Is this SSD degradation ?

When my laptop was new :




After reinstalling Vista




Laptop is Samsung X360 - SU9300

I believe there is no firmware on this Vista, possible available for W7

 

As I understand it, you should have done an 'secure-erase' before re-installing Vista to get the drives performance back up to 'new'.

At least its not slowing down to 8MB though? Do you see/feel any difference just using it?

And, yes, definitely degradation.

 

actually, to be fair, i much prefer the SECOND image. the first one has some huuge issues..

 

A "secure-erase" is that something in Vista ?

Can I still do it and reinstall Vista to get new performance ?

 

No a 'secure-erase' is what wipes the SSD controllers table clean so it doesn't have to track (and slow down, because of) used pages on the flash memory.

Yes, search for the tool (you have to do it on another system I think) and then re-install Vista.

davepermen, oh ya! First image looks scary.

 

WOW....this one is actually getting better over time! Well, at least by just looking at the graph....that's how it should look with an SSD, right? Instead of a downwards curve like with a "normal" HDD. The max. nrs in the first one are much better but it doesn't seem stable at all.

 

"SSD Degredation"??

What is that, exactly? I mean other than a pretty imprecise and dramatic terms for something.

 

Not imprecise nor overly dramatic (unless you're experiencing it).

http://www.anandtech.com/storage/showdoc.aspx?i=3531&p=8

 

the fact that ssds can get slower over time as they don't ever get deletes reported, and thus have to handle all the chaos that gets written to them.

but samsungs shouldn't, as they get delivered in "degraded state" out of the box, to make sure they don't get slower.

 

and, tiller, it is sort of dramatic, as the word can mean the ssd falls apart bit by bit. but it doens't. only it's performance gets a bit worse.

 

Effectively the same thing to me.

 

You have a Samsung PB-22J, should get 220 sequential read and 200 mb/s sequential write. What firmware does your drive have? Go device manager>disk drives> your ssd disk>properties>hardware ids in properties drop down. Should be VBM1 something.

 

no. one would mean the drive would really die, fall apart, getting non-functional step by step.

and, together with the fact that they do have a limited amount of write cycles, people can combine that word + that fact, into "they will die over time, not last long". and together with some randomly dying ssds, that results in panik

 

davepermen,

this is our second time discussing this issue - and yes! I would be panicking if that happened to my production workstation!

Totally a deal-breaker for me.

 

sure. but problem is, it is not like that.

the degradation thing is just, disk gets full, disk gets slower. that's all.

it is nothing about write cycles, about random dying ssds, about anything of that sort.

problem is, without trim, the disk never gets faster when you delete files, then, afterwards.

but a 'degraded' drive works perfectly well. it's just slower.


the problem is, people mix up and match for drama all the issues that exist, combine them into one big thing and use the worst word to name it.

 

Laptopaddict, the first graph looks very unstable, whereas the second seems a lot more stable and faster access time.

 

It is still one of the reasons i haven't got one yet (the other is $$) but to be honest, other drives can suddenly die as well....for various reasons, some of which SSD's are completely free from.

 

was your drive at the moment of test 1 around 60% full? because then, that would explain the graph: the empty parts in the second half are empty, so the drive just optimizes that, reporting very fast "nothing here, read further"

and after the second installation, that part got overwritten once, too.. from then on, it performs the same everywhere (which is why the second graph actually looks logical, but quite slow in general).

i have to check the specs of your drive if that's actually it's normal speed..

 

"degradation" as commonly referred is where the drives "free space is written to", which means the operating system shows free space but the controller does not "know" it is free so it has to erase data in real time as it is writing it, causing slowdowns. The newer firmwares have GC and TRIM which are similar in that they consolidate and erase deleted data so that the OS can always have free space to write to that doesn't need erasing prior.

All drives, contrary to what davepermen would lead you to believe still degrade in real time one way or another. The Intel drives just do it in a way that is supposedly the least detrimental to real world uses. Intel has made the decision to design their controller so that all of it's focus would be toward 4k random writes at the expense of sequential writes. If you throw 3 minutes of sequential writes at an Intel drive you will see gradual degradation. The drive will recover very quickly once you stop sequentially writing but not until you do.

The achilles heel of the Samsung drives is 4k random writes, if you throw 3-5 minutes worth of steady 4k data at it it will start off at about 14 mb/s for a few seconds then gradually decrease to about 5 in the first minute then level off at about 3. The rest of the Samsung benchmarks, seq read and writes and random 512 read writes actually gradually increase over the 3-5 minute IOMeter tests.

The reason most analysts have put so much emphasis on 4k random write performance, I think, is because the previous generation firmwares did not recover from degradation because there was no background GC or TRIM. 4k random write performance was the only way for a drive to be worth anything once a controller thought the drive was full. Since blocks have to be erased in 512k blocks all writes would become random 4k at that point resulting in permanent degradation until it was secure erased.

 

i never said something contrary, sgilmore. but i state that in nowadays drives, the good ones at least, it is not a problem.

and that's a difference.

and the other difference is, people without knowledge try to mix that up with the limited writes (which is the reason the degradation occurs, yes), and might think it's about the drive starting to die. others even mix it up with random failures and all such.

the correct term, thus, would be "performance degradation". to make sure that it has NOTHING to do with the drive's life at all.

 

+1

Performance degradation it is then: That's the show stopper for me.

 

depends on the drive :9 the intel g1 is even in degraded mode still snappy and fast, so it's unimportant.

but curious about that samsung here..

 

I agree 'theoretically', but for me;

depends on degraded performance offered compared to my 4 Raptor desktop.

 

It depends, I think what firmware is on the drive. The Samsung controllers that consolidate and erase deleted blocks only work when the drive is "idle" wheras Intel drives past a certain firmware do it in almost real time. It is often, mostly with notebooks, that the Samsung controller never see's the drive as idle. Most users with notebooks and background GC report restored performance after either logging off or S1 sleep mode.

 

Firmware is VMA0501Q

Is their an update available ?

 

Yes, the drive was more full for test 1 than test 2...(don't know if it was 60 %)

 

well, if you had first 60% filled, and the second time again at least 40% filled, then it is normal behaviour.

BUT the actual speed is much too low on BOTH.

 

I don't think that HD that laptopaddict has is a PBJJ-2. I've got a 1.8 in my TT and my speeds are different.
The second benchmarks seem quite in line with several that were posted in the following threads.

http://forum.notebookreview.com/showthread.php?t=413354

http://forum.notebookreview.com/showthread.php?t=430277

 

Does this BIOS update has anything to do with the firmware for the SSD ?

http://www.samsung.com/uk/support/d..._cd=05010300&prd_mdl_cd=&prd_mdl_name=NP-X360

you have to click on firmware..

 

yeah, it looks like my 1.8" 128gb samsung, which is not the 200MB/s one, too.. the 64gb one i have is (but feelable performance is the same anyways, so actually, it might not matter much to the OP)

 

I Googled MMCQE28GFMUP and got THIS and clicked the data sheet link and it showed PB22-J. I was just going off of Samsungs info, doesn't mean it is accurate, their website is lousy with info.

 

yeah. done the same. it looks like it should get 220MB/s, but it behaves like mine, which doesn't. now i guess we need crystaldiskinfo to show the real information? does that show if it's pb22-j?

 

Yes, Samsung is not always the most accurate with their speeds/labelling which is a little odd.
This is the 1.8 in my TT which has a PBJJ.
It seems to be only a recent thing for Samsung 1.8 drives that they have the 220/200 read/writes. The x360 has been out for quite some time now.

http://forum.notebookreview.com/attachment.php?attachmentid=40893&d=1258548383

http://forum.notebookreview.com/attachment.php?attachmentid=40894&d=1258548845

The ones that are MMCRxxxxxMXP look to be the newer drives. Mine is a MMCRE28GTMXP-MVBD1.

 

Old post I made a while back:
http://forum.notebookreview.com/showpost.php?p=5261134&postcount=20

Basically, if the model number ends in MUP then it's the gen 1 MLC drive, and if it ends in MXP it's the gen 2 MLC drive.

 

so the numbers are fine. and the high part in the first post was "empty space" that never got written on before. after the disk got written over once completely (doesn't mean to have been filled, just gotten 128gb written on to it, unimportant what got deleted in between), the numbers got more sane, and, if MUP is first gen, actually correct.

 

You can try this FW update utility from Lenovo. It shows where your SSD is supported but not your particular FW. It may not update your SSD if it does not recognise your FW ids as being a Lenovo FW ids.