Lexar NM790 4TB nvme no self-test & highly variable speeds. ok or not?

Experiences with hard disks, SSDs, USB devices, hard disk controllers, motherboards and so.
wwcanoer
Posts: 16
Joined: 2020.02.19. 16:55

Lexar NM790 4TB nvme no self-test & highly variable speeds. ok or not?

Post by wwcanoer »

What should I think of this drive?
Lexar NM790 4TB M.2 2280 PCIe Gen4x4 NVMe SSD
PN: LNM790X004T-RNNNG
Computer: Lenovo X1 Extreme Gen 1 20MF0015AD 2019

(Q1) There is no short or extended self-test.
Is this common or does it indicate a problem with the drive?
(I've never had a drive without this function.)
Error: "Error while starting the test. The error code is: NVMe-TB E: 1. The hard disk controller or its driver do not support this function.
I don't see any updates on the Lexar website and Win11 auto search for the latest driver says that the latest is installed.

I see an old response:
"Yes, I can confirm that the problem is related to the driver of the disk controller where the hard disk is connected.
Some drivers of disk contrllers do not support special functions like performing self-tests, adjusting acoustic level and so - and in this case, this or similar error may be displayed.
The E: 1 means "incorrect function" reported by the driver when the appropriate command is attempted - so the driver does not recognise the commands as a valid one (even if the hard disk supports that).
This is completely independent from the software and the hard disk itself: the driver blocks the command."

So, I assume that it's just one place they cut corners, but I'm surprised that a modern NVMe drive doesn't have this function.

(Q2) One test had very bad results, got one BSOD, others tests passed ok.
Is there something wrong? Should I do more testing?

1. Random Seek Test: 0.02 to 0.07 ms
2. Write + Read Test Random sequence, random data: run for 3 minutes, avg 1200 MB/s (800 to 1600), looks good.
3. Crystal Disk Mark speed tests: Passed, reasonable speeds.
4. Write + Read Test Random sequence, random data: full drive 2:26 hrs total, write avg 1100 MB/s (900 to 1300) read avg 1200 MB/s (1100 to 1300) Top half of the write chart has many darker squares.
5. Write + Read Test sequential, zeros: Write started at 1200 MB/s; at 100 GB dropped to 600MB/s; at 280 GB dropped to 124 MB/s then my cat stepped on the keyboard and computer crashed with BSOD 12 minutes into the test. (From Win logs: "The computer has rebooted from a bugcheck." but I can't decipher the cause.) I doubt that the cat caused the BSOD. I haven't had any BSOD problems with this computer.
Screenshot (1469).png
Screenshot (1469).png (257.58 KiB) Viewed 1818 times
6. Write + Read Test sequential backward, zeros: stopped at 2 min because only 138 MB/s! would run 22 hours.
Screenshot (1489).png
Screenshot (1489).png (251.86 KiB) Viewed 1818 times
7. Write + Read Test Random sequence, random data: run for 1 minutes because again only 120 MB/s!
Screenshot (1493).png
Screenshot (1493).png (250.2 KiB) Viewed 1818 times
8. Reinitialized and formatted the disk in Disk Mamagement.

9. Write + Read Test Random sequence, random data: ran for 1 minute. Now back to 1600 Mb/s!

10. Write + Read Test sequential backward, zeros: Write was a consistent 1700 MB/s (only 37 min to write 4TB!) but Read test varied between 800 and 1450 MB/s
Screenshot (1514).png
Screenshot (1514).png (286.67 KiB) Viewed 1818 times
Screenshot (1533).png
Screenshot (1533).png (252.23 KiB) Viewed 1818 times
Is this kind of variation normal for an nvme drive?

11. Write + Read Test sequential, zeros: write (56 min) mostly 1200 MB/s. read took 80 min, varying between 800 and 1400 MB/s
Image
Image
(seems to be max 5 attachments, so I will add these in the next comment.)

I'm surprised that there's so much variation in the read speed. I was not doing anything on the computer during these test but of course Windows could decide to download updates or virus scan or whatever in the background but I doubt anything that would significantly affect the test.

This is my only computer with NVMe, so I can't separate computer vs drive issues. I'll go back to look at the tests of my 970 Evo Plus 2 TB last year that I recall as being very consistent. I will copy that drive to this one and then test to see if it is still consistent.

Should I be concerned that there's some incompatibility with my computer or problem with the drive?

Should I further test by running reinitialize disk surface? Or other test?
wwcanoer
Posts: 16
Joined: 2020.02.19. 16:55

Re: Lexar NM790 4TB nvme no self-test & highly variable speeds. ok or not?

Post by wwcanoer »

(repeat with the last two images)
11. Write + Read Test sequential, zeros: write (56 min) mostly 1200 MB/s. read took 80 min, varying between 800 and 1400 MB/s
Screenshot (1548).png
Screenshot (1548).png (262.29 KiB) Viewed 1814 times
Screenshot (1562).png
Screenshot (1562).png (260.84 KiB) Viewed 1814 times
I'm surprised that there's so much variation in the read speed. I was not doing anything on the computer during these test but of course Windows could decide to download updates or virus scan or whatever in the background but I doubt anything that would significantly affect the test.

This is my only computer with NVMe, so I can't separate computer vs drive issues. I'll go back to look at the tests of my 970 Evo Plus 2 TB last year that I recall as being very consistent. I will copy that drive to this one and then test to see if it is still consistent.

Should I be concerned that there's some incompatibility with my computer or problem with the drive?

Should I further test by running reinitialize disk surface? Or other test?
wwcanoer
Posts: 16
Joined: 2020.02.19. 16:55

Re: Lexar NM790 4TB nvme no self-test & highly variable speeds. ok or not?

Post by wwcanoer »

12. Reinitialization test: Took 12 hours Started at 260 MB/s and later dropped to 80 MB/s (read) and 240 MB/s (write).
Screenshot (1623).png
Screenshot (1623).png (302.96 KiB) Viewed 1806 times
Screenshot (1625).png
Screenshot (1625).png (312.09 KiB) Viewed 1806 times
It passes again, so I guess that it's working ok despite the earlier problems with tests 5, 6 & 7 (above).

In comparison, my Samsung 970 EVO Plus 2TB tested reinitialization at a consistent 200 MB/s last year. Definitely a faster and more consistent drive.
20220612-162918_RI_Samsung_SSD_970_EVO_Plus__S4J4NM0RB08274Z_2B2QEXM7-surface-full-st.jpg
20220612-162918_RI_Samsung_SSD_970_EVO_Plus__S4J4NM0RB08274Z_2B2QEXM7-surface-full-st.jpg (505.04 KiB) Viewed 1806 times
20220612-162912_RI_Samsung_SSD_970_EVO_Plus__S4J4NM0RB08274Z_2B2QEXM7-surface-full.jpg
20220612-162912_RI_Samsung_SSD_970_EVO_Plus__S4J4NM0RB08274Z_2B2QEXM7-surface-full.jpg (634.6 KiB) Viewed 1806 times
wwcanoer
Posts: 16
Joined: 2020.02.19. 16:55

Re: Lexar NM790 4TB nvme no self-test & highly variable speeds. ok or not?

Post by wwcanoer »

FYI, After the reinitialization test on 4TB, I was surprised that the drive was partitioned but I assume that's normal.
Screenshot 2023-11-02 113021.png
Screenshot 2023-11-02 113021.png (196.77 KiB) Viewed 1805 times
But easily changed back to a single GPT drive.
Screenshot 2023-11-02 113326.png
Screenshot 2023-11-02 113326.png (57.55 KiB) Viewed 1805 times
wwcanoer
Posts: 16
Joined: 2020.02.19. 16:55

Re: Lexar NM790 4TB nvme no self-test & highly variable speeds. ok or not?

Post by wwcanoer »

I ran three more tests of read only (sequential, butterfly and random) that all consistently ran just below 3000 MB/s.

I partitioned 93% since I don't know how much overprovisioning is built into this drive, and now to start filling it as a data drive.
User avatar
hdsentinel
Site Admin
Posts: 3019
Joined: 2008.07.27. 17:00
Location: Hungary
Contact:

Re: Lexar NM790 4TB nvme no self-test & highly variable speeds. ok or not?

Post by hdsentinel »

(Q1) There is no short or extended self-test.
Is this common or does it indicate a problem with the drive?

No, this is not common (but can happen) - and it does not indicate any problem with the drive itself.

In general, the short self test, extended self test functions are the "built-in" hardware self test functions of the disk drives (hard disks or SSDs).
To access these tests, special functions used - which may be or may not be supported by the drive itself.
It is also possible that these functions (even supported by the SSD itself) may be blocked by the disk controller driver or something else (for example an USB-NVMe SSD adapter if used).

Not sure, but there is good chance that this happened: even if the SSD may support this function, it is blocked or "just" Hard Disk Sentinel did not receive the expected result after the test started.

I assume the SSD supports the Device self test, Short self test, Extended self test functions (you can verify on the Information page).
The Disk Controller section on this page (on the top) shows the current disk controller and its driver version. Not sure, but probably this is the Standard NVMe SSD controller driver of Windows.

It is also possible that the SSD (even if supports the self test function) may not provide the result (test status/progress) correctly.
You can verify: please use Disk menu -> Device Specific Information function which attempts to read the internal self test log / error log of the SSD.
This (like the self test) is a special function too, so everything above is applicable: the controller /driver may block accessing the result.
Ideally it may show that the self test started and successfully completed.

The possible "troubles" around hardware self test mentioned in the Help:
https://www.hdsentinel.com/help/en/62_testfaq.html
In some cases, these hardware tests (Disk -> Short self test, Disk -> Extended self test) are not available, not supported or they result in an error quickly even in relatively low number of problems. No further information is returned about the result, for example it is not possible to list the sector(s) which are damaged. In such case, an appropriate software testing method is required.


So yes, it is better to use the Disk menu -> Surface test functions in such case, exactly because these are more sensitive: reports possible variation of the speed (in addition to possible errors).

I suggest to use Report menu -> Send test report to developer option. Then it is possible to check the "raw" response of the SSD during detection / execution of the functions so it may give further ideas, thoughts.


> (Q2) One test had very bad results, got one BSOD, others tests passed ok.
> Is there something wrong? Should I do more testing?

It is possible that the SSD not installed correctly, may be it was not secured properly (?)
Or maybe reached VERY high temperature and disappeared from the system (?)
In other cases, BSOD should never happen of course.


> 3. Crystal Disk Mark speed tests: Passed, reasonable speeds.

Yes. This is the problem with such quick benchmark tools: they DO NOT properly test the complete SSD, just test a very small fraction.
Manufacturers know this and they only need to make a cache large enough to hold the "data" during the test executed and write data only later.
So these tools (while nice) can only provide very high random fake numbers - but not the real performance of the SSD - as the actual write to the SSD happens only after the test result displayed.

> 4. Write + Read Test Random sequence, random data: full drive 2:26 hrs total,
> write avg 1100 MB/s (900 to 1300) read avg 1200 MB/s (1100 to 1300) Top half of the write chart has many darker squares.

Yes. This is completely normal and expected - as this way you could verify the REAL performance of the drive, including possible drops of performance when the cache filled. Generally the darker green blocks show that the SSD is slower than in previous areas. Most SSDs have such / similar results or even periodic light-green + dark-green blocks alternating.

If you search the word for "darker" in this forum, you may see several results, showing "weird" patterns on SSDs: these confirm that the SSD seems fast (during writing data only to its cache) and then become slower (during flushing the cache, actually performing the write).

Some examples:
https://www.hdsentinel.com/forum/viewtopic.php?p=19743
https://www.hdsentinel.com/forum/viewtopic.php?p=18736
https://www.hdsentinel.com/forum/viewtopic.php?p=17002

On NVMe SSD, the performance drop is more noticeable.

This does not mean any problem with the SSD itself, it is completely normal, all SSDs work this way.
Manufacturers prepared for this, this is why the product says eg. "up to 4000 MByte/sec" but they never mention what is the LOWEST possible speed.


> Is this kind of variation normal for an nvme drive?

YES. Completely normal and expected.

The test should show no yellow / red blocks. This is the only important. Speed change is important if you prefer to use the drive for write-intensive application.


> Should I be concerned that there's some incompatibility with my computer or problem with the drive?

Probably not, but to say for sure, you can use the mentioned Report menu -> Send test report to developer option as then it is possible to check the driver installed which can cause troubles.
User avatar
hdsentinel
Site Admin
Posts: 3019
Joined: 2008.07.27. 17:00
Location: Hungary
Contact:

Re: Lexar NM790 4TB nvme no self-test & highly variable speeds. ok or not?

Post by hdsentinel »

> In comparison, my Samsung 970 EVO Plus 2TB tested reinitialization at a consistent 200 MB/s last year. Definitely a faster and more consistent drive.

Yes. This is absolutely true: according the experiences, yes, it is more consistent in terms of write speed.
User avatar
hdsentinel
Site Admin
Posts: 3019
Joined: 2008.07.27. 17:00
Location: Hungary
Contact:

Re: Lexar NM790 4TB nvme no self-test & highly variable speeds. ok or not?

Post by hdsentinel »

Slightly different thing but if you allow, just for illustration purposes: a SATA Lexar NQ100 240 GB SSD:

Image

Image

As you may noticed, during the write test the SATA SSD (advertised as "up to 550 MByte/sec" of course...) shows 300-350 MBytes/sec for a very short time and then the performance drops to 1/10 of the "start" speed, much slower than the "up to..." value.
Post Reply