The SSD Endurance Experiment: Finally, They Are All Dead

[dfsdfgfkjsefoiqzemnd]

So basically with normal use an SSD will outlast a HDD.

[patrickjp93]

840 EVOs are nothing but trouble, rave reviews out the gate, but a year down the road they've slowed down to less than mechanical speeds

 

Samsung's first fix takes care of 95% of that. I'm sure they'll get the last of it figured out. Also, you can just re-save old data and it'll be back up to speed.

[martinrox1568]

Nice test, but how much do I use per day is my question

Does anybody know about an application that could monitor the data written to a specific dry from when I push the button on my pc to when I'm about to shut it down for the night ?

It really doesn't matter. A pettabyte is 1000tb, and 1,000,000gb. You can reinstall your os everyday for the next 39 years and be fine. And that is considering that everytime you reinstall you install all the apps you use such as browsers and clients and some games that total together to 70gb writes per day. That is only considering 1 petabyte write that most of the drives exceeded by far

[Doobeedoo]

So basically with normal use an SSD will outlast a HDD.

Yeah.

I've seen some enterprise HDDs rated for like more than half of PT per year workload, not sure when they'd die tho.

While SSDs can last long what this thread shows good, they "burn" faster (ofc have # of cycles they can write) cause they are faster so you can burn them way quicker than HDD if you'd want to. But still while endurance of SSDs gained significant boost over time it's not there yet where you could get one and set it to endurance test and that it could last like that for 10y. But yeah, I'm pretty sure couple of years they'll be over top grade HDDs in terms of endurance for like data centers.

One thing I always asked my self... how come SSDs can't be like RAM, meaning have unlimited # of reads/writes. When will we see those

[Humbug]

One thing I always asked my self... how come SSDs can't be like RAM, meaning have unlimited # of reads/writes. When will we see those

Cause RAM is volatile?

No flash.

[zacRupnow]

And everyone says newtron ssd's are bad

#CORSAIR

not Newton, Corsair in general.

[KenJimmY]

Nice test, but how much do I use per day is my question

Does anybody know about an application that could monitor the data written to a specific dry from when I push the button on my pc to when I'm about to shut it down for the night ?

would like to know an answer to this myself .. tho .. i'll be checking with google soon if there's still no answer

[TooManyErrors]

Seems like I'll be replacing my 850 Pro due to speeds rather than hardware failure. 

[-CLASSIFIED-]

this eases my nervousness with my two kingston v300 120gb ssds , one with 3.5tb written on it , and the other tho older with only 2.8tb written to it 

[sylarspock]

Seems like I'll be replacing my 850 Pro due to speeds rather than hardware failure. 

 

what speeds are you getting ?

 

[Sakkura]

would like to know an answer to this myself .. tho .. i'll be checking with google soon if there's still no answer

 

Plenty of options. I like the simplicity of SSD Life, though I'm sure eg. Crystaldiskinfo has more features.

[Humbug]

Seems like I'll be replacing my 850 Pro due to speeds rather than hardware failure.

It may fail due to other reasons since all hardware including HDDs fails eventually... But it probably won't fail due to you writing data to it; which is what so many people are irrationally afraid of when it comes to SSDs.

[That Norwegian Guy]

840 EVOs are nothing but trouble, rave reviews out the gate, but a year down the road they've slowed down to less than mechanical speeds

 

For some people, sure. I believe it was 25% of drives.

 

I've written 55TB to mine, and it reads at 525MB/s and writes at 328MB/s.

 

That said, I plan to retire it from my main rig and stick it in my game server soon.

[TooManyErrors]

what speeds are you getting ?

 

Last time I checked I think it was around 550 - 575 MBPS. I'm more just referring to when SSDs get even faster.

[Paragon_X]

Nice post. This means the 850 Evo I got, might migrate in my next build in a couple of years.

[sylarspock]

Last time I checked I think it was around 550 - 575 MBPS. I'm more just referring to when SSDs get even faster.

 

oh, my bad.

Well, I see m.2/mSata or PCIe slot SSDs as the future forward

Most SSDs have more or less achieved a saturation point on the SATA interface

 

I have more or less decided that my Win 10 OS drive be a PCIe slot based SSD.

850 Pro for now while I run Win 8.1

[Sakkura]

For some people, sure. I believe it was 25% of drives.

 

I've written 55TB to mine, and it reads at 525MB/s and writes at 328MB/s.

 

That said, I plan to retire it from my main rig and stick it in my game server soon.

 

Sounds like you haven't understood the issue, since you mention writes. How are you measuring the drive's speed?

[That Norwegian Guy]

Sounds like you haven't understood the issue, since you mention writes. How are you measuring the drive's speed?

 

Samsung Magician of course. And when I mention reads I may as well mention writes, as it measures both at once.

[Sakkura]

Samsung Magician of course. And when I mention reads I may as well mention writes, as it measures both at once.

 

Samsung Magician doesn't test in a way that will show the problem, even when it is affecting the drive.

[Doobeedoo]

Cause RAM is volatile?

No flash.

Ye, I kno', just, wondering when will we get storage like that. That powering it off keeps data, and power failures doesn't affect it.

[That Norwegian Guy]

Samsung Magician doesn't test in a way that will show the problem, even when it is affecting the drive.

 

Yes it does, when the drive slows down it won't magically show more speed than it can actually deliver.

 

Sure, the flaw may still be there, but if it doesn't result in a decrease in performance yet, it's irrelevant whether there is or not.

[shujin]

had a conversation with a friend of my dad 2 month ago and he basically said that write/read is limited therefore ssd can't really be used on actual desktop I answered that yes they are limited but the amount of write/read is huge and cannot be reach unless you make it by purpose. Now I have the number and HOLY SHIT 2.4PB IS HUGE 

[Keystone Nyan Cat]

How many of each model were tested?

[Sakkura]

Yes it does, when the drive slows down it won't magically show more speed than it can actually deliver.

 

Sure, the flaw may still be there, but if it doesn't result in a decrease in performance yet, it's irrelevant whether there is or not.

 

Most storage benchmarks, including Samsung Magician AFAIK, measure performance by writing and reading its own set of data to the drive. Which is generally the right way to do things, you can control whether the data is compressible or not, and so on.

 

But the 840 Evo issue only affects data that has been sitting on the SSD for at least a couple of months. Read speeds slow down for that data, but will recover if the data is rewritten elsewhere or if it's deleted and replaced by new data.

 

That's why it took a while for the issue to be discovered - first it took a few months for slowdowns to happen, then it took another several months for anyone to run non-standard benchmarks and notice something weird was going on that most benchmarks weren't seeing.

[That Norwegian Guy]

But the 840 Evo issue only affects data that has been sitting on the SSD for at least a couple of months.

 

I reinstall (after format) Windows every 3 months.