Sorry to hear that. I've got a 1 TB one as my game drive. So far so good and about 2 years in with it. It stays loaded pretty close to max, I try to keep 100 MB free every now and then I have to do some clean up to get rid of something to get it back down.
I don't remember hearing about Intel recently. Good to know. They've been one of my go to for things important. The only SSDs I've really had problems with have been Kingston. Otherwise, I've seen a couple of cheap sandisk go but that was 1 or 2 out of 50+. At home I've got Toshiba, Sandisk, Samsung, Intel, and that Sabrent that are all still doing well.
I agree though, it's still a little too early to say with addlink but I checked out their Amazon store page and they've got some impressive looking PS5 Gen4x4 SSDs. Pricey but quality does cost. The heatsinks look sensible but beefy.
I don't really know much about them. I presumed (maybe incorrectly so) that they were just another brand putting their name on generic designs by Phison or some other controller maker. I mean, that's what Sabrent did. It was pretty much a reference Phison E16 design with Sabrent stickers on it.
The Sabrent drive saw mostly normal desktop use on my machine. Only very infrequent heavy writes, a good chunk of idle time, and being shut off every night. I bought it in December 2019 for my Threadripper build as it was the only one on the market I could find that had Gen4 capability, and folks on the Hardforums seemed very enthusiastic about the previous Phison controller, the E12, so I figured chances were good the E16 would be good as well.
That said, I was a bit uneasy about not going Samsung/Intel at the time, and maybe I was right and should have listened to myself
The Sabrent drive died with no warning at all. After 2.5 years it was barely broken in from a write cycle perspective, so that wasn't what did it. It shut down normally one night, and the next morning was not detected in the BIOS. I tried it in a few other computers as well as in a thunderbolt external enclosure, to no avail. Thing was a paperweight. No rescuing of data was possible.
I didn't lose anything i couldn't replace. At least not except game progress for games that don't sync saves to Steam. I always save my files on my redundant and backed up NAS. I forgot how long it takes to install, set up, and configure things the way I like them though. One bare metal Win10 install (my game install) one Linux install (my daily driver) and one Win 10 VM in Linux (which I use for work). Setting everything up, installing all the software I like, and configuring everything to play nice took me **** near an entire weekend. The one I was the most annoyed about was losing all of my optimized GPU overclock settings.
I guess I did lose one irreplaceable thing. My time.
Anyway, going forward I am going to try to remember to image my drives more often. I wish I could schedule something for shutdown time that would do a differential snapshot and block level backup of the drive images to my NAS. Sort of how Apples Time Machine works. That's the one thing on Apple I really appreciate. Its a good product, and I am constantly in disbelief that something similar doesn't exist for Linux or Windows.
(I mean, there is a Linux software called time shift which does something similar, but it refuses to back up to remote drives, which renders it mostly useless)
Anyway, my history with SSD's is something along these lines.
I bought my first one in 2009, an OCZ drive. 120GB Agility I think. I was so mesmerized with the awesomeness of SSD's that I started buying them for every machine I could. The experience was transformative. Every last OCZ drive I bought (and there were a bunch of them) and every last RMA replacement I received all died in less than 2 years. Yet for some reason I kept buying them for a while. Not sure why. Stockholm syndrome? I guess I was convinced that reliability issues were just a part of the cost to pay for awesome performance.
Actually, thinking a little bit more about it, I may have to take that back. My LAST OCZ RMA in 2015 from when I RMA's my 256GB Vertex 4 is still alive. I used it as a secondary drive for like a year after receiving it back, but then took it out and put it in a box, as I didn't trust it, and never used it again. It was of a newer post-bankruptcy and Toshiba OCZ acquisition though so it was probably more reliable.
When the initial Vertex 4 drive died in 2015, I didn't want to wait for the RMA process to conlcude to get my machine up and running again, so after some brief research, I pulled the trigger on a 400GB Intel 750 SSD. It is to this date one of the very few NVMe drives that actually had a traditional boot-ROM on it, so it could boot on motherboards without m.2 compatibility. Worked like a charm on my x79 motherboard with my i7-3930k.
That drive is still with me today, in my threadripper as a tertiary drive.
At some point in 2017 I think I added a 1TB Samsung 970 EVO as a secondary drive to my old x79 system, stuck into a PCIe to M.2 adapter card. I still had to boot from the Intel drive, but the 970 EVO made for a nice secondary game drive.
This drive is also still in my threadripper as a secondary drive.
When the Sabrent drive died, I hurriedly bought ymself a 2TB Samsung 980 Pro to get back up and running quickly. (Amazon had it with 1 day prime delivery!) When the RMA drive comes back, it will also go back in there, so I will have four NVMe drives in it:
1.) 2TB Samsung 980 Pro
2.) 2TB Sabrent Rocket 4.0
3.) 1TB Samsung 970 EVO
4.) 400GB Intel SSD750
It might actually be time to retire the intel drive. I think 5TB of NVMe storage on the desktop is way more than enough. The 400GB drive just won't be adding much value. I may move it back to my x79 system, which I use as a testbench system in my office. According to Smart stats, after being in use since 2015 (in non-primary drive use since 2019) it still has 60% write cycles remaining...
Code:
=== START OF SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED
SMART/Health Information (NVMe Log 0x02)
Critical Warning: 0x00
Temperature: 39 Celsius
Available Spare: 100%
Available Spare Threshold: 10%
Percentage Used: 40%
Data Units Read: 68,405,508 [35.0 TB]
Data Units Written: 39,501,997 [20.2 TB]
Host Read Commands: 953,743,755
Host Write Commands: 693,024,613
Controller Busy Time: 42
Power Cycles: 2,260
Power On Hours: 19,379
Unsafe Shutdowns: 317
Media and Data Integrity Errors: 0
Error Information Log Entries: 0
Error Information (NVMe Log 0x01, max 64 entries)
No Errors Logged
Anyway, that summarizes the desktop SSD history.
Outside of that, I've purchased a stupid number of SSD's when they have been on sale.
In 2014 I bought a couple of 128GB Samsung 850 Pro SATA drives which I used striped as cache drives for the ZFS pool on the server. In 2014 the server also got dedicated SLOG/ZIL drives. (they aren't write cache, but they speed up writes, it's complicated) Two mirrored 100GB SATA Intel S3700 drives.l At some later point, when 512GB 850 Pro drives were on sale, those cache drives were replaced with the 512GB drives, and moved to other roles. Once got beaten to **** as a live TV drive for my MythTV VM, the other served as a dedicated swap drive for the server for years. They finally came out and were used as laptop boot drives for a few years, before being replaced in the laptops when the 512GB drives came out of the server, at some point being replaced by 1TB drives.
Rinse and repeat many times. That server has had SSD's go in and out of it like it is going out of style, whenever there has been a good sale. The last SSD upgrade was the aforementioned NVMe drive upgrade lat elast year when I bought three four way M.2 to PCIe adapters, and stuck 12x NVMe drives (ranging in size from 256GB up to 2TB depending on the application) in there. It still has two old 512GB Samsung 850 EVO drives in there, mirrored for redundancy for booting/OS use, as the old server board can't boot from m.2 without a BIOS modification, which I don't think is appropriate for a server, which I want to be stable.
During the server years I also built several smaller single purpose systems. KODI HTPC boxes (which all got small 64GB and 32GB SATA ssd's, as they didn't need storage, they just needed to boot) a pfSense box which also got a small boot drive. I also bought various drives for family and friends systems. You know, the everyone thinks you're a wizard because your on sale SSD and RAM upgrade took a slow old laptop and magically made it feel new again. My better half got a desktop built after her iMac died, which also got a Samsung 970 EVO drive. My stepson's system at first used an old Server SSD, but when it ran out of space, I upgraded him to a 1TB SATA Samsung 850 EVO, and at some point I upgraded him to a 1TB 970 EVO as well, moving the sata drive to a secondary role on his build.
I honestly lose track of my SSD purchases, but if I do a search in my gmail for emails containing the words "tracking" and "SSD" and start counting, I start getting up somewhere around 40 or 50 of them. The prevailing theme is as follows:
- Everything Samsung and Intel, regardless of age, and how much it was beaten to crap in the server is still with me and working fine in some repurposed second, third or fourth life. Some are in the server, some in the desktop, some repurposed in my laptops, some in friends and family machines. etc. etc. These two brands have been bullet proof for me.
- 100% of the early OCZ drives are dead, and have been dead for many many years.
- The one non-Intel/Samsung drive I trusted in my desktop since the OCZ days, the 2TB Sabrent Rocket 4.0 is also dead.
- The 12x Inland Premium NVMe drives in the server are still going strong, but they are relatively young. The smallest of the batch, the 256GB drives (mirrored and used for VM drive images) have been consuming write cycles faster than expected, but at the rate they are going they should still be serviceable for another 6-7 years, and by the time they are worn out theri replacements should be dirt cheap.
I guess that is all the wisdom I can provide on this subject.
I don't have HUGE statistical sample sizes, but I think my n=40 to 50 somewhere is probably larger than most individuals anecdotal experiences, so take it for what it is worth.
****, that was more dense text than expected. I hope someone gets value out of this post
