Run a benchmark, examine the write speeds, if its about 50% of what you should get as most others experiened then yes it should nearly double by turning off the AX200 module "in bios" not in windows. I ordered an AX210 both M.2 Type E and PCIe 1x to tinker around, seems like a bit of a waste but I guess 6E is becoming the standard and I can try it out later once routers drop sub 500$ for it.
Secondary NVMe drives in M.2 slots running off PCH - 50% write speed @ PCI 4.0 4x
- Thread starter Solace
- Start date
Sitting at 512k Bus for drives.
How do you know the AX210 isn't also 128 byte bus? Is it published somewhere?
In my case, the FireCuda 520 is "first gen" Gen4, so it's not the fastest (5000/4400). Crucial P5 is only Gen3 (3400/3000). Looks like both are within 10% of rated speeds. I guess only newer Gen4 SSDs are impacted with 128 byte bus?
And well the SN850 is fine since it's not impacted anyhow.
Might be specific to 4.0 drives, I doubt enough complaints surfaced considering the majority of people are running a single 1tb/2tb/4tb drive due to the cost associated with it. If its negotiating properly then maybe its just a limitation on the bus and PCI 4.0 drives need 256bytes minimum.
Hard to believe this is the case if WD's firmware fix for the SN850 worked miracles if it really was a hardware/bus limitation. If you have the time/effort you could switch the drives and see if there are any implications for the sake of others specific to seagate drives (generally they are the lower tier of drives) and likely subject to the same issue. WD are pretty good with quality control and acknowledging issues.
Hard to believe this is the case if WD's firmware fix for the SN850 worked miracles if it really was a hardware/bus limitation. If you have the time/effort you could switch the drives and see if there are any implications for the sake of others specific to seagate drives (generally they are the lower tier of drives) and likely subject to the same issue. WD are pretty good with quality control and acknowledging issues.
think ill do a fresh install if performance is still in the 5000mb range ill see if they can offer a rma on both these drives no reason to suddenly lose 1500 mbs other then a failing drive.
Likely the OS/Drivers are at fault if something auto updated... Also what current version of windows are you on? Mine would be 21H1
Both are 512? Could you please give a snippet of the drive and AX200 when you have time as shown below... These are like mass produced and so standard that I'm curious as to what in the world could be causing the negotiation to be different considering the max payload size is defined at the time of the firmware compilation. I poped the 960 in to be wiped/sold and testing 3.0 for degraded performance as well. Likely it would not be affected since 128 probably handles up to almost max for PCI 3.0 x4
EDIT: 3.0 drives perform roughly the same -10% on writes, read is 100% only the gen 2 4.0 drives are probably impaired.
Last edited:
Id guess it depends on the amount of CPU lanes available, likely 2x is direct to CPU and 3rd is PCH. If 2/3 are PCH then it would likely degrade them all.
will dig into this soon as i get a chance only found one that showed 512kb its the latest version of windows did a media creation iso then regged with a key for pro.
also only pcie device in rig is the 3090.
also only pcie device in rig is the 3090.
Thx, the 3090 would be direct to the CPU using 16x of the 24 lanes the 5950 has I believe. there are 4x direct to the PCH controller and another 4 to other PCIe slots.
xentr_legal_notice_title
xentr_legal_notice_description