MPG X870E CARBON WIFI Beta BIOS

Sort by date Sort by votes
Yeah, haven't been a positive experience for me. Between the stuttering and my Roland Bridge Cast audio crackle on this board, I'm honestly contemplating ditching it.
 
Upvote 0 Downvote
I was able to fix the stuttering by setting the power management setting in windows to balanced and specifically the pci express setting set to maximum power savings.
 
Upvote 0 Downvote
RobsonByte said:
I upgraded, changed my 9950x to 9950x3D. I cleared the CMOS, installed the a29 BIOS. Every time the PSU is disconnected from the mains power, the computer does not turn on (error C5, OA, and Ab). To be able to turn it on, I have to press the Safe Boot button on the motherboard a few times, until the OA code appears, then wait almost 2 minutes to start training the memories. I decided to change a setting in the memory and it seems to have worked (activate Power Down, as it was disabled). This behavior was not observed with the 9950x processor and happened with my 9950x3D.
After some tests I realized that the processor doesn't show all its strength, because I can only get 890 points in single core on CPU-z, while some get 909. The same thing happens in CB23, I only get 45500 points and some get 46800. Maybe it's some limit that doesn't let the EDC go up, nor the PPT. While I only spend 230w, other people have better performance and use up to 300w.
Edit: By pressing the "Smart Button" on the back of the motherboard, I can enter "Safe Boot" mode. It's awful having to do all this ritual to be able to boot.
Click to expand...
Ugh, that does sound frustrating. I'm considering upgrading to a 9950x3d myself but have held off until I get this board working perfectly with the 9800x3d. I'm doing some production workloads but not enough to REALLY see the benefit just yet, and it's more important than I upgrade from my 3090 - once friggin' GPUs actually come to market in enough quantities that I'm not paying a scalper.
Give dropping your RAM down to 6000MT/s a try. The IMC on the x3d processors is MUCH more sensitive to frequency, it appears, but doesn't NEED as fast RAM with the 3d vCache. Until I scaled my 8000 MT/S RAM back manually, I was seeing a LOT of this sort of bug on startup/restart/etc.
 
Upvote 0 Downvote
sgmwright said:
I was able to fix the stuttering by setting the power management setting in windows to balanced and specifically the pci express setting set to maximum power savings.
Click to expand...
Setting the power management setting to Balanced does improve performance, specifically on dual CCD chips (so the scheduler works properly), but setting the PCI Express P/S to "Maximum" does seem like an odd setting for performance...
 
Upvote 0 Downvote
Zippos said:
Seems official A30 Bios is up but under the wrong name A3. Inside the zip file, the BIOS file name extension is A30 as expected. Anyone tried it?
Click to expand...
That is interesting. I hadn't noticed any suspicious CPU information being incorrect on my 9800x3d, so I suspect that it's mostly for the 9900x3d and 9950x3d's. Other patch notes seem identical to A27. Haven't tried A30 yet, but A29(beta) seems really solid on my test machine. Still going through testing on mu per-core undervolting but maybe I'll upgrade in the meantime just to get it on an "release" version.

No idea if there was anything addressed with what some folks are seeing, with the E1 PCIe socket getting stuck at lower versions than it should be, or if there are any speed increases on M2_1 and M2_3. I'll test the M2 slots and get back to everyone if there's any difference there, but A29 was actually a slight regression on that front.
 
Upvote 0 Downvote
Alyred said:
Ugh, that does sound frustrating. I'm considering upgrading to a 9950x3d myself but have held off until I get this board working perfectly with the 9800x3d. I'm doing some production workloads but not enough to REALLY see the benefit just yet, and it's more important than I upgrade from my 3090 - once friggin' GPUs actually come to market in enough quantities that I'm not paying a scalper.
Give dropping your RAM down to 6000MT/s a try. The IMC on the x3d processors is MUCH more sensitive to frequency, it appears, but doesn't NEED as fast RAM with the 3d vCache. Until I scaled my 8000 MT/S RAM back manually, I was seeing a LOT of this sort of bug on startup/restart/etc.
Click to expand...
I reduced it to 6000Mts, same behavior. Then I swapped the slot memories, one for the other. First boot was good, I will continue testing.
 
Upvote 0 Downvote
I updated to the new A30 from the original 1A10 today. Cleared CMOS before BIOS update and did the same after the BIOS update before i started making any changes.

I lost a lot of undervolt stability... I was stable at -30 on old BIOS. Tried -20 and windows froze while booting. Ended up stable at -15
Memory EXPO setting took without issue (6000mhz CL 30-36-36-80 @ 1.4v)
Not having any stuttering issues
Wi-Fi, Bluetooth, etc.. seem stable so far
No issues so far with my 5070 Ti and from what I'm reading that's a miracle in itself..
When running Cinebench r23 @ -15 undervolt, CPU VDDCR_SOC hit max 1.190v and hit max temp of 86.5c
While gaming, temps are much lower.. not seeing anything higher than 60c max

EDIT:
Cleared CMOS 3 more times using the button on the MB just to make sure the BIOS update didnt hang onto any old crap.. and low and behold im back to a stable -30 undervolt. Max temps in Cinebench r23 dropped a few more from 86.5c to with -15 to 83.2c with -30 undervolt. sitting at 39c idle

System:
9800X3D (using Thermalright Phantom Spirt 120 SE Air cooler)
MSI x870e Carbon
Kingston Furry Beast CL30 Ram (A-die) - 6000mhz CL 30-36-36-80 @ 1.4v
Asus Tuf 5070 Ti OC
2tb Samsung 990 Pro (NVMe 1), 2tb Samsung Evo Pro (NVMe 3), 4tb Samsung Evo Pro (NVMe 4) - Nothing in NVMe slot 2 to avoid GPU lane sharing
Corsair RM1000x

CrystalDiskMark - NVMe slot 1 with Samsung 990 Pro:
Read (MB/s)Write (MB/s)
SEQ1M / Q8T17387.356910.88
SEQ128K / Q32T17363.956834.77
RND4K / Q32T165739.495514.81
RND4K / Q1T175.97253.66
 
Last edited:
Upvote 0 Downvote
Indabox said:
I updated to the new A30 from the original 1A10 today. Cleared CMOS before BIOS update and did the same after the BIOS update before i started making any changes.

I lost a lot of undervolt stability... I was stable at -30 on old BIOS. Tried -20 and windows froze while booting. Ended up stable at -15
Memory EXPO setting took without issue (6000mhz CL 30-36-36-80 @ 1.4v)
Not having any stuttering issues
Wi-Fi, Bluetooth, etc.. seem stable so far
No issues so far with my 5070 Ti and from what I'm reading that's a miracle in itself..
When running Cinebench r23 @ -15 undervolt, CPU VDDCR_SOC hit max 1.190v and hit max temp of 86.5c
While gaming, temps are much lower.. not seeing anything higher than 60c max

EDIT:
Cleared CMOS 3 more times using the button on the MB just to make sure the BIOS update didnt hang onto any old crap.. and low and behold im back to a stable -30 undervolt. Max temps in Cinebench r23 dropped a few more from 86.5c to with -15 to 83.2c with -30 undervolt. sitting at 39c idle

System:
9800X3D (using Thermalright Phantom Spirt 120 SE Air cooler)
MSI x870e Carbon
Kingston Furry Beast CL30 Ram (A-die) - 6000mhz CL 30-36-36-80 @ 1.4v
Asus Tuf 5070 Ti OC
2tb Samsung 990 Pro (NVMe 1), 2tb Samsung Evo Pro (NVMe 3), 4tb Samsung Evo Pro (NVMe 4) - Nothing in NVMe slot 2 to avoid GPU lane sharing
Corsair RM1000x

CrystalDiskMark - NVMe slot 1 with Samsung 990 Pro:
Read (MB/s)Write (MB/s)
SEQ1M / Q8T17387.356910.88
SEQ128K / Q32T17363.956834.77
RND4K / Q32T165739.495514.81
RND4K / Q1T175.97253.66
Click to expand...
How about your cinebench scores are they on par with prior version? I do ask because I m used to see a little regression with each bios release , on single and multithreaded benchmarks.
 
Upvote 0 Downvote
Alyred said:
Setting the power management setting to Balanced does improve performance, specifically on dual CCD chips (so the scheduler works properly), but setting the PCI Express P/S to "Maximum" does seem like an odd setting for performance...
Click to expand...
it doesn't affect performance, I was getting "hitching" in Windows just watching movies, idk if it was because I had my power setting to Maximum or what but I read that the pci-setting to minimum stops the hitching. So I did both of that and it stopped.
 
Upvote 0 Downvote
RobsonByte said:
I reduced it to 6000Mts, same behavior. Then I swapped the slot memories, one for the other. First boot was good, I will continue testing.
Click to expand...
I had no progress, because on the third boot the problem returned. So if I disconnect the computer from the power and turn it back on after 3 minutes, for example, the boot works, but if I wait about 2 hours to turn it on, the boot doesn't happen (error OA or 15). I'm suspecting it's the pressure in the socket, but if I ease the pressure a little more, I get error C5.
 
Upvote 0 Downvote
On another note, just FYI, if you're using A29 along with AMD overclocking's "Auto Voltage" setting while overclocking the CPU with BCLK, pay close attention to the voltage set by the motherboard.
To my great surprise, I saw a whopping 1.394V at 4911 MHz (104.5 BCLK), something I never saw with any prior BIOS versions A21 - A26, where the voltage maxed out around 1.25-1.31V
 
Upvote 0 Downvote
vic123159d02e5 said:
On another note, just FYI, if you're using A29 along with AMD overclocking's "Auto Voltage" setting while overclocking the CPU with BCLK, pay close attention to the voltage set by the motherboard.
To my great surprise, I saw a whopping 1.394V at 4911 MHz (104.5 BCLK), something I never saw with any prior BIOS versions A21 - A26, where the voltage maxed out around 1.25-1.31V
Click to expand...
Good info to watch for, thanks. I played around with BCLK a little on my test system and it seemed stable on A29, but didn't push anything past 101. Went back to base 100 while testing per-core CO. Are you able to stabilize it by fixing the VCORE to 1.31V or is it still bouncing around when testing?

Indabox said:
I updated to the new A30 from the original 1A10 today. Cleared CMOS before BIOS update and did the same after the BIOS update before i started making any changes.

I lost a lot of undervolt stability... I was stable at -30 on old BIOS. Tried -20 and windows froze while booting. Ended up stable at -15
Memory EXPO setting took without issue (6000mhz CL 30-36-36-80 @ 1.4v)
Not having any stuttering issues
Wi-Fi, Bluetooth, etc.. seem stable so far
No issues so far with my 5070 Ti and from what I'm reading that's a miracle in itself..
When running Cinebench r23 @ -15 undervolt, CPU VDDCR_SOC hit max 1.190v and hit max temp of 86.5c
While gaming, temps are much lower.. not seeing anything higher than 60c max

EDIT:
Cleared CMOS 3 more times using the button on the MB just to make sure the BIOS update didnt hang onto any old crap.. and low and behold im back to a stable -30 undervolt. Max temps in Cinebench r23 dropped a few more from 86.5c to with -15 to 83.2c with -30 undervolt. sitting at 39c idle

System:
9800X3D (using Thermalright Phantom Spirt 120 SE Air cooler)
MSI x870e Carbon
Kingston Furry Beast CL30 Ram (A-die) - 6000mhz CL 30-36-36-80 @ 1.4v
Asus Tuf 5070 Ti OC
2tb Samsung 990 Pro (NVMe 1), 2tb Samsung Evo Pro (NVMe 3), 4tb Samsung Evo Pro (NVMe 4) - Nothing in NVMe slot 2 to avoid GPU lane sharing
Corsair RM1000x

CrystalDiskMark - NVMe slot 1 with Samsung 990 Pro:
Read (MB/s)Write (MB/s)
SEQ1M / Q8T17387.356910.88
SEQ128K / Q32T17363.956834.77
RND4K / Q32T165739.495514.81
RND4K / Q1T175.97253.66
Click to expand...
I have gotten -30 undervolt stable on many of my cores while others are stuck around -15. How are you testing to confirm stability other than booting? My system would often boot at -30 but I'd start to get errors that impacted latency, causing momentary freezes. The thing that I'm using now is Aida64 with FP checked while checking CPU, Memory and Cache. I'm counting 8-9 hours per test without an error as "stable" and with undervolting, I'm sometimes seeing failures running those settings after 5 hours. Saw one of my cores that I was testing at -34 this morning (that passed -30, it's testing now at -34) fail after 5 hours, 43 minutes of testing.
Two of my cores are stable at -35, but I don't think I'll push them any further. It's an eclectic mix of numbers.
My plan is to create processor sets in Process Lasso and run different games/processes that need performance under those core sets, while others that need less performance (web browsing) run on the less-performant cores.
 
Upvote 0 Downvote
Alyred said:
I have gotten -30 undervolt stable on many of my cores while others are stuck around -15. How are you testing to confirm stability other than booting? My system would often boot at -30 but I'd start to get errors that impacted latency, causing momentary freezes. The thing that I'm using now is Aida64 with FP checked while checking CPU, Memory and Cache. I'm counting 8-9 hours per test without an error as "stable" and with undervolting, I'm sometimes seeing failures running those settings after 5 hours. Saw one of my cores that I was testing at -34 this morning (that passed -30, it's testing now at -34) fail after 5 hours, 43 minutes of testing.
Two of my cores are stable at -35, but I don't think I'll push them any further. It's an eclectic mix of numbers.
My plan is to create processor sets in Process Lasso and run different games/processes that need performance under those core sets, while others that need less performance (web browsing) run on the less-performant cores.
Click to expand...
You are definitely running your tests longer than me. For stability I run 2 hour long cinbench r23 multi core, memtest, and Aida64. I played black myth wukong for 5 hours today and didn't have issues either. I don't stress test too much.. maxing out the system for 8-9 hours isn't the same load my system will see when im playing games anyway. In fact, most games I'm seeing 20-40% CPU utilization while the GPU sits much higher. Gaming does not heat up the CPU anywhere near stability testing either. While gaming the CPU runs about 20-25c cooler
 
Upvote 0 Downvote
vic123159d02e5 said:
How about your cinebench scores are they on par with prior version? I do ask because I m used to see a little regression with each bios release , on single and multithreaded benchmarks.
Click to expand...
My Cinebench scores have been fairly stable. Multi Core always in between the 23-23.5k range. Single core between the 2k-2.15k range. I'm also just undervolting.. no overclock.
 
Upvote 0 Downvote
Indabox said:
You are definitely running your tests longer than me. For stability I run 2 hour long cinbench r23 multi core, memtest, and Aida64. I played black myth wukong for 5 hours today and didn't have issues either. I don't stress test too much.. maxing out the system for 8-9 hours isn't the same load my system will see when im playing games anyway. In fact, most games I'm seeing 20-40% CPU utilization while the GPU sits much higher. Gaming does not heat up the CPU anywhere near stability testing either. While gaming the CPU runs about 20-25c cooler
Click to expand...
You'll definitely see lower temps during gaming, you won't be processor limited unless you're gaming at 1080p. So that's definitely expected behavior. Cinebench is a great benchmark but doesn't really do stress testing as it won't report errors. If you're mostly gaming this probably won't ever matter so long as you're not experiencing crashes or freezes.
I don't do a huge amount of production workloads, but when I do I can't afford random errors in calculations.
Can you run Samsung's performance tests from within Samsung Magician on your 990's? I don't think we have numbers from A21 in CrystalDiskMark to compare them to.A30, and they use different metrics for random IOPS.
 
Upvote 0 Downvote
Here's a quick test on my 4TB Samsung 990 Pro on A27 bios:

Samsung 990 Pro Magician share.png
 
Upvote 0 Downvote
Back
Top