Iometer

I would recommend you make read only sequential workload for testing purposes. It will generate plenty of heat still and wont wear the drive much. Sequential write draws the most power and will obviously generate more heat, but think how much data you will be writing so fast and how realistic a 30 min+ write is......you need to take a logical approach to this. 30+ minutes? You realize at 1 minute you will have written 120GB? 10 minutes is 1.2TB...if you exceed the drives capacity in testing, it is illogical if doing writes, unless you are mimicking a cache buffer type workload, but then that you would just set up in Iometer as a mixed workload. You will get throttling in like a minute or less if you have crap airflow. With ample airflow it could be a few minutes...

With Iometer you can test with or without a partition too (drive just need to be unformatted and MBR partition table). I can go into detail more so if you need help. You will only want 1 worker for sequential workloads and a QD of 32, 128, or 256 will work good. Suggestion, test workload on HDDs and get them perfect before you run them on the SSD.

Overall, just to get to testing your cooling...you shouldn't need 30 min+, overheating should occur a lot faster than that.

 

Quote:

Originally Posted by s1rrah

Hi ...
I have all the popular disk benchmark programs (HDTune, Crystal, etc.) ... but none of these run long enough.

Any suggestions?

just create a 1 GB dummy file (dd or similar).

Just write a simple script that loops through 100 times or so, moves the 1gb file to and from the ssd.

hdd stress test.bat

Code:

for /l %%x in (1, 1, 100) do (
        move my1gbfile.dat E:\my1gbfile.dat
        move E:\my1gbfile.dat my1gbfile.dat
)

place the batch file and 1gb test file anywhere on another drive.
change E: to your m.2 ssd drive letter.