They sent my broken product back.

[James]

Buy Intel P4500 SSD (PAID LINK): https://geni.us/1ZRtl

 

My RMA return from Intel finally arrived so I'm going to find out if they solved the problem I was having with these storage drives -and if so- what they changed.

 

 

[da na]

first reply

pog

[Hunter259]

Lol Gigabyte fixing things. They couldn't be assed to put proper ACPI and UEFI on a server board I was using. And that was just one board. Never liked any board from them.

[sub68]

Well that makes sense that they don't have time to try every single problem

[Kisai]

39 minutes ago, James said:

Buy Intel P4500 SSD (PAID LINK): https://geni.us/1ZRtl

 

My RMA return from Intel finally arrived so I'm going to find out if they solved the problem I was having with these storage drives -and if so- what they changed.

 

 

I'll state the obvious, but clearly the problem was the server in the first case. Had it been the drives, all the drives with the same firmware should have failed at the exact same time, every time. My wild guess here is that the gigabyte platform was not QA'd with PCIe 3.1 drives. Though what exactly they were QA'd with is anyone's guess. Probably whatever storage platform the server was intended to be sold with. To which, why didn't Gigabyte update the firmware?

 

Like a very basic troubleshooting step is hard to do in a server environment, which would be to try the drives in another unit, but outside of a data center, where would you find an identical server configuration?

 

[Ubersonic]

Hate to be that guy but technically speaking isn't this AMD's fault not Intel's? If the issue is due to a specific incompatibility between an NVMe drive from Q2 2017 and CPUs from Q3/4 2018 and it can be fixed via a BIOS update, then it's a CPU issue right? :S

[Nystemy]

The intricacies of PCIe makes for a lot of fun troubleshooting time that one wish one never needed to have.
Though, can't say that x86, and NVMe is making the situation any better on that front.

 

I usually refer to software as a jungle, but sometimes hardware is its own. Though, at least a poorly cable managed rack looks like a jungle so one clearly knows what one is up against. But incompatible hardware, or just software issues are far harder to gauge at first glance.

 

Though, a lot of systems are not built with simplicity in mind, but rather a scatterbrained feature creeping behemoth trying to do everything. _{(So when a small issue pops up, it tends to have a domino effect and suddenly one's computer doesn't boot due to a fault one suspects to be on the PCIe bus, bit is in fact created by a partly responsive USB controller fuzzing about with an intermittent connection to the USB hub in one's display on the other side of a cable that has seen the office chair's wheel run over it a bit too many times, all as the hub is trying to handshake with a thumb drive, so that the BIOS can see if the thumb drive that is first on it's boot list is there or not, since one forget that one used that thumb drive for installing the OS 7 months back. (This is though likely not a realistic scenario for most setups, but I would not be surprised if one can achieve this specific case with some odd combination of hardware... But the idea is that the BIOS handshakes with all storage devices to search for the first boot device on its list, it finds the thumb drive, attempts booting from it, but then the cable rears its ugly head and the thumb drive disappears, leaving the BIOS confused and bleeping out a boot error. And obviously one still has the OS install image on the thumb drive, since why would one possibly remove that for....))}

[Middcore]

"Maybe Gigabyte just couldn't be arsed to fix this?"

 

That fits my priors about Gigabyte.

[dogwitch]

3 hours ago, Middcore said:

"Maybe Gigabyte just couldn't be arsed to fix this?"

 

That fits my priors about Gigabyte.

lol just like asus and TR mobos....

[namibj]

Do we know what happened? The video doesn't appear to say who (AMD or Intel) didn't properly conform to the spec (I'm assuming for now that the spec is strict enough that they can't both be fully compliant given the symptoms we see). Yes, it seems like the interrupts arrive too early, but that might also be one drive's interrupt being mistaken for another drive's interrupt, with the second drive not yet being ready.

[RejZoR]

Why not sell them if they are problematic and buy some other SSD's?

[csm10495]

I used to work on SSDs at Intel. I can almost completely assure you that no firmware engineer looked into your issue. 

 

Have some fun stories about RMAs... for another day though. 

[Gagz9k]

RMA can be... funny. I had a 3700X RMA'd from Chile, that's where I live, because one core was defective and caused random reboots and they just send me a new one... with the SAME problem, and I know this is a common issue with AMD CPUs. I could work around that with undervolt and turning off c-state... But I really wanted my full AMD experience hahahaha So I contacted them again, really pissed this second time, explaining this was my workstation machine that I actually use to work/make money, and they actually send me a brand new 3950X

 

First RMA was long and unsuccessful.

Second try was half the time (a week and a half, between picking up and delivering) and really nice, because full blown upgrade.

[jdwx]

Did LTT ever figure this issue out?  Because we just ran into what looks like the exact same thing with an Epyc processor on a Supermicro board with Intel DC P4510 4TB U.2 SSDs.  (The LTT videos were some of the very few google results for the error message.)

 

In our case, we were able to make the problem go away by adding DIMMs.

 

It made the problem go from "every 20-30 seconds things would stall for 20-30 seconds" to "has not occurred even once on any drive in a week."

 

I don't have a good explanation for why that makes a difference.  The only even possibly relevant fact is that the Epyc CPUs have 8 memory channels and we went from 4 filled to 8 filled.  I wouldn't expect that to affect PCIe performance or capability, but here we are!

 

Curious if this is a general-purpose solution or a fluke.