Kingston XS2000 2 TB

[W1zzard]

The Kingston XS2000 portable SSD uses the blazing-fast USB 3.2 Gen 2x2 interface to achieve impressive transfer rates of up to 2 GB/s. Internally, a brand-new controller from Silicon Motion is used; it combines the USB bridge and SSD controller in a single chip. Pricing is great, too, just $285 for 2 TB is more affordable than any competitor.

Show full review

 

[Wooden Law - Black]

The flash seems be 512Gb as density (“512” in the part number of the package) and 8-bit (dies’ pins for I/O traffic) as mode. I guess it is Kioxia BiCS4 like the KC2000 and KC2500.

 

[zeljans]

How long this device can keep data without plugging it into a PC? To use as a backup. That is also an interesting subject for me, to test out and analyze how long data can be stored (HDD, TLC, QLC, MLC, .....) without be corrupted.

 

[Wooden Law - Black]

How long this device can keep data without plugging it into a PC?

The data retention can change from the temperature of the flash, from the model of the flash, architecture, etc.

 

[BSim500]

How long this device can keep data without plugging it into a PC? To use as a backup. That is also an interesting subject for me, to test out and analyze how long data can be stored (HDD, TLC, QLC, MLC, .....) without be corrupted.

I definitely second the need to see an updated article testing this stuff on new drives in practise. The last time tech sites bothered to look was 2D planar MLC all those years ago...

 

[W1zzard]

How long this device can keep data without plugging it into a PC? To use as a backup. That is also an interesting subject for me, to test out and analyze how long data can be stored (HDD, TLC, QLC, MLC, .....) without be corrupted.

I definitely second the need to see an updated article testing this stuff on new drives in practise. The last time tech sites bothered to look was 2D planar MLC all those years ago...

Any suggestions how this could be tested? I have flash drives and SSDs here that work fine after not being touched for 5 years. As soon as you connect the drive it will start refreshing cells, so it will become useless for future testing

 

[BSim500]

Any suggestions how this could be tested? I have flash drives and SSDs here that work fine after not being touched for 5 years. As soon as you connect the drive it will start refreshing cells, so it will become useless for future testing

It's hard to say. There was one guy on Anandtech (just a forum member not a reviewer) who was putting data on SSD's, leaving then unplugged for months then after immediately plugging them in, using SSD ReadSpeedTester. After several months of being continuously unpowered, let's just say some of the data was quite interesting, with significant slowdowns to just 20MB/s when reading "stale data". The older the data, the slower the speed which seemingly indicates aggressive ECC kicking in to deal with potential errors caused by voltage drift.

I'd love to see new drives tested as I find it woefully dishonest of manufacturers to pretend SSD's are "better than HDD's in every way" for external storage unpowered for extended periods when I've had HDD's read back after 20 years fine, but I very much doubt QLC drives would hold up anywhere near the same. Although I've also had some flash drives read back fine, I've also seen a few others come back completely blank (ie, not even the file system / partition information survived voltage drift to 0 on all cells) after a few years of being unplugged. It's all down to the flash being used, process size, SLC / MLC having more overhead than TLC / QLC, etc.

Edit: As for review advice, whatever you do you basically put "known" (non random) data (eg, large video files or RAR files with recovery volumes / PAR files), on that's MD5'd and have a copy of those .MD5 files backed up onto another drive. That way it at least makes it easier to verify data integrity / loss.

 

[W1zzard]

significant slowdowns

Maybe this was just Windows housekeeping that kicked in?

leaving then unplugged for months then after immediately plugging them in

and at that point the drive gets refreshed and will be unusable for further testing. Guess that means I have to buy 20 flash drives?

 

[BSim500]

Maybe this was just Windows housekeeping that kicked in?

It's hard to say. It's definitely one of the most difficult things to test since SSD's do stuff (garbage collection, TRIM, periodic refresh / wear levelling, etc), "internally" that the OS never sees. All I know is if I were to look for an external SSD for data backup that will be unplugged for long periods, I wouldn't touch QLC's with a barge pole and would be actively seeking out an older MLC if possible...

SLC (1 bit per cell) = 100% vs 0

MLC (2 bit per cell) = 100% vs 66% vs 33% vs 0

TLC (3 bit per cell) = 100% vs 86% vs 71% vs 57% vs 43% vs 28% vs 14% vs 0

QLC (4 bit per cell) = 100% vs 93% vs 87% vs 80% vs 73% vs 67% vs 60% vs 53% vs 47% vs 40% vs 33% vs 27% vs 20% vs 13% vs 7% vs 0%

PLC (5 bit per cell) = 100% vs 97% vs 94% vs 90% vs 87% vs 84% vs 81% vs 77% vs 74% vs 71% vs 68% vs 65% vs 61% vs 58% vs 55% vs 52% vs 48% vs 45% vs 42% vs 39% vs 35% vs 32% vs 29% vs 26% vs 23% vs 19% vs 16% vs 13% vs 9% vs 6% vs 3% vs 0%

...not out of paranoia, but basic laws of physics. All else (eg, process node) being equal, MLC's can handle a voltage drop of 33% whilst QLC's can't even handle 8% (and PLC not even 4-5%) before data gets altered. Not very reassuring for backup drives...

 

[duynguyenle]

I would be interested to see if you can work some sort of thermal assessment into your future testing (e.g temperature graph of the controller and NAND packages under long duration workload like the full drive fill). IIRC, these fast SSDs wrapped into a USB stick form factors don't always pay adequate attention to thermal design and dissipation (possibly because they're banking on the use case being quick bursts of writes, not full drive fill or stress tests).

I'd like to know what the internal component temperatures are like after a quick write task vs a long write tasks and any form of thermal throttling/temperature management if possible.

I don't think I've seen many people look into this, and unsure if this is even a problem at all, just basing my comments on this video I saw a while ago when a YouTuber took his defective SSD apart:

 

[W1zzard]

All chips involved have thermal protection, so worst case you lose a bit of performance after hammering it with 100 GB+ in writes. As mentioned, thermals were fine, handwarm with heavy write load.

I can certainly stick a thermal probe to the outside of the drive on future reviews and mention the temperature in the review