Pcie 4.0 m.2 to u.2 adapter

[agrimm1]

Is the redriver and fancy cable still the only option?

I would like to be able to hook a p5800x up directly to my cpu m.2 and not have to use a pci-e card.

[CJRoss]

I didn’t see anything in the bios about PCI AER but even with the slightly longer connector and basic adapter it linked up at PCIe 4.

I did a bunch of CrystalDiskMark tests on the drive and it got toasty, so I’m going to have to figure out a different mounting solution in order to provide some airflow.

[disk_diddler]

That pricing is I believe the word is OUCH… very ouch.

[disk_diddler]

It looks like the post up the top covers it for $116 USD, not cheap to be honest, considering how expensive that drive is.

[Jolly]

With a pcie 4.0 u.2 device?

[Jolly]

Same, let me know if you find one that works.

Maybe Delock Produkte 62984 Delock M.2 Key M zu U.2 SFF-8639 NVMe Adapter mit 50 cm Kabel ?

[Domrockt]

Yes Kioxia CD6 8TB drives.

[CJRoss]

I’ve been using these for a while now and I haven’t noticed any problems. Everything seems to have linked up at PCIe4.

https://www.amazon.com/dp/B073WGN61Y

https://www.amazon.com/dp/B0CDRM1FMV

[agrimm1]

This one looks interesting, but I noticed that it’s got the 2260 form factor & is too small for my consumer boards.

[Jolly]

That doesn’t make any sense. If you can handle a 2280 you can handle a 2260. What board?
Also from searching around, looks like @LiKenun got a p5800x working reliably w m2 adapter with redriver

How fast is a P5800X Optane?

I just solved a similar problem myself after about 2 months. It turns out that the best combination thus far has been an M.2 adapter with a redriver and a very good cable that will definitely carry a PCIe 4.0 signal. I used an OCuLink-to-OCuLink cable, but Micro SATA Cables also sells ones that have the other end connect directly to the drive. …

[Dutch_Master]

@Jolly

If you can handle a 2280 you can handle a 2260

Not necessarily. The holes in the board to support a shorter form factor drive should be present and not every manufacturer had accounted for the fact drives get smaller and smaller over time. What used to barely fit in a 2280 just 2 years ago, now easily fits on a 2240 drive.

[lemma]

+1

Out of curiosity, though, I just checked all the mobos I’ve built in the past couple years. 2230/2242/2260/2280 is the majority of M.2 sockets. There’s some 2230/2242/2260/2280/22110 placements and some 2242/2260/2280. The most restricted board has one 2242/2260/2280 and three 2260/2280s.

[Jolly]

Is there a way to determine which PCIe device might be causing AER reports?
I found and flipped on the AER option on one of my boards (x570 aorus xtreme) and it started spitting out errors:

A corrected hardware error has occurred.

Component: PCI Express Root Port
Error Source: Advanced Error Reporting (PCI Express)

Primary Bus:Device:Function: 0x0:0x1:0x1
Secondary Bus:Device:Function: 0x0:0x0:0x0
Primary Device Name:PCI\VEN_1022&DEV_1483&SUBSYS_14531022&REV_00
Secondary Device Name:

Interestingly, it doesn’t seem to be the optane that’s the cause.

Full error log:
EventData

ErrorSource 4

FRUId {00000000-0000-0000-0000-000000000000}

FRUText

ValidBits 0xdf

PortType 4

Version 0x101

Command 0x406

Status 0x10

Bus 0x0

Device 0x1

Function 0x1

Segment 0x0

SecondaryBus 0x0

SecondaryDevice 0x0

SecondaryFunction 0x0

VendorID 0x1022

DeviceID 0x1483

ClassCode 0x30400

DeviceSerialNumber 0x0

BridgeControl 0x0

BridgeStatus 0x0

UncorrectableErrorStatus 0x0

CorrectableErrorStatus 0x1000

HeaderLog 00000000000000000000000000000000

PrimaryDeviceName PCI\VEN_1022&DEV_1483&SUBSYS_14531022&REV_00

SecondaryDeviceName

[aBav.Normie-Pleb]

Edit 3:

Looked at an X570 system with HWiNFO which is easier to search through, VEN_1022&DEV_1483 (“AMD Family 17h/19h - PCIe GPP Bridge[7:0]”) are basically all devices that are directly connected to CPU PCIe, on AM4 with the platform’s 16 + 4 + 4 PCIe Lane configuration these are:

• The X570 chipset chip with 4 Lanes (that would mean a defective motherboard)

• The main M.2 SSD slot with 4 Lanes

• The big x16 and x8 CPU PCIe slots

• Remove one device after another from these PCIe slots until those errors stop being generated and you’ll find out what’s causing them.

[disk_diddler]

I may have a post quite related to this one, having just seen it.

https://forum.level1techs.com/t/optane-905p-p5800x-in-nvme-slot-at-what-point-do-i-panic-they-should-work-right

I’m in a world of pain, some significant dollars on 2 Optane drives, both of which aren’t working in 3 different configurations. Should the Original Intel Gen 3 Retail 905P Box cable work with a P5800X? (It works with neither 905P nor the P5800X so I’m having trouble identifying my exact issue)

[aBav.Normie-Pleb]

@disk_diddler

Should the Original Intel Gen 3 Retail 905P Box cable work with a P5800X?

• Definitely not, they only handle PCIe Gen3 without errors. Tested this myself with the ones that came with Optane 905P 480 GB SSDs.

• There currently are NO passive adapter chains with cables that can properly connect a PCIe Gen4 U.2/U.3 SSD to a motherboard’s M.2 slot, no matter what any item description claims. Unfortunately the PCIe world is full of incompetent people and scammers.

• I’ve successfully tested these passive PCIe PCB adapters with Gen4 SSDs, no PCIe Bus Errors after 24 h of sustained load: Delock Products 90091 Delock PCI Express 4.0 x8 Card to 2 x internal U.2 NVMe SFF-8639 - Bifurcation

[disk_diddler]

Let me clarify…

Should the disk even appear? I don’t care if it runs slowly, it doesn’t appear.
If you check my thread, I’ve used the Optane 905P retail cable that (appears… brand new) with a 905P and a P5800X in

A HP 400 ProDesk Mini, A Lenovo M75Q Gen 2 Tiny and an Orico USB-C to NVMe adapter, all using a powered SATA cable from a HDD USB kit and on all 3 attempts, with both drives, I get no drives recognised.

So either the Optane drives and or optane cable are faulty OR the optane drives and or optane cable don’t like a “Molex to SATA style” power delivery from my old, crappy, USB C HDD kit
(again see my thread to make sense of it)

As it stands I’m … yeah I don’t know what to order now.

Blockquote * I’ve successfully tested these passive PCIe PCB adapters with Gen4 SSDs, no PCIe Bus Errors after 24 h of sustained load: Delock Products 90091 Delock PCI Express 4.0 x8 Card to 2 x internal U.2 NVMe SFF-8639 - Bifurcation

They are no good in my situation, I need an NVMe to U2 solution.

[aBav.Normie-Pleb]

• You should make sure that your OEM system’s BIOS is at all even compatible with all third-party PCIe NVMe SSDs. There have been numerous cases where these kinds of systems have a white list in their BIOSes where the manufacturer blocks devices from third-party manufacturers from even showing up. Would never buy OEM systems if you like flexibility.

• Depending on the BIOS, if the motherboard can’t establish a stable PCIe interface connection with a device it can happen that it just doesn’t work at all and it not just downgrades to PCIe Gen3 or Gen2.

• Check if you can force the M.2 interfaces of the motherboard to Gen3 instead of “Auto” and try again.

• The only buyable solution for adapting M.2 PCIe to U.2/U.3 PCIe Gen4 SSDs is one with an active M.2-to-Gen-Z adapter and an Gen-Z-to-SFF-8639 cable:

1. M.2 M-key PCIe Gen4 with ReDriver to Gen-Z 1C(EDSFF) Adapter

2. PCIe Gen4 Gen-Z 1C Male to U.2 (SFF-8639) Cable -50CM

MCIO would be the better modern connector type for PCIe Gen4/5 but the corresponding parts don’t seem to be out yet. SFF-8654, 8643 or OCuLink (mechanical strength) are not that great for PCIe Gen4 and above.

[CJRoss]

What do you define as properly? I’m using the ones I posted earlier and haven’t noticed any issues. I don’t see an AER option in the BIOS but if I grep dmesg for it I get multiple AER enabled messages but no error messages.

pcieport 0000:00:01.1: AER: enabled with IRQ 28

-[0000:00]-+-00.0  Advanced Micro Devices, Inc. [AMD] Starship/Matisse Root Complex
           +-00.2  Advanced Micro Devices, Inc. [AMD] Starship/Matisse IOMMU
           +-01.0  Advanced Micro Devices, Inc. [AMD] Starship/Matisse PCIe Dummy Host Bridge
           +-01.1-[01]----00.0  Intel Corporation NVMe DC SSD [3DNAND, Sentinel Rock Controller]

[aBav.Normie-Pleb]

The first step is always asking yourself: “Might this be too good to be true?”

How do I check if PCIe AER is working properly without having USD 500,000 signaling test equipment:

1. Get one of the numerous passive M.2-to-some-SFF-port adapters and cables I know can reliably establish a PCIe Gen4 interface connection but introduces a predictable amount of PCIe Bus Errors.

(There are parts with thousands of errors per second on one end of the spectrum and ones with only a single digit amount of errors per hour - BUT there are no completely passive adapter chains with PCIe Gen4 or above with cables that don’t introduce any PCIe Bus Errors at all)

2. Perform various extreme load tests on the connected SSD.

3. Check for PCIe Bus Errors via AER in operating system logs (on Windows these usually are WHEA ID 17).

4. If there are no errors at all with parts with previously thousands of errors per second then something is wrong, likely the BIOS not notifying the operating system of PCIe Bus Errors.

5. Check that the tested PCIe Interface is coming directly from the CPU and go through the BIOS to look for AER options, “Auto” (the usual default option) does not mean enabled most of the time. Also I have never seen logged PCIe Bus Errors on motherboard chipset PCIe interfaces, only direct CPU PCIe.

Why bother if no errors show up and drives seem to operate at their maximum speed?

• This means an error correction is working on the PCIe connections since engineers thought of that. This can be fine up to a point but if you exceed a certain threshold when multiple PCIe devices are under heavy load this can lead to janky behavior like the machine intermittently stuttering, freezing or even crashing - seemingly without any reason and that sucks balls.