Question Asus Prime B550 Plus M.2_2 Port is slow

[Cartmon]

Hey, hoping to get some help here after 2 days of pulling my hair out trying to figure out a way to get anyway near the advertised speeds for the samsung 970 evo plus I purchased and installed into my system a couple days ago.
My system is an Asus B550 Plus prime series mobo, AMD Ryzen 5 5600x and RX6800 installed in PCIex16_1.
The system was a prebuilt that came with a 512GB SSD in M.2_1 and a 2TB HDD connnected to SATA6G_6.
I noticed my SSD was filling up and rather than dealing with the slower speed of a HDD I opted to buy a 970 evo plus 2TB.
I've installed the new SSD in M.2_2 and created a new volume on the SSD so that it would be read by samsung magician.
After getting it all setup and ready to go I carried out a benchmark to see what kind of speed I'd realistically be getting and I was sorely disappointed at how far off the advertised values I was getting.
I've since downloaded and installed the samsung nvme drivers, updated my bios to the latest non beta release, updated my chipset and checked for updates for all storage controller drivers. I've also gone through the bios options and made some changes such as disabling SATA as for a short time I thought maybe it was using SATA instead of PCIe for the SSD as the M.2_2 port can use both. After some googling trying to resolve this I found suggestions of removing and reinstalling the drive which I have done twice and after each time the SSD has benchmarked at the same values.
Samsung magician and crystal disk info reports that the SSD is running on PCIe3.0x4 however the speed some more in line with only using 1 PCIE lane.
Planning to try swapping the SSDs around to see if that has any effect when I get some time this week.
Any help or suggestions would be greatly appreciated.

 

[USAFRet]

What does the Performance test in Samsung Magician show?

 

[Cartmon]

Roughly the same values as crystal disk

 

[Maxxify]

If you know how to do it, you can swap the drives (slots) and see if that makes any difference.

 

[Bob.B]

Hey, hoping to get some help here after 2 days of pulling my hair out trying to figure out a way to get anyway near the advertised speeds for the samsung 970 evo plus I purchased and installed into my system a couple days ago.
My system is an Asus B550 Plus prime series mobo, AMD Ryzen 5 5600x and RX6800 installed in PCIex16_1.
The system was a prebuilt that came with a 512GB SSD in M.2_1 and a 2TB HDD connnected to SATA6G_6.
I noticed my SSD was filling up and rather than dealing with the slower speed of a HDD I opted to buy a 970 evo plus 2TB.
I've installed the new SSD in M.2_2 and created a new volume on the SSD so that it would be read by samsung magician.
After getting it all setup and ready to go I carried out a benchmark to see what kind of speed I'd realistically be getting and I was sorely disappointed at how far off the advertised values I was getting.
I've since downloaded and installed the samsung nvme drivers, updated my bios to the latest non beta release, updated my chipset and checked for updates for all storage controller drivers. I've also gone through the bios options and made some changes such as disabling SATA as for a short time I thought maybe it was using SATA instead of PCIe for the SSD as the M.2_2 port can use both. After some googling trying to resolve this I found suggestions of removing and reinstalling the drive which I have done twice and after each time the SSD has benchmarked at the same values.
Samsung magician and crystal disk info reports that the SSD is running on PCIe3.0x4 however the speed some more in line with only using 1 PCIE lane.
Planning to try swapping the SSDs around to see if that has any effect when I get some time this week.
Any help or suggestions would be greatly appreciated.

Does the 512 ssd benchmark good?
If it does swap the 2 ssd's and rerun the benchmark on both.

If it also tests bad then it would seem the problem is further back.

 

[Cartmon]

Does the 512 ssd benchmark good?
If it does swap the 2 ssd's and rerun the benchmark on both.

If it also tests bad then it would seem the problem is further back.

512 ssd benchmarks as good as I guess I could expect for a cheap generic drive that came in a prebuilt system, going to try swapping the drive slots when I get some time after work later today.

 

[Cartmon]

Found the time tonight to swap the drives around. With the drives swapped I got near advertised speeds for the 970 evo sdd and the generic 512gb ssd suffered a similar speed in the m.2_2 slot.
So looks like the drive itself is not the problem but it seems like the m.2_2 slot is only running at PCIe3.0x1 despite samsung magician and crystal disk info both saying that the drive in m.2_2 is using PCIe3.0x4 interface.
As far as I'm aware I've not ran out of PCIe lanes as I have no other expansion cards installed, am I missing something here?

 

[Bob.B]

Found the time tonight to swap the drives around. With the drives swapped I got near advertised speeds for the 970 evo sdd and the generic 512gb ssd suffered a similar speed in the m.2_2 slot.
So looks like the drive itself is not the problem but it seems like the m.2_2 slot is only running at PCIe3.0x1 despite samsung magician and crystal disk info both saying that the drive in m.2_2 is using PCIe3.0x4 interface.
As far as I'm aware I've not ran out of PCIe lanes as I have no other expansion cards installed, am I missing something here?

As a test unplug all sata devices.
Run the benchmark.

 

[Cartmon]

As a test unplug all sata devices.
Run the benchmark.

No sata devices are connected, I unplugged the sata cable from the motherboard that was going to the HDD when I installed the SSD.
I've also disabled SATA in the bios.
Ended up gaming for a bit tonight after swapping the drives and trying to change some other settings so tomorrow after work I'll be trying with just the OS SSD in m.2_2 and see if it benchmarks any better.
If the OS SSD still benchmarks slower in m.2_2 without the other SSD installed could this be a port issue I could avoid by using a PCIe to m.2 adapter?

 

[Bob.B]

No sata devices are connected, I unplugged the sata cable from the motherboard that was going to the HDD when I installed the SSD.
I've also disabled SATA in the bios.
Ended up gaming for a bit tonight after swapping the drives and trying to change some other settings so tomorrow after work I'll be trying with just the OS SSD in m.2_2 and see if it benchmarks any better.
If the OS SSD still benchmarks slower in m.2_2 without the other SSD installed could this be a port issue I could avoid by using a PCIe to m.2 adapter?

Try an adapter see what the results are.

 

[Cartmon]

I've removed the 970 from m.2_1 and left the 512gb ssd in m.2_2 and it still benchmarks slow in that port.
I'll order an adapter now and will update with results once it's arrived.

 

[Maxxify]

Sounds good, keep us updated. I'm curious as the issue.

 

[Cartmon]

Ordered and installed the below linked adapter with the 512gb ssd installed in the adapter, changed bios setting to force PCIex16_2 to run x4 and it still benchmarks at the same slow speed.
PCIe to M.2 Adapter
At this point I'm starting to think there's an issue with the chipset or the communication between chipset and CPU?

 

[SkyNetRising]

Check BIOS for available M.2 slot configuration options in BIOS.
(Advanced/Nvme controller/Nvme configuration).
M.2 drive has to be installed in M.2_2 slot for this.

 

[Cartmon]

Removed the adapter and installed the drive back into m.2_2.
NVMe configuration option in bios is greyed out and not able to access?

 

[mdd1963]

I wonder if perhaps some B550 boards are perhaps not capable of even simultaneous standard PCI-e 3.0 X4 lanes in M2_2 when/if a PCI-e 4.0 GPU is utilized?

 

[Cartmon]

My understanding of the b550 boards from hours of googling now is that the mobo has 20 PCIe 4.0 lanes that are dedicated for PCIex16_1 and m.2_1 and 4 PCIe 3.0 lanes as the connection to the chipset.
I've already tried a test with m.2_1 empty and the 512gb ssd connected in m.2_2.
I can set PCIex16_1 to run at gen 3 in bios for a test if you think that might make a difference but would that cause a noticable change in gpu performance?