|
There would only be a noticeable difference if you filled the drive up past 90%, in which case the 850 Pro's superior I/O consistency would win out. Otherwise, the 840 EVO's just fine, especially for the price.
|
# ? Aug 14, 2014 23:23 |
|
|
# ? Mar 29, 2024 07:25 |
|
I cloned my old SSD to a new one and did what Factory Factory said here to switch them in Windows:Factory Factory posted:You unplug the old drive and boot from the new one to make sure it works in the first place. Then change the desktop background as a sanity check that you are booting from the correct drive. After that, change the boot order in the BIOS/UEFI setup and/or move plugs so that the new SSD is plugged into the first SATA port. Then, when booted to the new SSD, then clean the old SSD, create a new partition of maximum size, quick format the partition as NTFS (this issues a TRIM command to the entire drive), and then you can repartition the SSD further if you like. But now Trimcheck says trim isn't enabled on the old SSD (which is now F.) I changed the boot order rather than swapping cables. Is this maybe just because the disc usage is currently zero, or is something not working? This is on Windows 8.1, and the defrag utility and RST both detect that it's an SSD.
|
# ? Aug 15, 2014 00:01 |
|
Factory Factory posted:So Is it time to rethink OCZ/Toshiba drives? At least, when/if they send it to AnandTech and we see if it's a convincing alternative to an 840 EVO. TR's consistency benchmarks are interesting, but I don't feel that they're as robust as AnandTech's performance consistency benchmarks. Aphrodite posted:But now Trimcheck says trim isn't enabled on the old SSD (which is now F.) I changed the boot order rather than swapping cables.
|
# ? Aug 15, 2014 01:57 |
|
Alereon posted:What motherboard do you have, and what SATA ports do you have the drives connected to? P8Z77-V LX. The new, trim working drive is on the second slot I guess since it gets picked up as disk 1. The old, previously working when it was the Windows drive SSD is disk 0. Trim worked on this drive when it was still my system drive, and it's still physically connected to the same place. Disk 1 did trim when I ran it manually from the defrag program though, so it is working. Just not automatically apparently.
|
# ? Aug 15, 2014 02:05 |
|
How full was the drive? You had just trimmed it in its entirety and it might not have erased everything yet, since a TRIM doesn't automatically mean a block is erased right away. It just means that the drive is free to erase it at the first convenience. Different controllers schedule erases at different times.
|
# ? Aug 15, 2014 02:39 |
|
It had 133gb on it, of whatever free space you get from a 256. It's a Mushkin drive. Defrag did trim it, so I'm not worried since I assume that means the daily/weekly/whatever scan that's scheduled will do a pass.
|
# ? Aug 15, 2014 02:44 |
|
I need a back up solution for my work-related data. I've already had 2 WD external HDs die on me in less than 2 years, so I'd like to explore some alternate options. I was thinking of buying an SSD dock (something like the Thermaltake Docking Station. Or even this Seagate adapter). Any recommendations for reliable SSD dock?
|
# ? Aug 15, 2014 07:55 |
|
That Seagate dock is ridiculously priced. Just get a USB3 one like this.
|
# ? Aug 15, 2014 09:18 |
|
td4guy posted:That Seagate dock is ridiculously priced. Just get a USB3 one like this.
|
# ? Aug 15, 2014 09:45 |
|
Just keep in mind that SSDs may not be the best solution for backup if you're not keeping them powered for extended periods of time (>1 year).
|
# ? Aug 15, 2014 12:51 |
|
Seconding smax. Head over to the packrat thread and check out a NAS.
|
# ? Aug 15, 2014 14:42 |
|
smax posted:Just keep in mind that SSDs may not be the best solution for backup if you're not keeping them powered for extended periods of time (>1 year). td4guy posted:That Seagate dock is ridiculously priced. Just get a USB3 one like this.
|
# ? Aug 15, 2014 15:32 |
|
Let's be absolutely clear, because I'm not sure you gathered this: If you put an SSD in a closet for one year, you will lose the data on it. SSDs must be powered on to store data long-term.
|
# ? Aug 15, 2014 15:39 |
|
Factory Factory posted:Let's be absolutely clear, because I'm not sure you gathered this: If you put an SSD in a closet for one year, you will lose the data on it. SSDs must be powered on to store data long-term. I thought they just require a constant source of heat, kinda like a lizard. I may be mistaken, though.
|
# ? Aug 15, 2014 16:50 |
|
Solid state drives have very little in common with a lizard.
|
# ? Aug 15, 2014 17:00 |
|
Lizards have been at the 10c/GB mark for while now.
|
# ? Aug 15, 2014 17:04 |
|
Factory Factory posted:Let's be absolutely clear, because I'm not sure you gathered this: If you put an SSD in a closet for one year, you will lose the data on it. SSDs must be powered on to store data long-term. melon cat fucked around with this message at 17:27 on Aug 15, 2014 |
# ? Aug 15, 2014 17:23 |
|
melon cat posted:I had no idea about that. So what should I be getting if my storage device is going to be spending a lot of time in a closet? Because that will probably be the case. I just back up my data every 3-6 months or so, then store the hard drive away for several months. Should I just get a high capacity USB thumb drive, like this one? Absolutely not. The problem is inherent to NAND flash, and a USB drive will have the same issue as an SSD because they both use NAND. Often worse, because the NAND in USB drives is crappy stuff that wasn't suited for SSDs. Hard drives have a similar problem for different reasons and should be fully rewritten every two years at most when stored long-term, as well as the data stored with data recovery information (such as parity info a la RAID 5) if possible. The drives should be powered on every few months to make sure the moving parts haven't stuck. If you want backups for archival, you have two choices: 1) Proper archival media, like tape or archival-grade optical discs. 2) Make it somebody else's problem and pay for a service like Carbonite.
|
# ? Aug 15, 2014 17:37 |
|
Factory Factory posted:Absolutely not. The problem is inherent to NAND flash, and a USB drive will have the same issue as an SSD because they both use NAND. Often worse, because the NAND in USB drives is crappy stuff that wasn't suited for SSDs.
|
# ? Aug 15, 2014 17:58 |
|
melon cat posted:Thanks for the info. But I have a question about anyone's experiences with cloud-based storage. What happens to your files if that company shuts down? I'm only asking because back in the '90s (holy poo poo I'm ancient) I got burned by Angelfire when they went belly up overnight. Without warning they just folded with the dot-com bust, and I lost all of the data that I stored with them. The whole experience made me weary of trusting other companies with my data. I know how strange that sounds because things are much different now, but it sucked to go through. Sounds like you know what could happen. Better to stick with big companies and hope for the best. Maybe use two?
|
# ? Aug 15, 2014 17:59 |
|
There was a pretty neat thread awhile back concerning archival and long term storage solutions(5 year+) and there really isn't a great answer for it. Making it someone else's problem is your best bet.
|
# ? Aug 15, 2014 18:04 |
|
melon cat posted:Thanks for the info. But I have a question about anyone's experiences with cloud-based storage. What happens to your files if that company shuts down? I'm only asking because back in the '90s (holy poo poo I'm ancient) I got burned by Angelfire when they went belly up overnight. Without warning they just folded with the dot-com bust, and I lost all of the data that I stored with them. The whole experience made me weary of trusting other companies with my data. I know how strange that sounds because things are much different now, but it sucked to go through. Use multiple companies if it's that important to you. It's not like they cost a huge amount of money.
|
# ? Aug 15, 2014 19:19 |
|
If you lose data because your backup company disappeared overnight, you've done something wrong.
|
# ? Aug 15, 2014 19:22 |
|
melon cat posted:Thanks for the info. But I have a question about anyone's experiences with cloud-based storage. What happens to your files if that company shuts down? I'm only asking because back in the '90s (holy poo poo I'm ancient) I got burned by Angelfire when they went belly up overnight. Without warning they just folded with the dot-com bust, and I lost all of the data that I stored with them. The whole experience made me weary of trusting other companies with my data. I know how strange that sounds because things are much different now, but it sucked to go through. Uhh, you should've backed up any data on Angelfire, if Angelfire AND your backup company had gone belly up at exactly the same time then well then you might as well play the lottery. Backup to one company, the odds that they go belly up without warning at the same time your computer dies are astronomically low. If even such low risk is out of the question for your precious datas you can do two companies. For my personal stuff I just use one company and keep multiple local copies (Crashplan to my local NAS and backblaze to ~~ the butt ~~).
|
# ? Aug 15, 2014 22:41 |
|
I think he means he only had the data on Angelfire with no local copy of it, which, if you're paying someone else to take care of it so you don't have to have your own safe full of tapes, seems reasonablesque. Can a processor be a bottleneck in the read speeds of an SSD or is it not really involved enough in the RAM<->Storage dance to cause issue?
|
# ? Aug 16, 2014 01:29 |
|
deimos posted:Uhh, you should've backed up any data on Angelfire, if Angelfire AND your backup company had gone belly up at exactly the same time then well then you might as well play the lottery. In either case, it seems like Carbonite's my best option, so I'll probably go with that. melon cat fucked around with this message at 02:53 on Aug 16, 2014 |
# ? Aug 16, 2014 02:42 |
|
Wulfolme posted:Can a processor be a bottleneck in the read speeds of an SSD or is it not really involved enough in the RAM<->Storage dance to cause issue? In theory, yes, a CPU can bottleneck, but in practice, CPU overhead on SATA has been a non-issue for quite a while (I think people stopped measuring it ca. Core 2 when it was reliably less than 2% under the heaviest workloads). Even software RAID 5 parity calculations are pretty trivial these days, except in enterprise environments where ever FLOP counts. On NVMe, CPU overhead is even less. And as for bandwidth, PCIe is generally slower than the CPU to DRAM interface, even if you're using all of the lanes. I think that system integrators are generally good at matching storage capability to CPU capability. Like, you rarely see Atoms with more than a couple SATA ports and a PCIe lane or two, whereas enterprise storage is always matched with many-core Xeons with 40 PCIe lanes and quad-channel RAM. E: Here's some enterprise numbers from AnandTech's DC P3700 review, http://www.anandtech.com/show/8104/intel-ssd-dc-p3700-review-the-pcie-ssd-transition-begins-with-nvme/5 A good enterprise SSD runs about 6000 IOPS per % of an i7-4770 (HT disabled). Another thought: I guess if you weren't using a NAS/SAN, this'd be a lot more relevant... Factory Factory fucked around with this message at 03:13 on Aug 16, 2014 |
# ? Aug 16, 2014 03:06 |
|
melon cat posted:This was during the dot com bust. You don't seem to be aware of the fact that there weren't many data back-up options available at that time. There wasn't a bunch of companies offering cloud storage like there are today, and affordable external HDDs sure as hell weren't available. All I was trying to say was that my experience with Angelfire has made me skeptical of relying on outside companies with my data. FWIW, I think goons mostly recommend CrashPlan.
|
# ? Aug 16, 2014 05:36 |
|
Factory Factory posted:Hard drives have a similar problem for different reasons and should be fully rewritten every two years at most when stored long-term Hm, it sounds like good practice, but I recently pulled out a 245MB Seagate that had been sat in my loft without power for over 10 years, and the data was all perfectly intact. Maybe because the density is low, it is more reliable.
|
# ? Aug 16, 2014 09:17 |
|
HalloKitty posted:Hm, it sounds like good practice, but I recently pulled out a 245MB Seagate that had been sat in my loft without power for over 10 years, and the data was all perfectly intact. Maybe because the density is low, it is more reliable. One 500GB SATA drive had smart errors, which I just must not have noticed before. One 160GB PATA drive just didn't spin up anymore. Both the 40GB and the 200GB PATA drives seemed to work at first and then started overheating on large transfers and dropping away (hardened lubrication increased resistance and the power needed to overcome it?). I've got a 320GB PATA and a 500GB SATA drive left that are doing nicely. The point was never that a drive never could survive being unpowered for a long time, it's that the statistical likeliness of it failing increases a lot. I also think data integrity wasn't as much the point as opposed to mechanical failure. If it spins up and keeps running, data will probably be okay. If I could lift the platters on the broken ones and read from them, my data would probably still be there too. Anyway, that's my anecdote and the reason I've added a three monthly calendar event to spin up those remaining drives. Fingers crossed it helps.
|
# ? Aug 16, 2014 11:03 |
|
Mechanical failure is a big part, but the magnetic encoding on a hard disk platter can also be gradually demagnitized (as in, the domain alignment reverts to an unaligned state). When that happens, the de-aligned bits result in a bad sector (which technically could be overwritten to make it a good sector again with a zero fill or something, but the data would be lost). Higher density media and exposure to magnetic fields can make this happen sooner rather than later.
|
# ? Aug 16, 2014 11:27 |
|
Right, right. Somehow I've managed to ignore the whole context of that post. Rewriting data is something I haven't planned for yet either.
|
# ? Aug 16, 2014 11:46 |
|
I worked at a place that backed up Synology units to HD's using a dock. Monthly backups would be put back in a static bag and placed in a fire safe. Weekly backups would rotate. There were times I would grab a monthly set and half the drives in the set wouldn't be recognized and had to be RMA'd. Anecdotal yes, but I don't trust hard drives that aren't powered on.
|
# ? Aug 16, 2014 13:12 |
|
So I'm working on a new OP for a thread reboot, my main goal is to front-load useful information and put theory and stuff farther back. Here are some sections and subsections, any comments? What SSD should I buy? -You (the person reading this) are a normal desktop user and should buy a Samsung 840 Evo -If you have a Mac get a Samsung 850 Pro -If you're hardcore buy a Samsung 850 Pro -If you like dirt cheap and janky maybe get a Crucial MX100 or PNY Optima -List bad brands How do I install my new SSD? -Fresh installation recommended, disconnect all other drives, update BIOS, connect SSD, install Windows, install drivers and software -Image over using Macrium or the Samsung software -Setup on older computers including overprovisioning How do I maintain and use my SSD? -Enable RAPID but don't use OS Optimization on Samsung drives -Don't overfill, discussion of performance vs fill -Don't move pagefile to other drives -Keep frequently used apps and system folders on the SSD SSD Reliability SSD Endurance Data lifespan/fade -Probably won't be an issue but act like it will be or eventually it will bite you in the rear end
|
# ? Aug 16, 2014 16:06 |
|
Alereon posted:SSD Endurance Please put this section in bold and with sirens to announce that people shouldn't worry about endurance with a link to the Techreport thing to back it up. And maybe elaborate a bit on the bad changes (Samsung's) OS optimization makes and how to return them to sane values if you read that advice too late.
|
# ? Aug 16, 2014 16:17 |
|
Alereon posted:Crucial MX100 I think that's a bit TOO janky. It didn't show up too much in AnandTech's review, but in TR's recent OCZ Arc 100 review, the MX100 had a terrible, terrible showing in their "time spent being lovely" metric.
|
# ? Aug 16, 2014 16:42 |
|
Geemer posted:Please put this section in bold and with sirens to announce that people shouldn't worry about endurance with a link to the Techreport thing to back it up. quote:And maybe elaborate a bit on the bad changes (Samsung's) OS optimization makes and how to return them to sane values if you read that advice too late.
|
# ? Aug 16, 2014 16:43 |
|
Alereon posted:Oh definitely, though do you know of any easy ways for someone to determine how much data they write a day? I keep seeing people wringing their hands because no matter how much we say they'll be fine, they worry, so it'd be cool if there was a Windows performance counter or something they could look at and say "oh it turns out I only write 1GB/day I'm fine." Unfortunately I can only think of this working if they already had an SSD. SSDs generally track host writes, but that's less than helpful for someone who doesn't own an SSD. You could get some examples and testimonials from folks. I've had this warranty replacement SSD for about 3 weeks, for example, and done ~370 GB of writes. That includes restoring an 80 GB image, pruning it to 40 GB, then doing 6 months of updates, 20 program uninstalls and as many new installs, to bring it back up to 60 GB, then daily work hours use (8 hours a day) since then. So a heavy 17 GB a day on average, or about 12 GB/day after getting set up. Jesus. Most of that is just Chrome cache-thrashing and page file use from leaky tabs. Well, that's roughly in line with Anand's use at his workstation that set the standard for the Destroyer benchmark.
|
# ? Aug 16, 2014 17:00 |
|
Factory Factory posted:I think that's a bit TOO janky. It didn't show up too much in AnandTech's review, but in TR's recent OCZ Arc 100 review, the MX100 had a terrible, terrible showing in their "time spent being lovely" metric. Haha, wow. Sometimes slower to initially serve a write than a hard drive. (The Seagate hybrids don't cache writes if I recall correctly).
|
# ? Aug 16, 2014 17:01 |
|
|
# ? Mar 29, 2024 07:25 |
|
Factory Factory posted:I think that's a bit TOO janky. It didn't show up too much in AnandTech's review, but in TR's recent OCZ Arc 100 review, the MX100 had a terrible, terrible showing in their "time spent being lovely" metric. E: Actually the new JMicron controller is more mediocre than astoundingly awful, so substitute your favorite poo poo controller+firmware combo of choice here that sells for peanuts. Phison drives, Marvell plus firmware somebody's brother threw together, etc. Edit 2: I'm at 509GB written in a month, but that is on a new install, and includes 100GB of Steam games. I currently have 151GB used...for a nearly identical 12GB/day average in transitory data. It'll be interesting to see how that stabilizes now that I'm mostly done installing and deleting stuff. Alereon fucked around with this message at 17:12 on Aug 16, 2014 |
# ? Aug 16, 2014 17:04 |