AMD X670E motherboard bug downgrades PCIe 5.0 SSDs to PCIe 1.0 speeds

MSI X670 motherboards
(Image credit: MSI)

One of the key selling points of AMD's Ryzen 7000-series and X670E motherboards was support for the PCIe 5.0 specification and SSDs with a PCIe 5.0 x4 interface. But some users with AMD X670E motherboards are encountering significant problems with some PCIe 5.0 SSDs, as they are being downgraded to PCIe 1.0 speeds, which drastically reduce performance, according to Wccf Tech.  

The issue has affected users across multiple X670E motherboard brands, including Asus and MSI, who initially experienced normal speeds before encountering significant slowdowns. The issue also affects various drives, including, but not limited to, Crucial's T700 and T705. In fact, engineers from Crucial confirmed that the issue persists on various motherboards and drives — so it's not specific to actual hardware, but rather to the platform itself (and/or the way PCIe 5.0 slots are implemented).

"We would like to inform you that we escalated your issue to our dedicated team for further investigation and they informed that the problem lies with the motherboard rather than the Crucial SSD," a statement by Crucial published over Asus's ROG forums reads. "This behavior has been observed across various motherboards from different manufacturers, and we were able to replicate it on our in-house systems as well." 

AMD's Ryzen 7000 processors support 28 PCIe 5.0 lanes: 16 to connect a graphics card (although there are currently no consumer-grade graphics cards with a PCIe 5.0 x16 interface), eight to connect two NVMe SSDs with a PCIe 5.0 x4 interface, and four to connect to a chipset (such as AMD's X670E). While AMD does not mandate PCIe 5.0 support on lower-end motherboards, premium X670E boards support all 28 PCIe 5.0 lanes. However, actual implementation of the supported PCIe 5.0 lanes depends on the motherboard's design. 

The issue is occurring when both the first PCIe 5.0 slot for graphics cards and the first M.2 slot with a PCIe 5.0 x4 interface are populated. In this case the system encounters random crashes and some users are even unable to boot into Windows. The system boots after several restarts, but it sets the M.2 PCIe speed to PCIe 1.0. Troubleshooting has shown that reverting SSDs to PCIe 4.0 speeds or using PCIe 4.0 drives entirely can resolve these issues, but this undermines the advantage of AMD's platform being mandated to support PCIe 5.0 for both graphics and SSDs. 

Some MSI motherboard users have reported success after applying a BIOS update (version 1.0c), which seems to resolve the speed and stability issues. However, this fix has not been universally successful. Asus reportedly claims that the issue is with SSDs — not with the motherboards. 

The root cause of the issues with AMD's X670E currently remains unknown. Wccf Tech claims that in some cases the first PCIe 5.0 slot for graphics cards shares the lanes with the first M.2 slot for PCIe 5.0 x4 SSDs. If this is the case, then PCIe implementation on such platforms is wrong (as PCIe lanes should not be shared in this case — although PCIe 5.x specifications support x12 mode and it is technically possible to share the lanes, albeit bifurcating them by the motherboard is challenging), but this at least explains why reverting to PCIe 4.0 speeds solves the issue in some cases. 

We do not have block diagrams of actual AMD X670E motherboard implementations, so we cannot say for sure whether some do indeed share PCIe 5.0 lanes between PCIe x16 and M.2 x4 slots. 

AMD and other motherboard makers have yet to officially weigh in on the problems. 

Anton Shilov
Contributing Writer

Anton Shilov is a contributing writer at Tom’s Hardware. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.

TOPICS
  • Amdlova
    I know the problem, have name ASMEDIA... When amd change the chipset for another brand or make the chipset, I will be able to buy an AMD cpu!
    Reply
  • endocine
    Amdlova said:
    I know the problem, have name ASMEDIA... When amd change the chipset for another brand or make the chipset, I will be able to buy an AMD cpu!
    those lanes feeding the GPU and the first m.2 slot are coming from the CPU though
    Reply
  • Notton
    Between AMD, Asus, MSI, and Crucial, I'm going to believe Crucial the most.

    I'd be suspicious with the mobo base clock running above spec (It's still 100Mhz for PCIe, right?) because that eeks out a little extra performance to look good on benchmarks.
    Reply
  • ROB_DF_MX
    Since there are currently no consumer-grade graphics cards with a PCIe 5.0 x16 interface, changing in the BIOS of the affected Motherboards the first PCIe Slot configuration to PCIe4 could not solve the problem and leave the first SSD slot working to PCIe 5 ?
    Reply
  • NinoPino
    ROB_DF_MX said:
    Since there are currently no consumer-grade graphics cards with a PCIe 5.0 x16 interface, changing in the BIOS of the affected Motherboards the first PCIe Slot configuration to PCIe4 could not solve the problem and leave the first SSD slot working to PCIe 5 ?
    From what I understand, the M2 and the first slot are on the same lanes, so maybe setting the slot to PCIe4 could influence also the M2 speed.
    Reply
  • NinoPino
    Notton said:
    Between AMD, Asus, MSI, and Crucial, I'm going to believe Crucial the most.
    Actually, MSI and Crucial promptly investigated and MSI also released a first fix that seems to work for the most part of users.
    Only ASUS remained on the starting block, not good at all.
    Reply
  • Hotrod2go
    Have they solved the heat issues with 5.0 drives yet?
    Reply
  • Amdlova
    Hotrod2go said:
    Have they solved the heat issues with 5.0 drives yet?
    they have a new turbo jet fan bring temperature to normal conditions
    Reply
  • Hotrod2go
    Amdlova said:
    they have a new turbo jet fan bring temperature to normal conditions
    Great, more noise to tolerate...
    Reply