My plextor's health seems odd

As it seems. Some of you may know that I got a motherboard replacement last year, and got my machine back in September. It had a new Plextor M6M SSD instead of my old M5M. I was amazed, and happy for it, but twice in the last week I have run into an issue where my "disk usage" according to Task Manager hits 100% on my OS drive alone (my Plextor) and I can no longer open or close anything, and right clicking on something causes it to freeze (including explorer.exe). I have had to hold down my power button each time after waiting over 20 minutes for a fix.

Now, I know not if that is relevant to this issue, but it caused me to download CrystalDiskInfo to check my SSD's health just in case. What I've encountered is this:

Spoiler: Plextor M6M

Spoiler: Samsung 840 Pro (PM841)

Now, my Samsung has been on FARRR longer as you can see. This plextor has been in my possession for just about 1 year. I think that's an unnaturally high wear, but it's so difficult to find any good information via quick google searches for checking SSD health. Most of the articles and data are from 2012 or early 2013, and I don't know how relevant they are now with changing SSD tech.

Another thing that bugs me is how many writes. That's a pretty big write count, and I'm not exactly sure why. I don't write much to it. Is that just Windows using the paging file? Or any idea what else that might be?

Also, is this something I should consider a RMA for, or would I wait until it gets worse, etc? My M5M before I sent my machine in had 100% health after its ~10 months of usage, so it's weird to see this one be low.

 

I have the Plextor M5 Pro Extreme 512GB and older version, I have almost 30TB of writes and my SSD works great and has 100% health, I do an secure erase before i do a clean install, some people look down on this, but it`s only 1 write cycle over the entire ssd, and resets the wear leveling.

A few thing to check below.

Check that Trim is enabled, Also garbage collection (though i don't know how you check this is on) if either of these is disabled you sdd performance will start to suffer.

Do you have the latest into RST driver.

http://www.intel.com/p/en_US/support/highlights/sftwr-prod/imsm

You could do an secure erase using the plextool and a clean install, or do a image of your ssd do the secure erase and then write the image back to the ssd, the clean install is the preferred was though in as you can an software error causing your problem, how is the plextool app showing you sdd health.

Also i hate to say this, but make sure you have the fastest firmware.

You could install crystaldisk info and check the smart info for problems.

http://crystalmark.info/software/CrystalDiskInfo/index-e.html

John.

 

The behaviour of your Plextor M6M reminds me of the Sandisk U100 which could cause my notebook to freeze for several seconds if hit with a pile of concurrent reads and writes. It had very poor write performance (which you can see from the CrystalDiskMark result here).

I would also note that your M6M has had much greater workload in its relatively short life. 9527GB writes / 128GB capacity = 74 while the Samsung 840 has accrued 6368/512 = 12. The M6M could well be struggling to keep up with its garbage management, particularly if the drive doesn't has much free space.

Tryimplementing Tiller's rule of 30% over-provisioning and have at least 30% of the M6M's capacity unpartitioned. Moving the virtual memory swap file onto the 840 may ease the burden on the M6M and check what else you can move over.

John

 

That's what I was thinking. It seemed like too much wear.

Both are enabled, and I manually make sure Trim has been run now and then too.

I don't know if I do; I'll check this.

This would be a last resort for me, my external storage is not reliable at all. I really would rather not rely on it. I've never secure erased it before.

I'll check this, though if I remember correctly, they both had the latest firmware last time I had looked.

My screenshots are from CrystalDiskInfo...

I've NEVER experienced the issue except for twice this week gone. It was that issue that caused me to check on my SSD's health to see if there were any bad sectors or anything like that, is all. It's just that it happened twice in a short period of time which made it seem odd; I've known Windows to do odd things before as a once-off. I even had one time on my D900F that I kept getting Driver IRQ not less or equal BSODs for about a week. Then it just went away. I didn't touch anything.

Yeah, its workload surprised me. I originally had the paging file on my HDD but whenever I'd use the drive up it would slow my system to a crawl (which is why I suspect my system froze up when my drive's usage maxed; as the paging file was basically impossible to access). But the 840 is a good option I was considering.

I can't leave that much unpartitioned. I have about 40GB of free space at any given time though. I leave programs on it but I don't put anything else I can help.

 

I suspect the basic problem is that 128GB isn't enough for Windows, a few programs, some user files and have enough spare space for garbage management. It's not helped by 128GB (and usually 256GB SSDs) having slower write performance due to the cache size / configuration.

However, you might be able to do some housekeeping and reclaim some space to keep you going until you get something with more capacity. There's a good guide here.

John

 

I have never used overprovisioning on my M5ProExtreme 512GB, the plextool ssd app does not even have an overprovisioning option, does the M6M have overprovisioning?

EDIT: Is your SSD an M6M 128GB as the latest firmware v1.08, your crystaldisk is showing v1.01

John.

 

I usually have 35-40GB of free space on my Plextor, so it should have more than enough for garbage keeping/TRIM.

I do not know what overprovisioning is, so I probably don't have it enabled.

I'll be updating its firmware soon then (when I wake up; I'm going to sleep in a bit). Firmware updates are done in windows via their applications, right? I don't need to boot into a utility or anything?

 

The plextool utility has an update firmware section, just get it to do a check, I only updated mine onece a few years ago so i am not certain what the procedure is, the firmware check might take you to the plextor website so you can download it, the webpage should have instructions on how to update the firmware.

I never used over overprovisioning on my Plextor SSD, I did not see it even mentioned on the Plextor website when i read the specs, and the plextool utility does not have an section to enable or disable it, so maybe Plextor SSD does not need it, as it`s a waste of valuable space in my opinion and with an 128gb SSD free space is a premium.

An guide to ssd`s is in the link below.

http://howtodoit.eu/ssd-provisioning-frequently-asked-questions-answers/

John.

 

Okay, I've updated the firmware (it was done after the lockup below). Plextool was worthless for it; it kept saying it couldn't extract the firmware. I had to manually download and install it from Plextor's site.

I got the PC lockup again; checked disk usage. No programs appeared to be using my drive in any capacity, however it was still stuck at 100%. I managed to (without right clicking) close a bunch of programs and after about 15 minutes, the system seemed to catch itself. I am even further thinking it's a problem with my SSD itself. If it happens again after my firmware update I'll consider a RMA.

Also, after updating the firmware, it seems my total NAND writes has doubled to 12,579 from the 6275 from earlier this morning when I took the picture, according to CrystalDiskInfo.

 

If it's still acting up after the firmware I would return it and get a Samsung sad 256gb or bigger if you can afford it,, 128gb is far to small really.

John

 

Well I don't have any money really. I bought the M5M with this laptop in 2013, and I had a motherboard replacement last year and they also replaced my M5M for a M6M. I have a 512GB for games and 1TB for storage and I've been able to manage the storage on the M5M/M6M with relative ease. I think it might really just be something with the drive, but thankfully I'm under warranty. I'll have to borrow external storage from someone or something to clone my OS though.

 

How much free space is on the M6M according to Windows? Did you try a disk cleanup or other housekeeping?

John

 

As I said originally, I always keep 35GB to 40GB free space on the M6M. Since I've been using it, it has been so.

This morning I noticed that about an hour after the lockup I described in my 2nd to last post, windows threw up some event logs under system claiming that a drive has a bad sector. Is it even possible for a SSD to have a bad sector?

 

Sounds like drive is failing. Best to get it replaced.

A simple way to OP is go into Windows disk management, shrink the partition, create another partition in the free space, then use Windows 8.1 or 10 defrag and optimization tool to initiate TRIM on both partitions. Or if you have Windows 7 download the trial version of O&O defrag (free version doesn't offer TRIM) and run the optimize button. You can then uninstall the O&O tool without any ill effect. Then let system idle for 15-30 minutes and then delete the small partition you created. This should help kick in the garbage collection routine. But if your drive is failing, this will only buy you some time.

 

What does Win8.1 built in ssd optimize do, is it just running Trim?

John.

 

I believe it's reTrim. The drive is supposed to support S.M.A.R.T. so apparently it's supposed to do garbage collection on its own. I've never had a problem before the last week or two, and checking my error logs for "disk" shows a set of errors from the last three weeks alone with controller errors and as of yesterday, bad sectors.

Yeah, that's what I was afraid of. Welp, I'll call up mythlogic tomorrow and explain the issue and hopefully they'll just send a replacement SSD without much hassle. I can't afford sending the system up again.

I'll try that too if it happens again (hasn't since firmware update... yet).

 

Bad sector errors on an SSD is unusual and not a good sign. Rerun CrystalDiskInfo and see if anything has changed in the past two days.

It looks like it needs to be replaced. You could try chatting up the supplier about having a 256GB version instead (which probably now costs whatever the 128GB version cost when you bought the computer). If necessary, offer to contribute towards the cost difference. A higher capacity SSD will also be better able to handle the workload.

John

 

Eh I'll see what I can get, but at this point I'd rather just have anything that works properly. All I know is that I'm glad I got a 3 year warranty on this thing. As of right now I'm into my third year; I'd have been SOL if I had taken a 1 or 2 year warranty.

 

Does Plextor ssd not come with more than a 12 month warranty anyway, Mine came with an 5 Year one (well it would have if i did not buy it second hand), though in the UK just can claim for up to 6 years (the sale of good act 1979) and it`s the retailer that you return the item too not the manufacture, they can choose to do an repair, replacement or refund, and if the repair does not fix the problem, or it takes over 30 days, you are entitled to a refund, though not many retailers know this, but trading standard are starting to question the sales staff on how they handle faulty products, and more and more retailers are training their staff, though i have yet to test the 6 year guaranty, In the UK from this month if you buy software like a game and it is buggy, and needs an update you can demand a full refund.

John.

 

Well unfortunately for me, I live in the caribbean, where laws like that don't apply. Our laws are so lax that you can be sold broken products and have people refuse to accept returns accusing you of breaking it yourself or all kinds of crap.

If I have a longer warranty on this SSD, that'd be nice. But I don't know anything about that, nor do I have documentation for it. As far as I know, everything goes through Mythlogic as far as RMAs go.

 

A SanDisk Extreme PRO 240GB costs under $ 120 now. A very good choice. Samsung 850 Pro is actually over $ 130. Old Crucial mx100 240gb is under $ 100. Has always recommended to use minimum 240gb ssd as Os disk. But many people don't want to hear this .

 

Oh no, I agree and I usually say that ~250GB SSDs are the minimum people should use, after going with my 128GB/512GB/1TB triple drive setup. It's simply easier to manage.

BUT I don't have any cash.

 

Not enjoyable. Hope this works out for you. I showed the options that are worth looking at. Here in Norway the price is over twice on those ssd's, so an +240gb ssd for under 100 or 120$ is cheap. Nothing is cheap enough when you lack money.

 

Exactly XD. That's why I'm hoping a warranty exchange goes smoothly. I'll still need to find reliable external storage to clone my disk to.

 

@tilleroftheearth @D2 Ultima @Matrix Leader @Ashtrix @Mr. Fox @TBoneSan @TomJGX
You're better off with a low-end MLC-based drive . Someone who still has a taste for cheap (small) SSD's and TlC nand? http://www.pcworld.com/article/2998...ng-problem-storage-makers-dont-advertise.html

 

If possible, I ALWAYS suggest MLC drives. Not only for longevity, but because their performance was plain and simple better. But this is some interesting and new-to-me news. Did you know about this @n=1 @octiceps

 

Have always stated that TLC is garbage.

See:
http://www.tomshardware.com/reviews/ssd-recommendation-benchmark,3269.html


And forget low end MLC too. The SanDisk Extreme Pro's are the only SSD to buy today, no matter how you use your system, even when price is considered.

People get excited when I state that a specific SSD will get worse than HDD performance... But that is exactly what happens with many SSD's (yesterday's models and todays). With some even stating that when SanDisk/Intel is recommended it is the same stale recommendations.

But the recommendations are not stale; the industry is.

And as you can see from any of the review sites, SSD testing and reviewing that has shown us a successor (for SATA3 drives at least) has yet to happen. And for many sites - they have simply given up reviewing new SSD's for a couple of years or more now...

TLC junk is fine for phones.

For a computer, MLC is still HIGHLY recommended.

And for the low prices SanDisk Extreme Pro's are going for... it doesn't matter how light a workload you have.

But, pick those SanDisk Extreme Pro's up soon...

As their continued availability is in question, right now.

 

SanDisk Extreme Pro is the top one for you, but which SSDs come after this one? Most people need to know what other option that is usable. Otherwise, many will continue to buy TLC garbage as they believe newest ssd's with this TLC **** has improved.

Don’t buy it at current prices
Only buy this drive or another TLC drive + (Samsung 850 EVO excepted) if there are serious savings involved. At this point there aren’t. So until prices drop, the Q300 and the Trion that the Toshiba put out under the formerly unsullied OCZ banner are two drives that should be grounded in your “no-buy zone.”

Avoid them entirely if you perform even moderately disk-intensive work.
http://www.pcworld.com/article/2993...-ssd-review-not-the-bargain-it-should-be.html
http://www.pcworld.com/article/3000...s-but-not-for-slinging-extra-large-files.html

 

Yeah, I haven't used anything but MLC drives

 

If you want a reliable, solid performance (not extreme, but solid) SSD, look no further than the Micron M600. They are DIRT CHEAP on ebay. I bought a brand new 256GB for 80 and a 512GB for 140. Proven rock solid MLC NAND from Micron and a well established controller. Really can't go wrong here.

 

I actually have seen many bad things about Crucial's M500 and M600 SSDs. I would not look for those as any sort of top-level choice.

 

Link me to stuff on the M600. I know the M500 had some speed issues on the smaller capacity drives, but the 15 M550s and 12 M600s that I've used have been fairly consistent in speed and 100% reliable

 

It's not speed issues. It's heating up issues and suddenly not being detected by the BIOS requiring power draining to show up, etc. I don't have "links". It's from people who have them that I talk to, and they've all had them quite long-term.

 

Interesting. I wonder what form factor? I've pretty much only used the 2.5" drives, and they typically sit between 28 degC on light tasks and 37-38 degC on load/file transfer. I wouldn't call that hot. I have not had the disappearing magic either, whether it be on several different thinkpads (T410, T420, X220, T510) or Alienwares (M17x R3, M18x R2, Alienware 18, M17x R4, M15x, 17 R1, 17 R2). Maybe the silicon gods favor me. Who knows? Not me.

 

2.5" drives only, though Ethrem I believe had a M600 mSATA.

The first disappearing act was months after owning. No idea what caused it. But yes, everything is silicon god lottery.

 

Do those even have a warranty?

 

Some offers are cheap enough to go with even without a warranty.

 

I'm not a vet in this but from my knowledge..

After knowing about the controllers, tech on Sandisk along with their ncache and the NBR's trusted SSD brand, maybe the Sandisk Extreme Pros come first in terms of performance & reliability (though they use larger node vs the other MLC alternatives except for the Samsung's 3D-VNAND 40nm)

Read about the samsung screwin up with the firmware releases with their Pro series heard from @Matrix Leader, replaced his 850Pros with Sandisk Extreme Pros iirc..but there are these industry first 2Tb drives from Sammy

If I were looking for a cheaper alternative - Crucial MX200 would be my option but the 500Gb and 1Tb 2.5" drives don't have that pseudo SLC a.k.a Dynamic Cache Accleration or nCache of sandisk type. Coming to the M600 the >512Gb doesn't have this tech again..

Spoiler: Diving deep into SSD space

http://www.custompcguide.net/the-differences-between-crucial-mx100-vs-m500-and-m550-ssd/
http://www.custompcguide.net/compare-samsung-850-pro-versus-840-evo-and-840-ssd-series/
http://www.custompcguide.net/the-differences-between-sandisk-ultra-plus-ultra-ii-and-extreme-pro/
https://techreport.com/review/27909/the-ssd-endurance-experiment-theyre-all-dead
http://www.pcper.com/category/tags/dynamic-write-acceleration

 

Is it really needed on those drives? With 128GB/256GB, it can give a boost. But with 512GB+ the SATA3 port is almost maxed out anyway.

How fast a drive feels with typical PC workload is also mostly controlled by read latency/throughput, which the SLC cache doesn't really help in a steady state.

 

Not that I am aware of, but then again, I don't remember the last time I had to RMA any of my Crucial/Micron drives. My sandforce SSD on the other hand....

 

Sorry, I don't believe anyone in his right mind would buy an SSD w/o a warranty unless it's pennies on the dollar, and those are not

 

K

 

i

 

tbh, Yep, the SATAIII + SSD is saturated, Crucial says those drives don't need the dyn cache accle perk...That should be fine.

 

Double post

 

Except... they're wrong.

 

Can you elaborate? At least in sequential writing they are very close to being saturated.

 

Getting close to the theoretical maximums in one aspect of a drives performance does not mean more performance cannot be extracted from the SATA3 interface. It is simply decided to not do so.

The 'four corners' of performance that is fed down our throats makes for nice cookie cutter 'review' text. But it doesn't begin to describe the performance envelope of almost any real world workloads.

When the best commercial 'BM' (PCMARK 8) to date rates drives at less than half their theoretical values, that is not saturated to me.

See:
http://www.tweaktown.com/reviews/6470/sandisk-extreme-pro-480gb-ssd-review/index7.html

That review was completed almost a year and a half ago...


Today?

See:
http://www.tomshardware.com/reviews/ssd-recommendation-benchmark,3269.html

So sure, the manufacturer's BS marketing (and their parrot-like online channel 'reviewers') keep telling us that SATA3 is saturated for a while now.

When in actuality, most of them have simply chosen to stagnate any progress for us consumers for a very long time. Both in performance and the closely related (to productivity), capacity.

Can't wait to see a SATA3 based SanDisk Extreme Extreme Pro soon...

 

I agree on this statement. But I have a bit of problem with the testing methods from benchmarks like PCM. They tend to go back and forth inside a address space range that's too small. This is not true steady state, just stabilized response to a unrealistic workload, and it's showing the best case scenario of SLC caching because instead of benchmarking the drive, you're just benchmarking fresh data in the cache. Real preconditioning would require going through most of the NAND on the drive repetitively and keeping the drive mostly occupied, or at least to a ratio you would expect in your use case. Once you do that, the drive would have to face the tricky balance between caching more data to improve read and keeping more erased NAND to prepare for write. Unless your OP is very high most of the old data would not benefit from SLC caching at all.

We could borrow heavy benchmarks like IOZONE from server space. But they are not exactly good simulations of consumer/workstation load. And I'm not sure a regular consumer would be comfortable writing a few TBs per drive just to get a benchmark done, not to mention the long report is much harder to interpret than a PCM score.


Meanwhile in the parallel universe with enough people willing to pay, there are modern version of the SS 301 SLC drive. But in our world, the closest alternative would be buying SanDisk Extreme Pros and use them with 60%+ OP. Not sure how many people actually do this.

 

I have a problem with all BM's that do not follow my exact workloads and platform setups!

But as you say, they are still setup to give 'best' results and even then the SSD's fall on their face...

Now, 60% OP'ing is easy. All it costs is mere $/GB available for the user...

But performance doesn't increase substantially above 33% (even if consistency does) - it is mostly to protect the nand from premature death and the SSD's controller from having an unrecoverable meltdown.