MPG X870E CARBON WIFI Beta BIOS

Svet

Svet

Well-known member
Administrator
Joined
Nov 20, 2003
Messages
83,839
Last edited:
Sort by date Sort by votes
After trying a different OC method, using CCD0 F lock and V lock, no PBO. Getting BIOS boot code A6 on windows restarts... Tried to fix this via manual M2_1 PCIE 5.0 setting.

Also the RAM kit I have fails to report its timings or speed correctly, until I manually enter the timings. The 6000 -> 6200mhz change does not even show up in Zentimings, but in BIOS it shows 6200... so not sure which to believe.

Reverting the 6200 -> 6000mhz RAM and 2067 -> 2000 FCLK did not fix the A6 boot code... I did upgrade the chipset drivers for AMD from what I have listed below to the "AMD_Chipset_Software_7.04.09.545" versions. Most of which were already up to date.
 
Last edited:
Upvote 0 Downvote
Blackbird7 said:
Everyone writing about CO -20 or even lower, but I managed to get my 9800X3D stable on +200 boost only with CO per core, in range 0 to -12. I guess I had very bad luck with getting so bad silicone. Maybe I could get more from it with lowering boost, I see you have +150, or play with curve shaper. But when I tried that, I just made it worse.
I'll do bit more tuning and hope new BIOS will be available soon and get me some improvements.
This is current config... It looks like FCLK 2133 was causing some rare freezes, after I reduced it to 2066, mem benchmark is slightly worse (2%), but I did lower some timings and that reduced latency about 2%.
View attachment 202693
Click to expand...
You are running C32 RAM at C28, even with the voltage boost the chips might not be stable with the 9800x3d's IMC. It seems to be very sensitive, and dropping the voltage via CO can quickly cause instability.
 
Upvote 0 Downvote
juan.delgado.riguera152d02cc said:
Sure, this is the SOC voltage I'm referring to:
View attachment 202635

I'm using PBO as well indeed, with scalar x5, +200 boost clock and negative 33 all core curve.
Click to expand...
I was having the same problem and checked my SoC voltage, it was in Auto and the damn BIOS was delivering 1.301 volts. I have decreased it to 1.25 manually and now it seems the problem is fixed and everything is stable...

I'm starting to think this BIOSes have a bad voltage regulation algorithm and the problem is caused by having the "auto" enabled, something like Intel's past gen problems...
 
Upvote 0 Downvote
-30 CO +200 flck 2200 ram @6400 CL 32 ,Im not using additional Curve Shaper and this about max I can go with my system tuning wise. I see people OC much higer and saying 100% stable for me if I touch the bus even 102 everything goes "tits up" games start stuttering and all kind of weird crap. Trying to fully manually overclock the 9800X3D is overly complicated, time-consuming, and doesn't seem to have any real-life effect only rising Temp and power unnecessarily.
 

Attachments

  • Zen.jpeg
    Zen.jpeg
    97.9 KB · Views: 48
Upvote 0 Downvote
Alyred said:
You are running C32 RAM at C28, even with the voltage boost the chips might not be stable with the 9800x3d's IMC. It seems to be very sensitive, and dropping the voltage via CO can quickly cause instability.
Click to expand...
Yes, it is 6400 C32, but with that EXPO I could only get it to boot, any stress test would quickly fail. Like I said, 6200 C28 worked very stable on old BIOS. I guess something in newer BIOS changed or I messed up my notes about old config. Third option is that now is much warmer in room, and even with added fan on sticks, it's not enough, I have seen now I do get close to 55C on long tests, so I'll have to reduce voltage, even if it means changing timings a bit more. Just now I'm running stress test with VDD=1.43 and VDDQ/VDDIO=1.3 and so far it's good.
 
Upvote 0 Downvote
Blackbird7 said:
Yes, it is 6400 C32, but with that EXPO I could only get it to boot, any stress test would quickly fail. Like I said, 6200 C28 worked very stable on old BIOS. I guess something in newer BIOS changed or I messed up my notes about old config. Third option is that now is much warmer in room, and even with added fan on sticks, it's not enough, I have seen now I do get close to 55C on long tests, so I'll have to reduce voltage, even if it means changing timings a bit more. Just now I'm running stress test with VDD=1.43 and VDDQ/VDDIO=1.3 and so far it's good.
Click to expand...
Yeah, I have 6400CL32 as well (though corsair Dominator Titanium rather than g.skill) and couldn't get it stable at 6400 either. 6200CL30 runs great and stable though, and was able to speed things up a bit more my lowering TRFC and raising TREFI to Hynix A-die safe values. CL28 significantly increases instability if the chips aren't rated for it, and combined with the 9800's IMC being SUPER sensitive can make stability plummet really quick when voltage overall changes. Since I require hypervisor getting latency lower is better (to make up for the penalty that having Hypervisor enabled incurs), but I also need solid, long-term stability.
55 degrees isn't spectacular but it's nowhere near extreme temp limits either. If you're stable there you're probably good so long as your processor can hold high boost clocks and runs stable. If it doesn't go over 60 1.43v is probably fine (and I believe JDEC standards say the maximum is like, 90 degrees).
Most of the BIOS updates have changed base voltages in the processor, so several of us have noticed changes in what we can do with curve optimizer, and it seems that has a large effect on the integrated IMC being stable with faster memory. Your observations on the lower CO maximum doesn't surprise.
 
Upvote 0 Downvote
I think lot of my problems happened due to DDR temp going too high and having maxed TREFI. Now with fan and adjusted voltage, temp is not going over 51 °C, and no more such problems (but still lot of tests to do). With curve shaper I managed to lower only Max frequency values (results in slightly higher single thread benchmarks), if I touch mid or high, stress fails. Maybe I have to recheck CO per core values...
 
Upvote 0 Downvote
Blackbird7 said:
I think lot of my problems happened due to DDR temp going too high and having maxed TREFI. Now with fan and adjusted voltage, temp is not going over 51 °C, and no more such problems (but still lot of tests to do). With curve shaper I managed to lower only Max frequency values (results in slightly higher single thread benchmarks), if I touch mid or high, stress fails. Maybe I have to recheck CO per core values...
Click to expand...
Ahhh, remember when building a new PC was just putting everything together, enabling XMP, and running burn in tests to make sure it wasn't going to crash under load?
Those were the days...
 
Upvote 0 Downvote
Alyred said:
Ahhh, remember when building a new PC was just putting everything together, enabling XMP, and running burn in tests to make sure it wasn't going to crash under load?
Those were the days...
Click to expand...
I remember the days when I could build a high-end gaming PC with all peripherals and a monitor for less than the price of a 5090.
 
Upvote 0 Downvote
lysyn12155d02d4 said:
Combo AM5Pi 1.2.0.3e beta and official 1.2.0.3d bios have appeared on the Asus website.
Click to expand...
From reading the Asus forums, it sounds like 1.2.0.3d has issues. Prob why they already have 1.2.0.3e in beta. I just really want MSI to fix the pcie slot booting below gen 5... It works fine with spread spectrum on for me, but I shouldn't have to enable it just for that.
 
Upvote 0 Downvote
Indabox said:
From reading the Asus forums, it sounds like 1.2.0.3d has issues. Prob why they already have 1.2.0.3e in beta. I just really want MSI to fix the pcie slot booting below gen 5... It works fine with spread spectrum on for me, but I shouldn't have to enable it just for that.
Click to expand...
I don't know what you're talking about, but I've never experienced this PCIe generation problem on my X870E Carbon WiFi. I have a mix of Gen 5 and Gen 4 SSDs along with a 5090, and everything works fine. So, take it with a grain of salt when you say "FIX," because from my point of view, there's nothing to fix in that regard. In my case Spread Spectrum is OFF maybe you live in EM congested area?
Which makes me think, in fact, it works fine for you with Spread Spectrum ON. Keep it on then; that's why it's there in the first place. What do you want them to "fix"?
 
Last edited:
Upvote -1 Downvote
DarkOmenX said:
I don't know what you're talking about, but I've never experienced this PCIe generation problem on my X870E Carbon WiFi. I have a mix of Gen 5 and Gen 4 SSDs along with a 5090, and everything works fine. So, take it with a grain of salt when you say "FIX," because from my point of view, there's nothing to fix in that regard. And In my Spread Spectrum is OFF maybe you live in EM congested area?
Which makes me think, in fact, it works fine for you with Spread Spectrum ON. Keep it on then; that's why it's there in the first place. What do you want them to "fix"?
Click to expand...
I didn't have any issues for months with the PCI lane speed locking. Then it just started, randomly, without changes in my environment and my house is basically a faraday cage. Suddenly my graphics card started getting "stuck" at PCIe 2.0 and not being able to boost back to 5.0, and the entire system goes to crap.

Installing my brand-new 9950x3d yesterday, it immediately did the same thing. Once I turned on Spread Spectrum, and the system runs as smooth as butter again. Better than with my 9800x3d, I'm hitting speeds and stability on memory that I've never seen with my 9800x3d.
Upgraded to the A41 Beta and it's been solid otherwise so far, aside from the PCI speed locking on the GPU PCIe. I think there's still some problem as it is pretty random, and evidently other vendors have figured it out.
 
Upvote 0 Downvote
Alyred said:
I didn't have any issues for months with the PCI lane speed locking. Then it just started, randomly, without changes in my environment and my house is basically a faraday cage. Suddenly my graphics card started getting "stuck" at PCIe 2.0 and not being able to boost back to 5.0, and the entire system goes to crap.

Installing my brand-new 9950x3d yesterday, it immediately did the same thing. Once I turned on Spread Spectrum, and the system runs as smooth as butter again. Better than with my 9800x3d, I'm hitting speeds and stability on memory that I've never seen with my 9800x3d.
Upgraded to the A41 Beta and it's been solid otherwise so far, aside from the PCI speed locking on the GPU PCIe. I think there's still some problem as it is pretty random, and evidently other vendors have figured it out.
Click to expand...
It might be the frequency clock generator causing more spikes than when it was brand new, possibly due to "overly" overclocking the system and pushing it hard, leading to some degradation. Now you might need to use Spread Spectrum to keep the spikes under control, or it could simply be component degradation due to lower quality.
However, what you state, I don’t see anything for MSI to fix, and I’m not very MSI "protective." On the contrary, I’m very critical and bashing. I have no sympathy for BS, but in your case, it actually seems to be just fine.
Enjoy your system and who the F cares if Spread Spectrum is ON or OFF. If it works without it fine if it has to be enabled due to stability reasons fine again that's exactly the purpose of it.
 
Upvote -1 Downvote
Alyred said:
I didn't have any issues for months with the PCI lane speed locking. Then it just started, randomly, without changes in my environment and my house is basically a faraday cage. Suddenly my graphics card started getting "stuck" at PCIe 2.0 and not being able to boost back to 5.0, and the entire system goes to crap.

Installing my brand-new 9950x3d yesterday, it immediately did the same thing. Once I turned on Spread Spectrum, and the system runs as smooth as butter again. Better than with my 9800x3d, I'm hitting speeds and stability on memory that I've never seen with my 9800x3d.
Upgraded to the A41 Beta and it's been solid otherwise so far, aside from the PCI speed locking on the GPU PCIe. I think there's still some problem as it is pretty random, and evidently other vendors have figured it out.
Click to expand...
I'm currently in a "reboot hell", my computer (windows) crashes many times in a raw (about 5-30 minutes random intervals), then it's ok for a few hours and it crashes again randomly again some time to time (seems to be on writing operations as it often corrupts app files).
The last two beta bios made it worse as spread spectrum or forcing gen4 don't seems to stabilize anything anymore (even if it was not perfect before either).
I have a 9800x3d and a gen5 crucial T705 (which seems to be a bad combo) and a geforce 3080 (for now).
The result is always the M2_1 getting "disconnected" sometimes even changing some settings in the bios and rebooting is enough to lose it (it quite recent for this one).
I tried another crucial T705 (under linux this time) and got the same issue.
All self tests of my NVMEs are ok.
Since the crucial compatibility is from gen3 to gen5, my guess is as soon a the PCIe regresses under gen3, the NVME get disconnected (and it's only my guess, I'm no expert).
Also, I'm not into OC, so I basically use the default settings.
I certainly won't buy anything MSI for a very long time (and certainly not the GC), it's been monthes with my expansive piece of junk now...
 
Last edited:
Upvote 0 Downvote
Yeah, trying to shrug off the PCI stutter issue like it's a user problem, having to do with unstable / 'hard' overclocks or 50 other things doesn't help, because it happens on stock configurations, happens with all the supposed 'fixes' presented and I've seen 10+ people on this forum alone report it. There's clearly something going on underneath but people are just trying to locate a commonality between them and MSI have said they are aware of it.

After I reset / update BIOS, running completely stock, the prevalence of it happening jumps up like 500% - So it has nothing to do with my conservative overclocks.

Though my latest test was setting manual voltage as suggested. Never had any problems with auto voltage, kept CPU within safe limits and performance has been good. Now I set manual voltage about a week ago and haven't seen the PCI stutter bug so far, though even before a week could easily pass without issue.

Funny thing is, when I set the manual voltage and saved, it kept doing the stutter bug. Every single restart. Then I had to shut it off completely, wait and then it was fine. So we'll see if this is the magical bullet.
 
Upvote 1 Downvote
Sethos812d002b2 said:
Yeah, trying to shrug off the PCI stutter issue like it's a user problem, having to do with unstable / 'hard' overclocks or 50 other things doesn't help, because it happens on stock configurations, happens with all the supposed 'fixes' presented and I've seen 10+ people on this forum alone report it. There's clearly something going on underneath but people are just trying to locate a commonality between them and MSI have said they are aware of it.

After I reset / update BIOS, running completely stock, the prevalence of it happening jumps up like 500% - So it has nothing to do with my conservative overclocks.

Though my latest test was setting manual voltage as suggested. Never had any problems with auto voltage, kept CPU within safe limits and performance has been good. Now I set manual voltage about a week ago and haven't seen the PCI stutter bug so far, though even before a week could easily pass without issue.

Funny thing is, when I set the manual voltage and saved, it kept doing the stutter bug. Every single restart. Then I had to shut it off completely, wait and then it was fine. So we'll see if this is the magical bullet.
Click to expand...

Unfortunately, this board requires a 9900 series CPU. Perhaps MSI set it up like this! I only realised this much later. I used it with a 7950x3D for about six months and encountered many problems. After switching to a 9950x3D, however, all the problems disappeared. :scratchchin:
 
Upvote 0 Downvote
interesting read over by GURU
"MSI has issued a BIOS update for select AM5 motherboards that brings support for upcoming AMD processors. This update is based on the AGESA PI 1.2.0.3e firmware, which provides essential enhancements to system compatibility and performance. Among the key improvements is the ability to support four 64GB memory modules, allowing systems to reach up to 256GB of RAM. This capacity is important for users who require large memory configurations for demanding workloads. The update also focuses on improving memory performance, particularly in a 2DPC 1R configuration. This means that when two DIMMs are installed per memory channel with single-rank modules, performance and stability are optimized. Additionally, MSI has worked on refining overclocking support specifically for Samsung 4G×8 memory chips, which are commonly used high-density DRAM modules. These improvements should help users achieve better stability and higher frequencies when pushing their RAM beyond stock settings....."
www.guru3d.com

MSI Releases BIOS Update AGESA 1.2.0.3e for AM5 Motherboards Supporting Future AMD CPUs

MSI has issued a BIOS update for select AM5 motherboards that brings support for upcoming AMD processors. This update is based on the AGESA PI 1.2.0.3e firmware, which provides essential enhancements to system compatibility and performance.
www.guru3d.com www.guru3d.com
 
Upvote 0 Downvote
Back
Top