With around 140 hdparm submissions after our original post and an absolute myriad of drives tested, we learned quite a bit. We originally posted the challenge after seeing the quick read test in some Intel documentation as a quick and dirty way to validate drives were performing approximately as expected. One of the biggest things we saw were a number of SSDs performing quite poorly.
I posted on the interim results about the phenomenon we saw with two SanDisk SSDs but we received no less than five notes from readers who ran the test and saw the same issues. It seems like a secure erase was fixing the problem, but great to know that this is happening on “enterprise” SSDs. Just to give one an idea of the magnitude of the issue, 5 drives of 140 or so tested means that approximately 3.5% of the drives users tested and across different brands and drive capacities.
After receiving a solid data set we wanted to make it available to anyone. We have put the numbers into an Office 365 Excel sheet. Our advice is to copy the data and cut it as you please.
Here is a link to the raw data. We did do a bit of cleanup on the data. We also removed the usernames from the data. The result is a nice resource with a wide variety of disks that all were issued the same two commands. While this is far from a reliable indicator of performance, it does give one a ballpark idea regarding relative read speeds. It is also a quick test to run in order to determine if your SSDs are running at an appropriate level of read performance. The 3.5% of drives we saw with significant issues were turning in results less than 1/3 of what we expected. Those were easy to spot in terms of issues with the drives.
As one can see we did see the most dries from:
Intel
Samsung
Crucial
Western Digital
Seagate
Which is interesting because the majority of results were SSD tests. We also had about 9% of the population already showing as NVMe SSDs which is certainly higher than the average population out there.
Let us know if you like these little community tests and results and how we can improve the process in the future. If there is something you want to see let us know as well either in comments or in the forums.
As expected, the Samsung XS1715 is a much better read ssd, while intels are better write ssds. The thing I don’t understand about Samsung’s recent management is this: when you have the best read SSD on the market(XS1715), why couldn’t you be more price transparent and let people buy it on newegg or amazon instead of having it as a DELL OEM only product. It is a mistake to have the entire Samsung marketing department focusing on consumer TLC markets and 1TB+ models.