Question Crucial BX500 SSD 1TB - Sick or not?

[Malcolm L]

Hi, I'm hoping someone can help me interpret some write performance tests against my Crucial BX500 SSD CT1000BX500SSD1 (1TB).
Bought brand new in Jan this year.

I have the SSD as a secondary SATA drive and I run a Windows VM on it via Hyper-V. The VM has has suddenly gone slow to the point of unresponsiveness and after various experiments I've pinpointed that my Crucial SSD is the most likely culprit. So I'm trying to run performance tests against the drive to prove that it's problem but I need some help interpreting the results.

I've now got CrystalDiskMark v8, ATTO Disk Benchmark v4, and Hard Disk Sentinel v5.7 installed.
SMART results show 97% Health.

CrystalDiskMark results on my machine (using 1GiB total size, and 1M block size for sequential tests):
* Sequential Q32T1: 544MB/s read, 15MB/s write
* Sequential Q1T1: 486MB/s read, 23MB/s write
* Random 4K Q8T8: 157 MB/s read, 15 MB/s write
* Random 4K Q1T1: 25 MB/s, 13MB/s write

In contrast, results from Tom's Hardware review (https://www.tomshardware.com/reviews/crucial-bx500-ssd,5377-4.html):
* Seq Q32T1: 561MB/s read, 337MB/s write
* Seq Q1T1: 488MB/s read, 442MB/s write

My write speeds seem significantly worse.
This seems to be explained by the transfer rate graph produced by Hard Disk Sentinel, which shows a clear trend of about 4 to 5s burst at circa 500MB/s followed by the majority of the transfer at about 20MB/s.

I'll follow-up with some screenshots.

Is this reasonable behavior for this SSD or indicative of problems?

 

[Malcolm L]

Here's the CrystalDiskMark test while running the first of its write tests.
In the Disk Sentinel performance chart in behind, you can see that the transfer rate hit a 4-5 sec burst around 500MB/s and then slowed down to the approx 15MB/s speed that we got as the average.

Here's the ATTO benchmark tests, transferring 256MB files using different block-sizes:

In contrast, the review showed these results for Sequential writing - see the red line:

 

[Bob.B]

Here's the CrystalDiskMark test while running the first of its write tests.
In the Disk Sentinel performance chart in behind, you can see that the transfer rate hit a 4-5 sec burst around 500MB/s and then slowed down to the approx 15MB/s speed that we got as the average.

Here's the ATTO benchmark tests, transferring 256MB files using different block-sizes:

In contrast, the review showed these results for Sequential writing - see the red line:

Show a screenshot from crystaldiskinfo.

Wag.....is the optimize app enabled for this ssd?

Device manager/disk drives/policies test with write caching on/off see if it makes a diff.

 

[Malcolm L]

CrystalDiskInfo:

Results with write caching disabled look very similar to the results from before where it was enabled:

The only real difference is that you don't get the initial write-cache advantage. Only on the very first write did I see a higher speed 4 to 5 sec burst. This time only at about 250MB/s:

 

[Malcolm L]

Ok, this I think has to be the clincher - I've run CrystalDiskMark with the smallest file size (16MiB) - which should give the best results because even 5 writes of that should still fit within its write cache. But the results are no better:

Here's the problem. Still exactly the same appearance of a short burst of high speed writing. And again here it only occurs for the very first write test.

 

[Bob.B]

Ok, this I think has to be the clincher - I've run CrystalDiskMark with the smallest file size (16MiB) - which should give the best results because even 5 writes of that should still fit within its write cache. But the results are no better:

Here's the problem. Still exactly the same appearance of a short burst of high speed writing. And again here it only occurs for the very first write test.

Optimize app enabled for this disk?

Just for yuks try a different sata port on the mobo.
Try a different sata cable.

 

[MWink64]

I'm not surprised. This seems to be a common issue/flaw with the Crucial BX500. I speculate it may be due to rapid degradation of the data in NAND, resulting in very slow reads, which can also affect writes because of how flash media works. My BX500 produced the same confusing, inconsistent benchmark results. I'd suggest using HDDScan and running the (sequential) Read Data to Host test. In the details, look at (and preferably post a screenshot of) the Graph tab (not the default Map tab). I'll bet there are some deep valleys. Normally, a SATA SSD will read somewhere in the ballpark of 500MB/s, from beginning to end. Drives affected by this issue have huge drops, sometimes into the single digits.

In my experience, the only way to resolve the issue is to rewrite all the data, and even then it's only a temporary fix. It amazes me that Crucial has continued to sell this seriously flawed drive and that it has persisted through major hardware revisions. Mine is the old TLC version, from 5 years ago.