SSD Write Endurance 25nm Vs 34nm

[johnw]

it looks like OS X could be next (as I've got a Mac) and on top of that I need to get up to speed with Linux)

It should be possible to write code that would only need minimal changes to run under OS-X or linux, if you choose a GUI library that is available for both OSs.

[bluestang]

We have a new contender...

Samsung 830 64GB will be up for testing sometime early next week (I think - I mean, as soon as it gets here it's getting trashed).

The 64GBs do come with Norton Ghost and Batman: Arkham City (which I won't play because [A] it isn't Skyrim and [B] uses GFWL).

It's a good deal, and sequential writes are rated at 160MBs.

I expect the 64GB version to average around 110MB/s under endurance testing, but I can't wait to find out if I'm wrong. With practically no 64GB model reviews, I'm just guessing.

Here's hoping it will last a little longer than the Mushkin.

Nice! Wonder what the WA will be on the 830 compared to the 470

[Christopher]

I was waiting for 520 as well (but who isn't?). November came and went with nary a whiff of a new Intel drive, and Octane launched but only in > 128GB varieties. Either one of those drives in a smaller capacity would have been great for testing. It's really disappointing that Intel couldn't manage a late December 520 launch (I'm sure someone will get me a SSD for Christmas, and sadly it won't be Cherryville). For less than $100 you get the Samsung 830 and Norton Ghost (which I haven't used since 2004), plus a new AAA game title (if you're in to that sort of thing). It's a pretty good deal overall for that 0.00006 percent of the population who are considering buying Ghost, a Batman game, and a SSD.


I for one think ASU on Linux wouldn't be a horrible idea for endurance testing. After all, you could run the installation from a USB drive or Live CD and mitigate the need for a system drive.

[Anvil]

We have a new contender...

Samsung 830 64GB will be up for testing sometime early next week (I think - I mean, as soon as it gets here it's getting trashed).
...

Great!
We should agree on some standard way of reporting from now on. (like Ao1 suggested some time ago)
(before you both start testing the new drives)

Been following from Day 1. Looking to buy SSD soon, but data so far doesn't translate to buy this, stay away from that.
Think it's about time to finalize charts and reach some type of consensus?
There is so much data and with new SSD lines coming out. I'm afraid all this hard work will get lost in the deluge of information already obtained.

Nothing is lost, looking at the big picture we can already tell that all drives will endure way beyond MWI.
There will be a summary in post#1 within reasonable time but the test is ongoing and I'm expecting that some of the drives will last way into 2012.

I'm not sure that you should expect that there is a winner here, they are all winners imho, they just work differently and the way they work leads to how long they will last. (e.g. in particular the Samsung 470)

Next is to find what data retention does for the drives when MWI is at "0" and we have already started doing some tests on this and all is a result of what has happened so far in the thread.

If you are looking for an SSD in the near future you can safely select any of the modern drives used in this test.
So in general, the drive to go for depends a bit on your usage, if you do a lot of incompressible data then you should go for a drive without compression.

It should be possible to write code that would only need minimal changes to run under OS-X or linux, if you choose a GUI library that is available for both OSs.

Yepp, both a GUI and a general purpose library is needed.
(haven't started looking for anything yet)

[Christopher]

The drive won't be here for several days, so there will be time to re-establish new testing parameters. What should be done differently from day one? I was taking screen shots of smart data every day with the Mushkin, aside from retention testing later in the drive's life, I'm not sure what else to do.

[Anvil]

I suggest we plot the data into an Excel worksheet, nothing fancy just keeping a log on the important values/attributes.

I'll make a template we can discuss over the next few days.

[johnw]

I suggest we plot the data into an Excel worksheet, nothing fancy just keeping a log on the important values/attributes.

I'll make a template we can discuss over the next few days.

You could make a spreadsheet in google docs and everyone could enter their data into it themselves...I think...I've never tried to make a google docs spreadsheet with lots of people entering data.

[Anvil]

Google could work!
I've tried it but have not checked for concurrent users/if it detects that someone is already having a write lock on the document.

[Christopher]

Microsoft's SkyDrive is also an option. I think it may be more geared toward what we are trying to do. Plus, you get real Excel in your browser and integration with the desktop version.

It supports multiple users editing the same document. I'm trying to post a picture for others to see who are not familiar with it.

I'm sure most have some familiarity with it, and Google Docs is good too, though I've not used it in quite some time.

[Eternalightwith]

[QUOTE=Anvil;5006827]Great!
We should agree on some standard way of reporting from now on. (like Ao1 suggested some time ago)
(before you both start testing the new drives)

Next is to find what data retention does for the drives when MWI is at "0" and we have already started doing some tests on this and all is a result of what has happened so far in the thread.

If you are looking for an SSD in the near future you can safely select any of the modern drives used in this test.
So in general, the drive to go for depends a bit on your usage, if you do a lot of incompressible data then you should go for a drive without compression."

As a scientist I love this stuff, but as a consumer I'm still left with questions.
Windows Skydrive excel would be great.

I guess for me, I want to use a drive with Intel's SRT. I want to get two drives one for OS install and one for caching my 5900 drives. I don't want it so small as Intel's 30gb drive, but OCZ Synapse Cache is not up my alley either.

So I guess if I took a stab at the graph, with shear number of writes (1Petabyte)and data retention, that Crucial M225 looks like the ticket. I like the M4 the best but it just up and dying like that has me cringing, "N" of 1 yeah I know, probably an isolated case.

[johnw]

So I guess if I took a stab at the graph, with shear number of writes (1Petabyte)and data retention, that Crucial M225 looks like the ticket. I like the M4 the best but it just up and dying like that has me cringing, "N" of 1 yeah I know, probably an isolated case.

Are you planning on writing more than 500TB to your SSD? Because if not, the m4's death is hardly relevant to your case.

I would guess that the m4's failure is NOT an isolated case for m4s that have had 500+ TB written to them. If you write past the specified MWI for the SSD, way past it in this case, then you really need to expect that the unpowered data retention time is liable to be very short.

But it IS a rare occurrence for normal usage, or even for extreme usage that people would put the SSD through when they are not intentionally trying to write enough to the SSD to make it fail. I doubt one person out of 1000 who uses that SSD for real work (or play) is going to write more than 100TB over the lifetime of the SSD.

[Christopher]

Is there any utility which could log SMART values at certain intervals and then be exported to a CSV?

[bluestang]

M225->Vertex Turbo 64GB Update:

864.40 TiB (950.42 TB) total
~2215 hrs (Torture), 3183 hrs (Power-On)
16816 Raw Wear
117.27 MB/s avg for the last 20.28 hours (on W7 x64)
MD5 every 20 loops on 1.59GB file = OK
C4-Erase Failure Block Count (Realloc Sectors) from 21 to 24.
C5-Read Failure Block Count (uncorrectable bit errors) from 6 to 7.
1=Bnk 6/Blk 2406 - Erase Failure C4
2=Bnk 3/Blk 3925 - Erase Failure C4
3=Bnk 0/Blk 1766 - Erase Failure C4
4=Bnk 0/Blk 829 - Erase Failure C4
5=Bnk 4/Blk 3191 - Erase Failure C4
6=Bnk 7/Blk 937 - Erase Failure C4
7=Bnk 7/Blk 1980 - Erase Failure C4
8=Bnk 7/Blk 442 - Erase Failure C4
9=Bnk 7/Blk 700 - Erase Failure C4
10=Bnk 2/Blk 1066 - Erase Failure C4
11=Bnk 7/Blck 85 - Erase Failure C4
12=Bnk 4/Blk 3192 - Erase Failure C4
13=Bnk 7/Blk 280 - Erase Failure C4
14=Bnk 3/Blk 2375 - Erase Failure C4
15=Bnk 7/Blk 768 - Erase Failure C4
16=Bnk 7/Blk 765 - Erase Failure C4
17=Bnk 7/Blk 182 - Erase Failure C4
18=Bnk 5/Blk 939 - Read Failure C5
19=Bnk 5/Blk 1115 - Read Failure C5
20=Bnk 5/Blk 1011 - Read Failure C5
21=Bnk 7/Blk 3549 - Read Failure C5
22=Bnk 7/Blk 3556 - Read Failure C5
23=Bnk 4/Blk 1961 - Erase Failure C4
24=Bnk 7/Blk 1862 - Erase Failure C4
25=Bnk 7/Blk 111 - Erase Failure C4
26=Bnk 5/Blk 902 - Read Failure C5
27=Bnk 7/Blk 560 - Erase Failure C4
28=Bnk 7/Blk 827 - Erase Failure C4
29=Bnk 6/Blk 482 - Erase Failure C4
30=Bnk 5/Blk 167 - Read Failure C5
31=Bnk 7/Blk 1771 - Erase Failure C4

So a total (B8+C3+C4+C5) of 41 Bad Blocks out of 32768.

[Ao1]

Skydrive would work great. Personally I think each person testing should input data into the excel file and one person should be responsible to create a master set of graphs from the input data. Only those testing should be able to modify the excel file. (I think it is possible to set permission with Skydrive) It would be great however if the file could be downloaded by anyone so they can play with the data and maybe spot something of interest.

A standard will be difficult to achieve when all drives use different SMART stats, but I guess a standard could be developed for each controller.

I think we are missing out by not capturing key metrics, so I hope this can soon be rectified.

[Christopher]

Skydrive would work great. Personally I think each person testing should input data into the excel file and one person should be responsible to create a master set of graphs from the input data. Only those testing should be able to modify the excel file. (I think it is possible to set permission with Skydive)

I think we are missing out by not capturing key metrics, so I hope this can soon be rectified.

With the Mushkin, I took a screen shot every day of the smart data. I could easily go back and insert the daily smart data into whatever new reporting metric is chosen. I say we should come up with with a new system within the next few days, even if its just logging smart attributes into a spreadsheet.

...but I vote for SkyDive.

[bluestang]

FYI...I've requested Vapor's files to see if we can try and continue his enormous efforts of tracking the data.

[Anvil]

Kingston SSDNow 40GB (X25-V)

521.79TB Host writes
Reallocated sectors : 05 14
Available Reserved Space : E8 99

MD5 OK

33.02MiB/s on avg (~54 hours)

* reallocated sectors increased by 1 again, it's the second movement in just a few days.

--

The Corsair is back up again after being disconnected for ~54 hours.

Corsair Force 3 120GB

01 120/50 (Raw read error rate)
05 2 (Retired Block count)
B1 54 (Wear range delta)
E6 100 (Life curve status)
E7 10 (SSD Life left)
E9 463344 (Raw writes) ->452TiB
F1 618039 (Host writes) ->604TiB

MD5 n/a

---.--MiB/s on avg (~- hours)

power on hours : 1775

I'm restarting the test in a few minutes.

[B.A.T]

Todays update:
Kingston V+100
276.4895 TiB
1206 hours
Avg speed 25.29 MiB/s
AD still 1.
168= 1 (SATA PHY Error Count)
P/E?
MD5 OK.
Reallocated sectors : 00


Intel X25-M G1 80GB
87,6873 TiB
19324 hours
Reallocated sectors : 00
MWI=89 to 79
MD5 =OK
45.24 MiB/s on avg


The Kingston is back. I'd to disconnect the power to it to get it back up and now it runs ok. The speed has been very low for some time now but I don't know why.
I also got my m4 in the mail. Of course I'm all out of SATA cables so I need to get one tomorrow before we start. You guys have been discussing changes to the test. Anyting special we should agree on before I start?

[johnw]

...but I vote for SkyDive.

Why jump out of a perfectly good airplane?

[Christopher]

Why jump out of a perfectly good airplane?

Man, I've been on a couple of commercial flights where jumping out of the airplane seemed the only course of action.

Of course, the problem was being stuck on the tarmac for hours on end, so the fall would have only maimed and not killed.

[Christopher]

The Samsung 830 will be here on Monday, barring any freak blizzards.

I'll run some benchmarks on it, then put it to work.

For now, I think I'll continue to take a screen shot of SMART attributes. Then, I'll put each day's values in an Excel spreadsheet. I figure that approach will leave me covered until we can devise a new plan.

I was thinking that for a while we could utilize SkyD(r)ive, but just dump each drives' values into it's own sheet in the workbook.

A truly fantastic addition to something like SSDlife would be a way for it to dump SMART values into a CSV text file so you could easily automate the process... but for now we'll just have to do it the old fashioned way.

For the record, I'm expecting the 830 to perform very similarly to a 64GB M4, with possibly the exception of some higher sequential writes. I'm just guessing, but I'd say that it will sustain about 8TB a day in ASU.

[johnw]

For the record, I'm expecting the 830 to perform very similarly to a 64GB M4, with possibly the exception of some higher sequential writes. I'm just guessing, but I'd say that it will sustain about 8TB a day in ASU.

Even more similar to the Samsung 470. For writes, the two SSDs should be quite close, I think. Unless they changed the write amplification and wear leveling algorithms in the firmware.

[Christopher]

Even more similar to the Samsung 470. For writes, the two SSDs should be quite close, I think. Unless they changed the write amplification and wear leveling algorithms in the firmware.

I believe that if you're using a SATA II only motherboard or laptop, the 470 is actually faster than the 830 [on SATA II].

The 470 is fantastic, and I wish I had bought one when they were still using 3xnm NAND (I believe Samsung switched out 3xnm for 2xnm in the 470s and gave them new SKUs).

From the reviews I read it appears that the 830 gives up random performance over the 470, uses more power, and comes in 7mm sizes so that you need a spacer for 9.5mm Z-height applications. If you're using SATA III then it's great compared to it's predecessor, but it doesn't appear to be a warmed over 470 with 6gbps tacked on. Perhaps they've made some concessions for enhanced WA and wear leveling at the expense of random performance for the OEM crowd?

Anyway, I'll post some benchmarks from ASU, CDM, and possibly AS-SSD when the drive arrives. It's difficult to find the most basic of benchmarks for 64GB drives these days, and most review samples are either 256GB or 512GB sizes for newer drives. Two 830 64GBs in RAID 0 might be a pretty good setup... if you're into that sort of thing

[Anvil]

The computer froze last night, about 5 hours after restarting the test.
Looking at the available data the F3 is the guilty part, there were 0 files in the test folder and so it happened while TRIMming the deleted files.

Kingston SSDNow 40GB (X25-V)

522.31TB Host writes
Reallocated sectors : 05 14
Available Reserved Space : E8 99

MD5 OK

--.--MiB/s on avg (~- hours)

--

Corsair Force 3 120GB

01 94/50 (Raw read error rate)
05 2 (Retired Block count)
B1 54 (Wear range delta)
E6 100 (Life curve status)
E7 10 (SSD Life left)
E9 465410 (Raw writes) ->455TiB
F1 619460 (Host writes) ->605TiB

MD5 n/a

---.--MiB/s on avg (~- hours)

power on hours : 1780

I'll restart again in a few minutes.

--

I'll create the template today, should be ready for discussion within a few hours.

[Anvil]

Gathering info pt1

First I'd like to get an overview of the drives used in the test, so can we agree on something like this.
(do comment if more info is needed)



The main point is to get an overview of what was tested and when the test started/ended, what comes in between will follow.