—this could also affect the EXPO settings. That’s the strategy for Thursday; stay tuned!
It is mentioned that you possess an EXPO configuration. Even though your system includes several AMD CPUs, I personally do not have any EXPO memory installed. When updating the BIOS, certain adjustments from the previous version may be lost. For instance, I often notice that the BIOS update disables "Virtualization," which is necessary for Hyper-V to function. After such an update, it's recommended to restore system defaults and aim for a consistent first boot. You may need to adjust the default boot drive if your machine has multiple bootable partitions. Additional configurations could involve choosing a different GPU card instead of the integrated one, turning off the full-screen logo, enabling third-party ROMs, or modifying fan speed profiles—though these should not compromise stability. In your situation, it appears this hasn't occurred yet; if you still encounter issues with a cold boot, consider whether you're pushing your memory to 6000MT/s. Temporarily suspend EXPO and revert the RAM to its JEDEC standard setting (likely 4800MT/s). Power on and off the system several times, then disconnect the power supply to ensure a proper cold start next time. Test for stability and observe if the problem lies with the memory overclock. If you're comfortable tweaking memory parameters manually—adjusting CL/CAS values in small steps—you might identify the cause. Ryzens often spend time in new builds or after BIOS updates trying to recalibrate DDR5 memory. If successful, the system may stabilize after a few sessions as the BIOS fine-tunes settings.
Unless you're skilled in manual memory adjustments to fine-tune Primary, Secondary, and Tertiary timings by hand, it's advisable to let the BIOS handle these configurations automatically from the SPD menu on the DIMMs. Many issues stem from excessive XMP/EXPO overclocks, while others arise from mismatched CPU, RAM, or motherboard combinations.
You've made thoughtful choices with your RAM and EXPO settings. Others have encountered stability problems with DDR5 modules in AM5 platforms by using Corsair Vengeance DDR5 without EXPO tables. Check online forums for similar setups to see if others face comparable challenges.
Hello Everyone,
Thank you for staying engaged throughout this process and sharing your knowledge.
Today I performed a cold boot and it appeared the computer managed to turn on successfully from cold start, which aligns with the 1 in 20 probability mentioned earlier.
After restarting and setting the system to default 4800MT/s, there were no further problems with Cold Boot.
For each day, I plan to continue cold booting the machine as usual when the issue began, but with these adjustments:
I’m uncertain whether it was better to adjust the Memory Frequency directly without considering timings or voltages, so I followed this approach instead:
Loaded the EXPO profile:
Memory Frequency: Auto → 6000 MHz
Tcl: Auto → 30
Trcd: Auto → 38
Trp: Auto → 38
Tras: Auto → 96
Dram VDD Voltage: Auto → 1.35
Dram VDDQ Voltage: Auto → 1.35
PMIC Voltages: Auto → Sync All PMICs
Memory VDD Voltage: Auto → 1.35
I hope this matches your suggestions, Misgar? If not, I’m unsure about the consequences of changing timings and voltages.
Then I lowered the Memory Frequency to 5400MHz, matching the system’s intended operation.
Based on how the cold boot occurs, I’ll adjust it in 200Mhz steps until I find a stable setting.
There was one Reddit thread mentioning an issue with Asus MB boards and BIOS 3057, but the settings varied widely—some users reported stability problems with older BIOS versions, while others didn’t encounter issues after updating.
Lowering the clock speed to 5400MT/s while maintaining the faster 6000MT/s settings will simplify operations for the IMC channels within the CPU. It may not be the approach I initially took, but stability is what matters most.
It seems your system experienced instability at a slower 4800MT/s, but functioning at 5400MT/s is acceptable. My Kingston Fury DDR5 DIMMs offer two configurations at 4800MT/s: one with the standard non-XMP setting and default RAM voltage of 1.20V, and another with XMP (still at 4800MT/s) but using a higher voltage of 1.35V. This additional 0.15V enhances stability during overclocking.
In my DDR5 RAM, I found it unusual to have two settings at the standard 4800MT/s, since extra voltage isn’t typically required, but I’m not an expert on memory or BIOS details. Usually, the JEDEC default frequency for DDR5 (around 4800MT/s) remains stable at 1.20V. Instability tends to appear only when I enable higher frequency XMP overclocks (without EXPO on any of my DIMMs), even with increased voltage. At that point, I reduce the RAM speed until it stabilizes.
It might have been beneficial for me to update the BIOS on both Asus boards—3800X using DDR4 and 7950X with DDR5—to see if it improves performance. I don’t use PBO overclocks on either system for long-term stability during extended rendering tasks.
On a different note, I assume you’re using Windows 11 (or possibly 10)?
You might want to turn off Windows “Fast Startup” and test whether that helps. It’s unlikely to directly fix your memory issues, but it could ease boot problems. Fast Startup saves certain system settings to a file on the boot drive (like hiberfil.sys), which theoretically speeds up future boots. However, this can interfere with multi-drive setups where multiple Windows versions are stored on separate drives. I’ve therefore disabled Fast Startup.
https://www.windowscentral.com/soft...-e...windows-11
https://www.windowscentral.com/how-disab...st-startup
When installing Windows 10, I turn off all power-saving features: screen saver, disable disk shutdown after inactivity (after 20 minutes), power down USB hubs, Ethernet chipsets, WiFi chipsets, network controllers, etc.
I remove the hiberfil.sys file from all desktop PCs, even if it contains only one boot drive, but keep hiberfil enabled on laptops.
https://www.howtogeek.com/885071/how-to-...indows-11/
Even if I’m struggling, disabling Fast Startup could lead to more consistent launches. You can always re-enable hiberfil and fast startup later if needed.
Just to confirm, you meant to say that the fail-to-boot screen never appeared, even though the system seemed stable at default settings. That day I got a bit excited and went beyond just using base settings. I also felt a bit frustrated because my system couldn't seem to manage the advertised EXPO profile now.
Today I performed a cold boot at 5600MHz, but it didn’t start up properly, so I’m reverting to 5400MHz. Earlier, I did a cold boot at 5600MHz and it worked fine.
I promise it’s been the most inconsistent boot I’ve ever had.
I’m currently using Windows 10 and am worried about upgrading to Windows 11 next August. Following your advice, I disabled Fast Start-up. However, if I hadn’t checked the link you shared, I might have assumed the issue was related to the BIOS Fast Boot setting. Maybe it’s worth experimenting with that now too?
Overall, at this point I’ve lowered my RAM speed back to 5400MHz and turned off Fast Start-up.
I've finally got around to checking the officially supported maximum memory speed for the 7800X3D in CPU World and it's DDR5-5200.
https://www.cpu-world.com/CPUs/Zen/AMD-Ryzen 7 7800X3D.html
This doesn't mean you (or other people) can't overclock RAM faster than 5200MT/s on a 7800X3D, but 6000MT/s may not be easily achievable with this particular CPU, unless you're experienced at manual tweaking, or win the "Silicon Lottery".
Just because you buy memory rated at DDR5-6000 doesn't mean you'll be able to achieve 6000MT/s with all combinations of CPU and motherboard. In general, DDR5 tends to clock faster on Intel than on AMD. You'd probably stand a much better chance of an easy 6000MT/s XMP overclock with a modern Intel CPU.
When I'm overclocking RAM, be it DDR3, DDR4 or DDR5, I check the various timings in the SPD table (a small chip on each DIMM) using AIDA64 (Motherboard, SPD option). I then check the actual timings set by the BIOS in AIDA64 (Motherboard, Chipset option) to determine which set of values the BIOS has pulled out of SPD for the specific XMP overclock speed I requested.
Usually there's a very close (or identical) match between a set of values in the SPD table and the actual BIOS values, which helps to confirm that "what you've asked for" is "what you've got".
My suspicion is that if you've selected the 5400MT/s EXPO or XMP setting, the BIOS will have chosen the most appropriate timings out of SPD for that speed, i.e. you'll be using "tighter" 5400MT/s timings, not "looser" 6000MT/s timings.
It's a tricky concept to get to grips with, but AIDA64 and other programs can help you to puzzle things out. It's easy to check. At lower speeds, timings will appear as a set of smaller figures, with fewer (slower) clock cycles. As you increase the memory speed, timings will appear as larger numbers, with more (faster) clock cycles per second.
On the 5600X in front of me now, DDR4-2400 SPD Primary timings are 12-15-15-28, but DDR4-3600 SPD timings are 18-22-22-48. The numbers for DDR5 are bigger, but the clock frequencies are twice as fast, e.g. 2400MT/s is a typical JEDEC speed for DDR4, but 4800MT/s is a typical JEDEC speed for DDR5.
Since the 7800X3D officially supports only 5200MT/s, I'd be inclined to call it a day and stick to 5400MT/s if it's stable, unless you feel like exploring manual tweaking. You're already 200MT/s above the "guaranteed" maximum, but I'm sure more is possible in expert hands. Put a Liquid Nitrogen pot on the CPU and you might get 7000MT/s. Even faster with Liquid Helium if Ryzens can operate down towards Absolute Zero. I've not tried extreme cooling on CPUs, but I have used LN2 at work.
https://www.storagereview.com/revie...d-...000-series
I doubt it'll make any difference. All that switching off Fast Boot in the BIOS does is to spend a little longer checking things like RAM. Still, there's no harm in trying.
You probably won't benefit from switching off Windows Fast Startup either, but who knows?
True, but during my initial investigation I examined the parings and through the two AMD systems I've used (this being my second one), the EXPO profile functioned at the rated speed once the EXPO was activated.
I suppose that’s why my surprise emerged, since I hadn’t anticipated that RAM which had already operated smoothly at 6000 MHz would eventually become unstable after a BIOS update, and only in this particular case (warm boot didn’t cause any problems). It’s disappointing to miss out on the gaming performance I had invested in, but ensuring the computer starts reliably is more important. (I appreciate the adaptability of a PC, but I have some basic experience in assembling and getting them to work.)
I concur with your view about simply concluding this task. EDIT: no, the instability issue has now appeared at 5400 MHz, so I reduced it to 5200MHz. Interestingly enough, the BIOS resolution was different this time, and Fast Start-up was turned back on.
At this point, it seems like we’re just trying out various settings. Although I prefer disabling Fast Start-up because it makes shutdowns smoother—adding about three seconds to the Windows startup screen—I’ve noticed the problem now. I restored it to normal.
Reply from Asus:
"BIOS Flashback button to flash the BIOS again on the motherboard. Fail no post.
If the motherboard BIOS is flashed and it persists, report the issue."
If I were to interpret this, they suggest using the non-bios method to update the existing BIOS version again. This implies just connecting a USB and seeing the system status light for success or failure. I’m not very sure about this outcome. They also recommend sending the board back for replacement, but I don’t think it’s a reliable fix since the problem seems to be with the BIOS itself, not my actions.
Looking online, it looks like the CMOS button is needed next. Otherwise, I won’t touch the BIOS and will have to work with the current memory speeds until possibly after a few BIOS updates."