SSD Speed/Reliability

That can happen with a bad interface chip. I only had one do that, shirtly after purchase. Kingston IIRC

Probably. Did they all go at once?

Ive had zero issues with any ssds since that one went. many years ago,

The BX is the budget version, so not surprised that it's slow. It also has lower endurance, but still plenty.

I have a pair of BX drives that have been running for several years with no problem. I'm relaxed enough, as they are in a mirror.

With benchmarks, it seems your choices are the obviously wrong ones (the old version of HDTune), or, the other tools give rather optimistic values and tightly clustered at that.

CrystalDiskMark seemed to produce a decent result. But it is less good than a graph style thing, for determining health.

The GnomeDisks bench in LinuxMint gives nice numbers, but the drives are awfully close to one another during read, and this is suspicious, because the NS100 drive included in the test, is normally "a few paces off the others". In practice, those three drives are not the same.

The seek dots seem to indicate a little wear around the edges, and that's probably the most indicative thing in the graphs.

The first bench at the top, I didn't let that finish, because it didn't look like a keeper. But even my "special" quiet-mode run using a Macrium Rescue CD as an OS, still could not get into the 500MB/sec domain.

Summary: Get yourself a copy of CrystalDiskMark and pretend all is grand.

Paul

Paul

What does the QNAP NAS look like, in terms of protocols ? Does the OS "like the taste of that connection" ? Or does it require some sort of special driver ?

There were at one time, NAS and NDAS, and the second one had a custom driver you installed on each PC.

Benchmarking tools, can work at a couple different levels. They can work at the physical (sector) level and require Administrator to do so.

Or, they can run at user level and work through the filesystem, which produces a pessimistic value for "speed". Going through a filesystem can be slower, to a lot slower.

I don't have a NAS here, so don't have any first hand experience with such.

As for testing recently purchased devices, I would not put them in the NAS at first. I would check them out in my desktop machine (assuming a port is available for connection) and then run my bench. On the assumption that, at least the connector on the NAS works in much the same way, and any other slowdowns on the NAS side, are caused by the software stack in the NAS. That's why when I ran the Linux test on my daily driver, I put the three SSDs on the three SATA cables, and the OS was a Live Session that I did a TORAM=yes on so that the distribution files were in RAM.

When TLC storage devices sit on the shelf at the store, the TLC gets "mushy", the ARM Cores are busy doing error correction on every sector, the situation looks "bad". But when the device is written once from end to end, that "makes the numbers look better". At that point, the sectors do not need correction and should be in better shape (for the next three months).

Paul