|
Any SMART reporting tool will do it, but the standard Windows/no-nonsense answer Crystal Disk Info standard edition. The Shizuku edition has anime, if that's your thing.
|
#
¿
Aug 18, 2014 05:49
|
|
|
|
| # ¿ Apr 19, 2024 16:22 |
|
|
You see those two disks labeled Caution? Get out your backups and your RMA pen.
|
|
#
¿
Aug 18, 2014 07:30
|
|
|
PCIe SSDs are expensive and/or rare. Full PCIe card SSDs are the expensive part, and good ones that aren't hyper-expensive enterprise drives are rare. Alternatively, an m.2 PCIe SSD could be plugged in with a simple adapter card, but there aren't many m.2 PCIe SSDs out yet, and even fewer worth buying. There's the Plextor M6e, but it has lovely performance consistency and will be lovely in a Mac.
|
|
#
¿
Aug 19, 2014 00:38
|
|
|
Wulfolme posted:Wasn't it a thing once upon a time to put something like 20 DIMMs in a box, pretend it was a hard drive, and just always keep it powered? I know I remember something resembling that from somewhere. I know I saw a picture of an enclosure with a dozen DIMMs in it with room for more. Generally, that's a RAMdisk on a SAN. There's also a company named, wait for it, RAMSAN that makes RAM SAN appliances.
|
|
#
¿
Aug 19, 2014 07:10
|
|
|
Diovanti posted:I didn't see any mention of this yet, so I have to ask: If I use truecrypt to encrypt my SSD do I have to over-provision it by 20% in order not to kill the performance? Do you have to use TrueCrypt? Almost all drives will self-encrypt with 128-bit or 256-bit AES if you set an ATA password. Besides that, a number of drives are compatible with TCG Opal and MS eDrive, which make the a compatible OS (like Windows 8.1) aware of the SSD's hardware encryption and delegate key management duties to the OS, rather than relying on an ATA password. The only thing TrueCrypt would add is the plausible deniability partition obfuscation thing. But yes, if you do FDE in software and need to maintain the full-disk nature of the encryption, not only does that instantly fill the drive, but TRIM won't function correctly, so you need to overprovision the drive.
|
|
#
¿
Aug 19, 2014 17:15
|
|
|
Hadlock posted:Wasn't TrueCrypt compromised and/or Officially Abandoned by the developers? Officially abandoned, suggested (maybe) that there was an NSA backdoor, and immediately forked into a new project.
|
|
#
¿
Aug 20, 2014 23:36
|
|
|
Without eDrive, there's even less performance penalty - encryption-capable SSDs always encrypt. An ATA password just enables checking for the correct password hash at boot before unlocking the drive. Fun fact: A Secure Erase on some encryption-enabled drives just erases the encryption key and then issue a drive-wide TRIM.
|
|
#
¿
Aug 21, 2014 03:32
|
|
|
You know that phrase, "even if you're one in a million, in New York there are eight just like you"? Samsung SSDs have a very low failure rate, but if every goon bought a Samsung SSD, you'd still have hundreds of dead drives. Wear level count of 1 - I would not worry - it's just a used-refurbished drive rather than a DOA-refurbished drive. Wear level of 1 means that on average, all of the NAND has been written to once. You still have, worst case, 999 to go. The "Current" value for that attribute will decrease from 100 to 0 based on Samsung's conservative estimate of when to replace the drive (i.e. at 0). POR Recovery is for when the drive is subjected to a sudden power-off, like an unsafe shutdown or pulling the computer's plug. It's more about watching the drive's environment than checking the drive itself for errors.
|
|
#
¿
Aug 24, 2014 16:13
|
|
|
AnandTech reviewed the OCZ Arc 100 today and the reviewer talked a bit about why the MX100 gets recommendations from big tech sites that I think we missed: features. The drive may be a somewhat cruddy performer, but it implements ultra-low-idle power states and TCG Opal/eDrive, making it a bottom-of-the-barrel, cheap drive that nevertheless has a full feature set suitable for laptops and FDE. Notably, the Arc 100 does not do low-power idle, and FDE is only done through an ATA password. So that's a mystery solved for why the MX100 clings on.
|
|
#
¿
Aug 26, 2014 16:06
|
|
|
According to how the Unaligned Access Count value is calculated, you've had a decent number of unaligned reads and writes. If that's true, you're getting lower performance and higher write amplification than you should be - not a huge deal, but worth fixing. Let's check drive alignment. Open a command prompt and run this annoying command: wmic partition get BlockSize, StartingOffset, Name, Index If the StartingOffset (which will be a long number) is not evenly divisible by 8192, you got alignment problems.
|
|
#
¿
Sep 1, 2014 00:48
|
|
|
Well, the Samsung drive will be faster for everyday use thanks to RAPID, but otherwise not really.
|
|
#
¿
Sep 1, 2014 04:07
|
|
|
Well, my ThinkPad's Ultrabay caddy broke and now my secondary hard drive falls out whenever I look at the laptop wrong. I decided to solve two problems and get a larger mSATA SSD, an 840 EVO, which will let me stick the hard drive back into the primary bay, stop me from feeling constrained by 120 GB of space, and also give me a spare Intel 320 to work with I guess. How does the Samsung data migration software stack up vs. e.g. Macrium Reflect? Does it image the same way, or is it more "copy data, let the OS figure out what changed"? Same question, different phrasing: Does Windows ask you to reboot to apply changes after using it?
|
|
#
¿
Sep 7, 2014 06:50
|
|
|
DaNzA posted:Only a few months away? Did they work out the boot problem with the bios/operating systems itself? Which boot problem? If you mean basic support, Intel 9-series boards have been/are being validated to boot against PCIe drives like the Samsung XP941, since these boards have SATA Express port options. Bottom line, booting is a driver issue, both in the OS and the firmware. For OS support, Windows 8.1 added driver support for booting from NVMe drives, so that's set... as long as you have Windows 8.1. 9-series boards either have PCIe/NVMe boot support already, or the appropriate BIOS updates are in development.
|
|
#
¿
Sep 7, 2014 10:10
|
|
|
The species has been around 200,000 years and still the best way we save stuff for a long time is by making marks in stone.
|
|
#
¿
Sep 9, 2014 03:11
|
|
|
This has nothing to do with the NAND. TechReport has written nearly a petabyte to an 840 vanilla and it's performing exactly as expected - well above the numbers being listed here. This is something weird, somewhere else.
|
|
#
¿
Sep 12, 2014 20:59
|
|
|
bUm posted:Vaguely related: I am, however, a little sad that the new thread's OP did become so Samsung heavy, dropping many considered-to-be-competitive Intel/Sandisk offerings. Kind of seems strange to go through pointing out all the bad brands without giving a fair shake for the other good ones as well for people to more readily research beyond on their own. Unlike the system building thread, this thread is much more geared to "Just buy this, 99% chance you'll be happy, now stop bothering us." And that's absolutely fair - SSD prices fluctuate in a way that a thread OP can't handle and a person not looking to buy would be hard-pressed to keep up with. That's why there's a deals thread in Coupons and Deals for that kind of stuff. And for almost all consumers, merely having an SSD at all is so much better than a hard drive that the difference between a mediocre SSD and the best SSD is almost entirely unnoticeable. Updating OPs is a lot of work, and Alereon erred on the side of a steady recommendation so that his frequent updates would be manageable. Lots of drives were left out, but virtually none of them did anything the 840 EVO doesn't. v  v-- Anyhoo, this 840 EVO bug thing comes at an odd time for me. I literally just switched my laptop over to a 250 GB 840 EVO mSATA. 
|
|
#
¿
Sep 13, 2014 02:33
|
|
|
The MEX controller in the 840 EVO is the same one as in the 850 Pro. I'm sure they have incentive to track it down, at least.
|
|
#
¿
Sep 13, 2014 17:49
|
|
|
I run without RAPID (it broke on suspend until a reboot), but my drive is days old. It would take me a while to get to a testable situation.
|
|
#
¿
Sep 15, 2014 21:01
|
|
|
GreatGreen posted:From what I was reading on overclockers, the problem with the EVOs are specifically limited to the 128/250 GB models. They have found that a secure erase, meaning using a specialized app to totally zero out the drive, will restore performance to brand-new-drive levels. Technically, a Secure Erase on an 840 EVO is an encryption key erasure followed by a drive-wide TRIM, not a zero-fill. But it does eventually leave the drive filled with zeroes. You just can't run, say, DBAN and get the same effect even though in both cases the drive ends up with zeroes all over. In general, Secure Erase on TRIM-enabled drives doesn't do a write pass except as part of post-TRIM garbage collection. If a Secure Erase returns the drive to like-new performance, then we're definitely looking at something related to the controller rather than the NAND. That indirection table hypothesis is looking good.
|
|
#
¿
Sep 16, 2014 02:08
|
|
|
Normal? Years and years and years. Commonly available? I'd bet on Skylake, which will up the number of PCIe lanes off the CPU to 20 - 16 for a GPU, and 4 for whatever (place your bets on an SSD vs. a Thunderbolt link).
|
|
#
¿
Sep 16, 2014 04:38
|
|
|
Bob Morales posted:Here's a dismal benchmark from a 240GB Crucial M500 in my work PC (Dell i5-something or another) That's the same overall pattern as my 840 EVO mSATA:  SATA 3 Gbps interface. Two days ago, this looked the same, but the valleys were not as bad. Checking with defraggler, those dips correspond with LBAs where data is stored, with the spike at the 90 GB mark being the page file. I know this is a sample size of two models, but it looks like the 840 EVO and M500 SSDs have a pattern of being slow on areas where there is old data. I wonder what other drives show. Here's the standard pattern for a hard drive:  I don't know what that big spike is, but it's standard for the drive to be slower on average as it reaches the inner tracks. Here's a GIS image for an Intel X25-M:  Doesn't look like an M500, an 840 EVO, or a hard drive. It's just flat. I don't know if that's new or not. AnandTech's post-trim consistency checks:  840 EVO  Sandisk Extreme II (TLC + SLC cache) Anyone got an in-use SSD of a different model they can run HD Tach against to get a "used" benchmark? I did the Long Bench, so use that for apples-to-apples. I'll run it against my Crucial M4 when I get home.
|
|
#
¿
Sep 17, 2014 23:44
|
|
|
Here's my Crucial M4 256 GB: The peak speeds in the back third seem to correspond with free space. I need a better LBA mapper than Defraggler. I haven't even put this stuff into an Excel spreadsheet, but I'm thinking that slowdowns on old data are not a Samsung-exclusive phenomenon. Certainly the 840 EVO results we've seen are the junkiest results so far, but it's a difference of degree rather than kind. I'm not sure if I'll do this, but if I were to investigate this further, I'd dump all these results into Excel and see what kind of flash was being used, what controllers are showing this effect, and (if possible) try to quantify the crappiness. E: even on gut feelings it's hard to generalize yet - the M4 above me was flat and mine is not. TRIM is definitely working on mine, TRIMcheck-verified. Factory Factory fucked around with this message at 03:44 on Sep 18, 2014 |
|
#
¿
Sep 18, 2014 03:41
|
|
|
Cawd Rud posted:I think I win  I have an inkling of a hypothesis: that what we're seeing is data retention related. Charge leaks from NAND cells slowly but surely, making reads harder. Many controllers have multiple levels of error correction: at least fast and slow, and sometimes heroic. As charge leaks, heavier error correction is needed on reads. The 840 EVO is the worst offender so far because TLC NAND is less tolerant than MLC of this kind of natural degredation, and it also has the slowest read-erase-program times when a slow read is wear-leveled. If this is true, then... *sigh* yes, the TLC NAND would be part of the problem. This does not explain why I am getting this behavior on a drive less than a week old, however. That's a ludicrously short time for data degradation in a system that's been powered every day. Plus the EVO is also equipped with a DSP that makes it much more tolerant of TLC's wandering voltage levels. So the indirection table hypothesis and other possibilities are still definitely in the running. Factory Factory fucked around with this message at 03:57 on Sep 18, 2014 |
|
#
¿
Sep 18, 2014 03:55
|
|
|
I tweeted a couple of these at AnandTech's SSD guy. Especially that  -y 1 TB EVO reading at single-digit MB/s that Cawd Rud posted.
|
|
#
¿
Sep 18, 2014 04:12
|
|
|
Tech Report's 840 got to 300 TB before registering an unrecoverable error. A Secure Erase cleared that and the drive plowed on happily to 800 TB, though it doesn't seem long for this world as it's been logging burned-out blocks for the past hundred TB or so (I'm not looking it up to check). So hopefully you have quite some life left in the drive.
|
|
#
¿
Sep 18, 2014 04:46
|
|
|
dissss posted:Are you guys getting those results consistently? That's an excellent question.  Some spikes are transient, but the pattern is remarkably consistent save for varying magnitudes. -- AnandTech should have something "by the end of the week" on this, so let's chill on the HD Tach graphs I guess.
|
|
#
¿
Sep 18, 2014 14:49
|
|
|
It really varies. But most program loading feels slowest because of random I/O, and even 20 MB/s is still hugely faster than a hard drive. So it's plausible that it'd still feel fast while being bullshit (maybe?!?!?!).
|
|
#
¿
Sep 18, 2014 17:51
|
|
|
WhiteHowler posted:Well poo poo, I just ordered a 500gb 840 evo. We don't know. We are expecting at least AnandTech to say something today about whether this is real or a benchmark artifact, how bad it is, why it's happening, what can be done about it, and what other drives might be affected.
|
|
#
¿
Sep 19, 2014 14:41
|
|
|
BlueInkAlchemist posted:I'm considering getting an SSD for the following reason: My 1TB HDD apparently has a bad boot sector. When I fire up the PC, the only message I get is "A disk read error has occurred. Press Ctrl+Alt+Del to reboot." From what I've found, my data is perfectly fine, but something on the disk hosed up the boot sector so it won't work as a primary drive. So my thinking is to get a new drive, toss Windows on it, and use the 1TB as a slave, with all of my data and games and media on there. Will this work? Is this a good solution? It seems the most straightforward idea, especially if the SSD only has the OS, but I want to be sure before I spend my money. I think this got lost: Your hard drive has failed. Period. Get an SSD if you like, but either one big enough to replace the hard drive, or get a new hard drive as well. Bad sectors that destroy data mean the drive is no longer reliable for storing data.
|
|
#
¿
Sep 19, 2014 23:40
|
|
|
Siochain posted:How the hell are you getting those speeds off a platter drive? My 1tb WD Black drive starts at 140 and drops to 70'ish. My 840 evo is at 50'ish at worst, but most of the drive is at 400+. Probably density. They're up to 1 TB per platter now.
|
|
#
¿
Sep 20, 2014 22:57
|
|
|
Or an 850 Pro.
|
|
#
¿
Sep 21, 2014 15:17
|
|
|
In retrospect, I have pretty poo poo taste in SSDs considering that I've been trying to go with the most problem-free models. I got an Intel 320. I never got the 8 MB bug, but it did fail (and the "drive size: 8 MB, no access to user data" thing seems to be its standard failure mode). I got a Crucial C300 and had constant problems with its multiples-of-60-second race condition bugs that took forever for Crucial to diagnose and partially fix. The C300 also failed, and Crucial gave me an M4 instead. And I just replaced the 320 with an 840 EVO for more capacity and mSATA so I could move the laptop's hard drive out of the Ultrabay and into the shock mount. I also had a 64 GB Kingston SSD of some description. It failed and I sold the RMA replacement NIB. The only entirely non-problematic SSDs I've ever had were V+100 (Toshiba controller) I sold with a netbook that got surplused out of my life, and the X25M I put in my mom's laptop. Two of six different models. That's a lousy percentage. How did I manage to do that when I was trying to buy for reliability?
|
|
#
¿
Sep 21, 2014 19:51
|
|
|
Swagger Dagger posted:Can you expand on this? My desktop HD died so I was going to pick up an SSD today, and the MX100 has very good reviews, especially for the price point and considering the Evo issues. It's just a bit slow.  Great features - low-power idle and TCG Opal encryption support, notably. But cruddy performance for an SSD. That said, it's still good enough that it's a huge step up over a hard drive.
|
|
#
¿
Sep 22, 2014 17:54
|
|
|
Sometimes I feel like the best way to have a computer that works well is to never install software.
|
|
#
¿
Sep 22, 2014 22:17
|
|
|
fookolt posted:Uh, so my Samsung EVO 840 has been on for 6103 hours and it says it's written 20 TB in that period of time...That's not normal, right? That's 78 GB per power-on day. How many hours per day do you use the machine? Because 8 hours a day of use is 26 GB per day - a smidge high but totally reasonable for heavy use.
|
|
#
¿
Sep 23, 2014 03:31
|
|
|
Xenomorph posted:Companies like Intel know what the gently caress they are doing, and they make 480GB SandForce drives. Do you really think a company like Intel would make 480GB SandForce drives if there was something wrong with them and they "don't work well"? I like Intel, but this is the company that made a CPU that couldn't divide, a SATA controller that murdered itself, GPUs marketed for media playback that couldn't play back a DVD at standard NTSC framerates, an SSD that lost its poo poo if you jiggled the power cord wrong, a uniquely-identifiable serial number on your CPU that spat itself out after every CPUID instruction, a headlining HPC feature that doesn't work on its debut silicon, Itanium IA-64, and Netburst. Meanwhile, before TRIM, when JMicron drives had IOPS measured in tenths and the X25-M was but a glimmer in Intel's eye, Samsung SSDs never stuttered and had top-class performance consistency and access times. I'm getting pretty sick of the tech warfare around here the past few days. Factory Factory fucked around with this message at 07:59 on Sep 23, 2014 |
|
#
¿
Sep 23, 2014 07:57
|
|
|
That's what I did. That said, I'm not using RAPID because it would break and disable itself every time I slept my laptop. But I'm pretty sure that's a one-off problem rather than an inherent problem of downloaded Samsung Magician.
|
|
#
¿
Sep 25, 2014 07:03
|
|
|
Some hard drives just have exceptionally long lives. It skews the mean time before failure away from the median time before failure. Same thing happens in MBTF calculations for light bulbs, where most will fail before the MBTF has come to pass, but then you get outliers. E: VVV I acknowledge your correction. Factory Factory fucked around with this message at 16:37 on Sep 30, 2014 |
|
#
¿
Sep 30, 2014 16:34
|
|
|
Agreed posted:Super quick, is Clonezilla a tool to get and learn to use for migrating my system? Someone else mentioned it and that isn't the first time I've heard it said here. Will it migrate data with soft links in place, or do I need to try to unfuck that mess myself and then back up & restore/resize? Macrium Reflect will copy shortcuts, symlinks and junctions as such without following them and hitting the files they point to. Crashplan does, too, incidentally; it might be the default handling for backup programs that use Volume Shadow Services for backup.
|
|
#
¿
Oct 7, 2014 16:04
|
|
|
|
| # ¿ Apr 19, 2024 16:22 |
|
|
fookolt posted:Thanks; I was actually going to move it to my Crucial M4 (which is working fine). Multiply the 120 GB version by 8, pretty much.
|
|
#
¿
Oct 11, 2014 10:17
|
|