MPG X870E CARBON WIFI Beta BIOS

Sort by date Sort by votes
Mr.Fox said:
Sounds like you're "stronger" power supply is defective and needs to be exchanged if it's still under warranty.
Click to expand...
It's possible that he's just not ventilating it properly, or has the fan disabled - Seasonic power supplies have a switch to turn off the fan if you're venting air from inside the case, which isn't a great option IMO, but there we go.
At either rate, unfortunate that the P/S was shutting off and causing the crashes that got blamed on the BIOS stability, throwing a red herring into the testing and communication here.
 
Upvote 0 Downvote
Alyred said:
It's possible that he's just not ventilating it properly, or has the fan disabled - Seasonic power supplies have a switch to turn off the fan if you're venting air from inside the case, which isn't a great option IMO, but there we go.
At either rate, unfortunate that the P/S was shutting off and causing the crashes that got blamed on the BIOS stability, throwing a red herring into the testing and communication here.
Click to expand...
Yes, before the shutdowns I had the error codes on the motherboard and the infinite loops until I stabilized with the BIOS a25. With the Bios A25 it worked fine for 2 days, then the only symptom was the shutdown according to the load. Now I've changed the switch that controls the fan from the power supply and haven't had any unexpected shutdowns yet. We need to continue investigating any and all behavior of the equipment, I don't usually rule out any hypothesis. My case was an example and will be recorded, being able to help other people.
 
Last edited:
Upvote 0 Downvote
I had only headaches with this motherboard, BIOS A23 seemed to have performed better except for random loop reboot..(imagine that, being almost glad that it was the only problem i had left)
I'm looking forward to try A25, but "maybe" one day MSI will release a non-beta release of their bios ?
I actively discouraged already a bunch of people(at least 5 that I know for sure and maybe others that read me..) from buying that board and will continue to do so, until we get what we paid for..
 
Upvote 0 Downvote
vic123159d02e5 said:
I had only headaches with this motherboard, BIOS A23 seemed to have performed better except for random loop reboot..(imagine that, being almost glad that it was the only problem i had left)
I'm looking forward to try A25, but "maybe" one day MSI will release a non-beta release of their bios ?
I actively discouraged already a bunch of people(at least 5 that I know for sure and maybe others that read me..) from buying that board and will continue to do so, until we get what we paid for..
Click to expand...
Unfortunately problems happen with all manufacturers, currently we managed to stabilize with the BIOS a25. The biggest challenge is to identify the source of the problem in order to fix it. The only problem I had after the a25 BIOS was unexpected shutdown, but I already found out it was the temperature of the PSU.
 
Upvote 0 Downvote
RobsonByte said:
Unfortunately problems happen with all manufacturers, currently we managed to stabilize with the BIOS a25. The biggest challenge is to identify the source of the problem in order to fix it. The only problem I had after the a25 BIOS was unexpected shutdown, but I already found out it was the temperature of the PSU.
Click to expand...
Yes, sure, but I’ve never had a motherboard this unstable, never.
Just to provide some context for my criticism of MSI, after five months since its release, we still don’t have a final, non-beta BIOS version.
Here’s a list of the issues I’ve encountered with this board:
  • USB disconnections
  • Unable to boot with a USB stick plugged in
  • No display when connecting via DisplayPort
  • Voltage instability when overclocking
  • Reboot loops
1 week before buying this, I had to RMA a Tomahawk X870 that wouldn’t even power on ( that was early december), still not had any news from the board... MSI




 
Upvote 0 Downvote
There are many problems to be solved, I also don't like the big difference in performance between equipment granted for influencer testing and equipment sold to end consumers. Even with the a25 BIOS I still have problems with USB disconnects on the Hub 3.0.
 
Upvote 0 Downvote
I have cold boot issues with 25 bios 3 beeps and stays on code 15 on power up . When it beeps I just hit the restart button and it boots.all setting are good bios do not it reset. Settings. I went back to 24 and everything if good.9800x3d with 6000 gskill running at 8000.if you want me to test something let me know.
 
Upvote 0 Downvote
dlomond said:
I have cold boot issues with 25 bios 3 beeps and stays on code 15 on power up . When it beeps I just hit the restart button and it boots.all setting are good bios do not it reset. Settings. I went back to 24 and everything if good.9800x3d with 6000 gskill running at 8000.if you want me to test something let me know.
Click to expand...
Have you done the complete CMOS cleanup? Code 15 with 3 beeps is related to RAM.
 
Upvote 0 Downvote
Yes full CMOS on every bios update .that is why I posted about 6000 to 8000. I did not do to much checking or even adding more voltage to it just info for the next bios update or if someone else has the problem.
 
Upvote 0 Downvote
Did most of you pickup the boards around launch? I have been following a lot of posts as I just got my x870e carbon. I must of won the lottery because I'm having 0 issues.. WiFi works, Ethernet ports work, all USB ports work (tested thumb drives, nvme enclosure, SATA drives etc..). I have the 9800x3d with a -30 undervolt (no OC) and am getting right at 24k scores in cinbench r23. Ran 1 hour stress tests with various tools and system is fully stable. All this with the BIOS that came pre installed 1.A10

In stress tests with the -30 undervolt I have not seen the CPU go over 77c temps while staying at max boost 5,250. Before the undervolt I was getting up to 85c

I don't use any USB hubs though and that seems to be be a main culprit of some issues.

Below is my setup:
  • Motherboard: MSI x870e Carbon Wi-Fi
  • CPU: AMD 9800X3D
  • CPU Cooler: Thermalright Phantom Spirit 120 SE Black with 2x 120mm high speed be-quiet fans (was thinking AIO but with this CPU it's not needed imo)
  • GPU: MSI 3080 Gaming X Trio - will be removed from current rig (plan to update to 5080 Ti / Super whenever they come out)
  • DDR5 Ram: Kingston Fury Beast 32gb (2x16gb kit) KF560C30BBEK2-32 (A die)
  • Power Supply: Corsair RMx Series RM1000x CP-9020271-NA
  • M.2_1: Samsung 990 Pro 2tb (Boot/OS Drive)
  • M2._2: Leaving empty
  • M.2_3: Samsung 990 Evo Plus 2tb
  • M.2_4: Samsung 990 Evo Plus 4tb
  • Case: Antec Flux Pro with 3x 140mm front intake, 2x 120mm bottom intake, 3x 140mm top exhaust, 1x 140mm rear exhaust
  • OS: Windows 11 pro 24H2
  • Monitor: Alienware 39" ultra wide, Samsung 27" on DP & 65" 4k Sony TV over HDMI 2.1
 
Last edited:
Upvote 0 Downvote
I got the board I have no issues had the network problem first when I had the board first week but now I have no problems pbo -20 and overclock to 200 overclock.. ram at 6000 overclock at 8000 and 3090 I have a 5090 supreme on order.
 
Upvote 0 Downvote
Indabox said:
Did most of you pickup the boards around launch? I have been following a lot of posts as I just got my x870e carbon. I must of won the lottery because I'm having 0 issues.. WiFi works, Ethernet ports work, all USB ports work (tested thumb drives, nvme enclosure, SATA drives etc..). I have the 9800x3d with a -30 undervolt (no OC) and am getting right at 24k scores in cinbench r23. Ran 1 hour stress tests with various tools and system is fully stable. All this with the BIOS that came pre installed 1.A10

In stress tests with the -30 undervolt I have not seen the CPU go over 77c temps while staying at max boost 5,250. Before the undervolt I was getting up to 85c

I don't use any USB hubs though and that seems to be be a main culprit of some issues.

Below is my setup:
  • Motherboard: MSI x870e Carbon Wi-Fi
  • CPU: AMD 9800X3D
  • CPU Cooler: Thermalright Phantom Spirit 120 SE Black with 2x 120mm high speed be-quiet fans (was thinking AIO but with this CPU it's not needed imo)
  • GPU: MSI 3080 Gaming X Trio - will be removed from current rig (plan to update to 5080 Ti / Super whenever they come out)
  • DDR5 Ram: Kingston Fury Beast 32gb (2x16gb kit) KF560C30BBEK2-32 (A die)
  • Power Supply: Corsair RMx Series RM1000x CP-9020271-NA
  • M.2_1: Samsung 990 Pro 2tb (Boot/OS Drive)
  • M2._2: Leaving empty
  • M.2_3: Samsung 990 Evo Plus 2tb
  • M.2_4: Samsung 990 Evo Plus 4tb
  • Case: Antec Flux Pro with 3x 140mm front intake, 2x 120mm bottom intake, 3x 140mm top exhaust, 1x 140mm rear exhaust
  • OS: Windows 11 pro 24H2
  • Monitor: Alienware 39" ultra wide, Samsung 27" on DP & 65" 4k Sony TV over HDMI 2.1
Click to expand...
Nice setup! Good that it's stable with the A10 BIOS, though you won't be getting the newer AGESA code from AMD for the processor itself. That typically has some performance fixes and code optimizations.
USB hubs definitely had some problems on A23, though they did seem more stable in A24 and A25 seems to have solved the issue altogether. There's still some performance issues on the Samsung 990 drives with A25 that they'll hopefully iron out, though they do seem stable. Just not getting the performance in random IOPS that you're paying for. Strangely this seems to hit the M2_1 slot more than the M2_3 slot - I have 990 pro 4TBs in both those slots and there is a notable IOPS performance improvement on the M2_2 slot over the M2_1. like... 200000 IOPS difference. Neither slot is getting the roughly 1.5 million IOPS they should be and even the M2_2 seems to be stuck at around 1.2 million. A21 BIOS seems to be the last BIOS that performs at that full speed.
Otherwise, my 9800x3d is running stable in OCCT Extreme/Large Load/AVX512 and AIDA64 CPU/FPU/Memory for 3+ hours. RAM gets warm during the toughest burn-ins but not dangerously, in the low to mid 60's. CPU temp spiked at 86 once or twice (limited PBO max temp to 85 in the BIOS) but average was around 72 at the highest settings. Idle at low 40's (43 right now with all the chrome tabs, discord, and some other random things running, actual core temps in the low 30's).
The biggest problems I had when I got the boards as combo packages with 9800x3ds is that the XMP settings for the DDR8000 RAM from their QVL list causes a lot of issues with stability just at XMP. Had to do a deep-dive into how to manually tune DDR5 RAM as well as how to adjust some of the voltages for the CPU. Took several weeks to get it stable and then A23 with its USB-related boot loops came out and really threw a spanner in the works. Additionally, the 10.73 network drivers were causing massive system-level stuttering and latency spikes. Took me several weeks to find someone posting that rolling back to the 10.72 version of the drivers solved that issue and that worked for me. Not sure if those driver issues were related to the particular BIOS/AGESA of the CPU/Chipset or what, but haven't tried to roll forward to the 10.73 drivers yet. Unsure what I might be missing out on there, 10.72 runs the 5GB/s port at the expected speeds as checked using iPerf3 against one of my UnRAID servers running 10GB/s cards over a 10GB/s switch.
Other than those hiccups, happy with the hardware itself. Hoping MSI cleans up whatever is causing the slowdowns with the M2 slots.
dlomond said:
I got the board I have no issues had the network problem first when I had the board first week but now I have no problems pbo -20 and overclock to 200 overclock.. ram at 6000 overclock at 8000 and 3090 I have a 5090 supreme on order.
Click to expand...
How do you mean 6000 overclock at 8000? Are you pushing RAM specced at 6000 up to an 8000 overclock? What sorts of speeds and latency are you getting with that? I have a 48GB DDR8000 kit but it can't run that stable, will fail Prime95, OCCT CPU+RAM and AIDA64 CPU+FPU+RAM tests due to the IMC on the 9800x3d being unable to push those speeds properly. Looks like some others on the forum with 7900x and 7950x can push more memory bandwidth to get faster speed and lower latency.
Where does everyone keep being able to get 5090's from? I'd order one in a heartbeat if SOMEONE would get enough stock to let me order one, haha I skipped the 40 gen but running a venerable EVGA 3090 FTW3 which is starting to show its age. My new P/S is ready to push that ATX3 2x6 juice...
 
Upvote 1 Downvote
I was ready to pull my board out and send it back, I have Lian-Li wireless fans and needed more USB headers than the motherboard has so I bought two NZXT hubs. When using A21 or A23 BIOS I was getting boot loop. I installed A25 and now all my devices are plugged in and working great. I did have trouble entering BIOS, I had to boot into troubleshooting mode, If I hit DEL during startup it locks up on code 64 with the green LED on and will not continue. edit: If I wait until the popup saying to hit delete shows up and then hit delete I can get into BIOS.
 
Last edited:
Upvote 0 Downvote
Anaksagoras said:
I recommend you try these settings, I use them without any problems:

Optimum ram sets:
UCLK=MEMCLK
FLCK: 2133
CPU SOC: 1.25V
DRAM VDD: 1.4v
DRAM VDDQ: 1.35v

Optimum PBO Settings:

(You should make all your PBO settings from the AMD Overclock menu. This way you get the most optimum result.)

AMD PBO menu
PBO: Advanced
PBO Limit: Motherboard
PBO Scalar: 5X
CPU Boost: Manual +100
Platform Thermal Throttle: 75C
Curve Optimizer: Negative 15
Click to expand...
I tested some of these settings and what I noticed was that I had almost 0% performance change. I tried using CO of -30 on all cores and it is having the same performance as -15, very confusing. Since I did not feel comfortable, I went back to using the specific setting for each core.

AMD Chipset Software Install Summary (AMD Chipset Driver 7.01.08.129)
Name : AMD GPIO Driver (for Promontory)
Version : 3.0.3.0
Install : Fail
Name : AMD Interface Driver
Version : 2.0.0.21
Install : Fail
 
Last edited:
Upvote 0 Downvote
Alyred said:
Nice setup! Good that it's stable with the A10 BIOS, though you won't be getting the newer AGESA code from AMD for the processor itself. That typically has some performance fixes and code optimizations.
USB hubs definitely had some problems on A23, though they did seem more stable in A24 and A25 seems to have solved the issue altogether. There's still some performance issues on the Samsung 990 drives with A25 that they'll hopefully iron out, though they do seem stable. Just not getting the performance in random IOPS that you're paying for. Strangely this seems to hit the M2_1 slot more than the M2_3 slot - I have 990 pro 4TBs in both those slots and there is a notable IOPS performance improvement on the M2_2 slot over the M2_1. like... 200000 IOPS difference. Neither slot is getting the roughly 1.5 million IOPS they should be and even the M2_2 seems to be stuck at around 1.2 million. A21 BIOS seems to be the last BIOS that performs at that full speed.
Otherwise, my 9800x3d is running stable in OCCT Extreme/Large Load/AVX512 and AIDA64 CPU/FPU/Memory for 3+ hours. RAM gets warm during the toughest burn-ins but not dangerously, in the low to mid 60's. CPU temp spiked at 86 once or twice (limited PBO max temp to 85 in the BIOS) but average was around 72 at the highest settings. Idle at low 40's (43 right now with all the chrome tabs, discord, and some other random things running, actual core temps in the low 30's).
The biggest problems I had when I got the boards as combo packages with 9800x3ds is that the XMP settings for the DDR8000 RAM from their QVL list causes a lot of issues with stability just at XMP. Had to do a deep-dive into how to manually tune DDR5 RAM as well as how to adjust some of the voltages for the CPU. Took several weeks to get it stable and then A23 with its USB-related boot loops came out and really threw a spanner in the works. Additionally, the 10.73 network drivers were causing massive system-level stuttering and latency spikes. Took me several weeks to find someone posting that rolling back to the 10.72 version of the drivers solved that issue and that worked for me. Not sure if those driver issues were related to the particular BIOS/AGESA of the CPU/Chipset or what, but haven't tried to roll forward to the 10.73 drivers yet. Unsure what I might be missing out on there, 10.72 runs the 5GB/s port at the expected speeds as checked using iPerf3 against one of my UnRAID servers running 10GB/s cards over a 10GB/s switch.
Other than those hiccups, happy with the hardware itself. Hoping MSI cleans up whatever is causing the slowdowns with the M2 slots.

How do you mean 6000 overclock at 8000? Are you pushing RAM specced at 6000 up to an 8000 overclock? What sorts of speeds and latency are you getting with that? I have a 48GB DDR8000 kit but it can't run that stable, will fail Prime95, OCCT CPU+RAM and AIDA64 CPU+FPU+RAM tests due to the IMC on the 9800x3d being unable to push those speeds properly. Looks like some others on the forum with 7900x and 7950x can push more memory bandwidth to get faster speed and lower latency.
Where does everyone keep being able to get 5090's from? I'd order one in a heartbeat if SOMEONE would get enough stock to let me order one, haha I skipped the 40 gen but running a venerable EVGA 3090 FTW3 which is starting to show its age. My new P/S is ready to push that ATX3 2x6 juice...
Click to expand...
Check your drive PCIe speed in Magician. I am noticing on the X870E Tomahawk with the latest BIOS, the M2_1 slot will keep reverting to PCIe 1 x 4. I can get it to run full speed if I go into BIOS and manually select PCIe Gen 4 or 5 for the slot, but it will revert again after a shut down and cold boot. Basically, I have to keep going between Gen 4 and 5 setting on every boot to get the drive to run full speed. Very annoying! I'm guessing the Carbon probably from the same issue. Samsung 990 Evo Plus 1TB.
 
Upvote 0 Downvote
RobsonByte said:
I tested some of these settings and what I noticed was that I had almost 0% performance change. I tried using CO of -30 on all cores and it is having the same performance as -15, very confusing. Since I did not feel comfortable, I went back to using the specific setting for each core.

AMD Chipset Software Install Summary (AMD Chipset Driver 7.01.08.129)
Name : AMD GPIO Driver (for Promontory)
Version : 3.0.3.0
Install : Fail
Name : AMD Interface Driver
Version : 2.0.0.21
Install : Fail
Click to expand...

GPIO (for Promontory) will always fail because that's the codename for the chipset for AM4 (Promontory). I've read that the AMD Interface driver installs but always shows fail because of an installer log issue, but that apparently fails on pretty much everyone's installation.
Undervolting on your cores won't improve performance specifically, it allows you to raise your scalar and boost clocks for longer while keeping your temperatures lower. You could also increase your base clock but that's more complex and a bit more "dangerous" to stability to keep the base clock running at higher frequency. Ideally you'll want to set your core COs individually as each core will have different tolerances for undervolting. Finding an all-core undervolt that is fully stable means that there are probably some cores that could be undervolted a little more but that takes time to test.
Memory clock that @Anaksagoras mentioned are ideal but not every CPU/RAM can reach those numbers. I also have been looking into FCLK settings and there's apparently a synchronicity when the FCLK is set to an even ratio to the UCLK/MEMCLK, at 3:2 it will lower your latency by a few nanoseconds. I've seen that in tests myself, though it's less than 2ns. Which ain't nothing, but not mind-blowing either.
You can test your bandwidth and latency by downloading AIDA64 as a demo, it'll show you your read speed and latency. You can also use it to stress test your system's combination of CPU, FPU and RAM. Throwing that at it will let you know if you have issues lurking when the CPU starts getting warm or gets saturated with a lot of work. You can have a system that runs on day-to-day stuff fine but gets unstable when working hard. The problems aren't really the crashes in those situations but the "errors" - think it returning a wrong number when doing math stuff, where it'll return the wrong information but won't throw an error.
 
Upvote 0 Downvote
I never said I had a problem overclocking beyond standard PBO, but the voltage is completely erratic. Sometimes it boots at 1.245V, then on the next reboot it hits 1.28V, and I’ve even seen it push to 1.34V! Even with BIOS A25, it’s almost impossible to boot if I leave a USB stick plugged in.
I’m running a –34 curve with a 103 BCLK and 2100 FCKL; the max boost is 5587 MHz, and my RAM runs at 8240 MHz at CL36. The chip never exceeds 73°C even during OCCT.
It can also achieve an all-core overclock at 5625 MHz with SMT disabled, but then it hits 88°C, so I reverted to the standard boost. This motherboard has no external clock (imagine that for a $500 board), so 103 BCLK is about the maximum achievable,103.5 is possible, but my RAM would need significantly more voltage (around 1.56V) for stability.
I don’t consider using standard PBO +200 mhz boost even with a negative curve as overclocking; most $200 boards are capable of similar performance. If I’d be happy with only a 200 MHz boost on PBO, I would have bought one of those cheap boards.
 
Last edited:
Upvote 0 Downvote
jlsmith.01712b902af said:
Check your drive PCIe speed in Magician. I am noticing on the X870E Tomahawk with the latest BIOS, the M2_1 slot will keep reverting to PCIe 1 x 4. I can get it to run full speed if I go into BIOS and manually select PCIe Gen 4 or 5 for the slot, but it will revert again after a shut down and cold boot. Basically, I have to keep going between Gen 4 and 5 setting on every boot to get the drive to run full speed. Very annoying! I'm guessing the Carbon probably from the same issue. Samsung 990 Evo Plus 1TB.
Click to expand...
Thanks for the hint. I did just check and it does seem to be maintaining 4th gen between reboots. I do automatically set the M.2, chipset, and PCIe slots as fixed at PCIe v4 in the BIOS, I hate letting the board determine it automatically and have some weird response set it down to 1.1 or something. :D So at least on the carbon wifi it's not experiencing that issue. If the Tomahawk is, that's a fairly significant issue that a lot of folks might not notice. I wonder if your BIOS upgrade went wonky or something? Any other folks that you've spotted reporting this? I was unable to find a Tomahawk x870 motherboard beta BIOS page...
In my (and another commenter's case a few pages back, I forget which right now) the speeds are definitely PCIe v4 level, but random write IOPS is much lower than it should be on the first M.2 slot:
1739746300713.png

This was from just after I upgraded to A25. The random write on A21 is around 1.2 million IOPS with the same hardware.
Strangely, the M2_3 slot has better performance in Random Write (which is what you really want over Sequential), but a sequential read that is slightly lower than it should be, even when factoring in the overhead from going through the chipset:
1739746460255.png

I expect to see a slight difference, like the difference between the random read on the two drives, but everything is fairly significantly lower on that D drive, except the random read IOPS which is about the delta it should be (around 3%).
 
Upvote 0 Downvote
vic123159d02e5 said:
I never said I had a problem overclocking beyond standard PBO, but the voltage is completely erratic. Sometimes it boots at 1.245V, then on the next reboot it hits 1.28V, and I’ve even seen it push to 1.34V! Even with BIOS A25, it’s almost impossible to boot if I leave a USB stick plugged in.
I’m running a –34 curve with a 103 BCLK and 2100 FCKL; the max boost is 5587 MHz, and my RAM runs at 8240 MHz at CL36. The chip never exceeds 73°C even during OCCT.
It can also achieve an all-core overclock at 5625 MHz with SMT disabled, but then it hits 88°C, so I reverted to the standard boost. This motherboard has no external clock (imagine that for a $500 board), so 103 BCLK is about the maximum achievable,103.5 is possible, but my RAM would need significantly more voltage (around 1.56V) for stability.
I don’t consider using standard PBO +200 mhz boost even with a negative curve as overclocking; most $200 boards are capable of similar performance. If I’d be happy with only a 200 MHz boost on PBO, I would have bought one of those cheap boards.
Click to expand...
If you're running a 9800x3d it sounds to me like you're pushing it to its very limits by boosting it's bclk and then dropping it's voltage via CO to about the maximum available, and then having stability issues. I agree, this board should have features the tomahawk doesn't for the extra money spent on it. There's still problems to be worked out and MSI is still probably just spinning up from Chinese new year. But you're really pushing RAM and Processor to the limits there.
 
Upvote 0 Downvote
Back
Top