New order with 7TB already put through an SSD

Ah, ofc

Interestingly my M2 drive has 11tb of writes to it as well. I've had it for over 2 years though.

Stop using the Schneier method to erase your browsing history?


Seems my 1TB 840 Evo is up to 80TB. I got that Black Friday 2014. I think the writes are from using Mod Organiser and Skyrim/FO4.

Tony1044 said:

Yes indeed - I was assuming the same sized drive to be fair, which would make the 12.xTB around 9%

Another analogy is if you bought a brand new car, with a 60,000 mile, 5 year warranty only to find it was delivered to you with 3,000 or 6,000 miles on the clock... Something tells me you'd be less sanguine about that. And yet, no one would suggest that said car would be any more likely to suddenly fail, either.

Click to expand...

I do understand the reason for the complaint. I think we really need to wait for PCS to determine whether that level of writes is necessary to properly test these drives or simply an unwanted effect of the speed of modern drives and the (current) testing methodology.

Sorry, I'm probably being thick here - but how does the speed of a drive affect the amount of data being written to it, surely that only determines how quick it does it. Do PCS not use the same test algorithm on all drives, regardless of speed/size, or am I wrong to assume that. The biggest problem is with lower capacity drives (i.e. 128GB) that typically have only 74/100TBW, then 7-11TBW does become more significant.

You're assuming the test PCS do has a fixed amount of data. Whereas it seems to involve writing data to the drive for a fixed amount of time.

So if you write data to the drive continuously for 8 hours, the faster drive will have written more data.

The 250gb 970 Evo has upto 1500MB/S write and the 1TB has upto 2500MB/S write. Users have reported 7TB written to the 250gb one and 12TB for the 1TB one. So allowing for a little variation between how drives of different size classes sustain write performance, those numbers kind of line up for me.

Oussebon said:

You're assuming the test PCS do has a fixed amount of data.

Click to expand...

Your assumption is correct - and I would think that would be the best way to do it.

Well, apparently, they don't

I agree that it's not a lot in the grand scheme of the life of the SSD but as Tony1044 said, it's the equivalent of 28 full disk writes just in testing.

I sent a message but had no reply for two days so I phoned up tech support and the guy said I must have been given a second-hand SSD. This was surprising because if he says it's second-hand it concerns me with what components are used second-hand on a brand new build. He said he would send me a replacment SSD.

I did finally get a reply from my message and they confirmed it was from the testing process that causes this.

Thank you for getting in touch and raising this to us. I can confirm that this occurs as part of our testing procedure and to ensure the endurance of the machine. You are correct that SSDs can have a limited write count though this is typically a far higher amount that the typical amount a consumer would write to the drive far beyond the warranty period.

The full cycle of testing will read and write to the drive as well as pushing other components to the limits for an extended test to ensure that we can approve and warranty the order once it is dispatched out to you.

Based on the forum thread, we have opened communications with drive manufacturers to discuss the points some forum members have made. Ultimately we will provide the warranty on the system, any remaining manufacturer warranty will carry on after our warranty per component. I do want to assure you that we do not have any example of a drive which has failed due to exceeding it's endurance rating.

Click to expand...

simonball598 said:

Thank you for getting in touch and raising this to us. I can confirm that this occurs as part of our testing procedure and to ensure the endurance of the machine. You are correct that SSDs can have a limited write count though this is typically a far higher amount that the typical amount a consumer would write to the drive far beyond the warranty period.

The full cycle of testing will read and write to the drive as well as pushing other components to the limits for an extended test to ensure that we can approve and warranty the order once it is dispatched out to you.

Based on the forum thread, we have opened communications with drive manufacturers to discuss the points some forum members have made. Ultimately we will provide the warranty on the system, any remaining manufacturer warranty will carry on after our warranty per component. I do want to assure you that we do not have any example of a drive which has failed due to exceeding it's endurance rating..

Click to expand...

That they know of.

But it's still a reasonable response and it will be interesting to see the reply from the manufacturers.

A previous PCS comment alluded to the drive having had so much written to it because of the speed.

PCS are not writing a fixed amount of data to the drive, or filling the drive a fixed number of times. It just seems to be a case of writing as much data as possible to the drive over a fixed duration of X. This would be why a 250gb 970 Evo has ~7TB writes and a 1TB 970 Evo had ~12TB writes. And why PCS said the amount of writes was down to the speed of the drive.

Is it therefore PCS's position that all drives need to be tested at full pelt for a fixed length of time, regardless of size, model, capacity, of the SSD before declaring it fit for duty?

I wonder if the manufacturers would have any guidance on that for system integrators?

Not counting HDDs / secondary storage as users have fairly often reported those not showing up in Windows because they weren't formatted, so they don't look like they get tested.

I wonder what testing the SSDs receive from the vendor before they're delivered to PCS? Assuming that they have been properly tested by the vendor I'd have thought it sufficient just to test that the drive is working when fitted to a PC/laptop?

ubuysa said:

I wonder what testing the SSDs receive from the vendor before they're delivered to PCS? Assuming that they have been properly tested by the vendor I'd have thought it sufficient just to test that the drive is working when fitted to a PC/laptop?

Click to expand...

I was thinking that myself.

It seems to be that it's an all-or-none situation as well from PCS' perspective.

If you're going to test, it needs to be sufficient amounts of data to ensure that even with wear levelling and trim, it writes to the entire drive. A small amount of data won't cut it. And that leads us to this situation.

Otherwise, I'd have thought, that as you say, installing it and installing even the test OS should be enough to assume it's a working SSD.

it needs to be sufficient amounts of data to ensure that even with wear levelling and trim, it writes to the entire drive.

Click to expand...

But if that's the case, should the qty of data not be proportional to the size of the drive? Which it currently is not.

Oussebon said:

But if that's the case, should the qty of data not be proportional to the size of the drive? Which it currently is not.

Click to expand...

No that's kind of my point but I didn't put it across well - you cannot do a simple "let's write a 20GB file" as it won't be enough. You'd need to tailor it to each size SSD, so a 250GB would need a 250GB file, for a 500GB a 500GB file etc. That's the only way you could guarantee you are a) writing to the entire drive and b) with enough data to ensure it can't use wear levelling but fills the drive entirely.

I'd be interested to know how the test is done anyway as the more I think about it, the less likely it seems to be entirely valid - just pumping through as much as possible as quickly as possible will still utilise trim and wear levelling so you've presumably no guarantee you're testing the entire drive equally.

The more I think about this the more I think it shouldn't be done - as ubuysa says, I can't imagine they leave the manufacturer untested. I'd think dropping Windows on (even the test version) is enough to verify the drive is ok.

I don't know for certain but I imagine these tests would be more for conventional hard drives where you can have sector issues etc? Rather than SSDs. I don't know if you actually need to fill an SSD to test it fully?

I would like to think that one or 2 capacity writes would be more than enough for any drive.

I wonder if PCS check drive thermals. Probably not? Or even if they do I don't think you'd need to fill a drive 28 times to work out whether there was a problem with the heatsink if it even has one.

Also, we've seen a few systems on the forums where PCS put the SSD in the 'wrong' M.2 slot (I think on an Intel board where it affected Sata port availability, and in an AMD mobo where it nerfed the speed). So whatever the testing is, it doesn't register / get affected by that.

The_Scotster said:

I don't know for certain but I imagine these tests would be more for conventional hard drives where you can have sector issues etc? Rather than SSDs. I don't know if you actually need to fill an SSD to test it fully?

I would like to think that one or 2 capacity writes would be more than enough for any drive.

Click to expand...

Hard drives ship with spare sectors, it's these that are brought into use when chkdsk /r 'repairs' a faulty sector. I don't know whether SSDs ship with spare cells/blocks but I doubt it.

With my limited understanding of how SSDs work internally I would have thought that if you're going to properly test one you need to write enough data to fill every cell/block AND to give the wear levelling mechanism time to operate and thus prove that it's working properly. That might mean a lot of data writes. As mentioned though, I'd expect the vendor to do that before shipping?

Given that there are manufacturer warranties on these drives I'd have thought a handful of write/reads at PCS would be enough to show that the drive is installed and working? If it later fails with the customer and PCS have to replace it I'd assume that PCS could use the manufacturer's warranty to recover the cost (of the replacement drive)?

Oussebon's point about the testing not revealing that some drives were fitted to the 'wrong' slot is a good one, so perhaps PCS's testing should be aimed at ensuring the PC/laptop delivers the performance expected from it rather than just burning in the components?

Had a look at my ~3 month old 250GB 970 Evo after seeing this thread and it's had 8.6TB written to it. Not really sure how significant this is but the car analogy from earlier in the thread comes to mind, with this being about 5% of the TBW warranty.

It would seem that PCSpecialist are looking into their testing methodology. I paid for a fast track build and when the status ‘In Building’ remained unchanged after 4 working days I emailed to query any problems. I had a reply this morning (to be fair, very promptly) to say that they were experiencing delays due to issues in their testing procedures. Sadly, they have not given any indication as to how long the delay will be.