Hard Light Productions Forums
Off-Topic Discussion => General Discussion => Topic started by: est1895 on October 22, 2011, 06:24:38 pm
-
I was wondering what kind of SSD drive to get, since I have heard many bad things about them. :confused:
It is for an AMD and Intel System. (2 different towers) :lol:
-
I have this (http://www.newegg.com/Product/Product.aspx?Item=N82E16820148447) Solid State Drive drive and it's serving me well. Solid State Drive drives really are good pieces of hardware. You can't beat a good Solid State Drive drive!
-
They're still so expensiiiiiive.
Well... compared to hard drives.
-
too expensive, better get a meaner gpu.
-
I dunno I personally think it's worth being able to boot in eight seconds.
-
Yea my PCs boot times never get old.
Though lets be honest, most of us leave our computers on 24/7 anyway so boot times aren't that important...
Load times on your favourite game though.... hell yes!
-
I prefer this SSD drive.
(http://images4.wikia.nocookie.net/__cb20090123043656/starwars/images/thumb/5/5e/Executorendor.jpg/364px-Executorendor.jpg)
-
Lol.
Even if you leave your machine up all the time, fewer moving parts means longer life.
-
Unless write cycles exceed the amount the SSD is specified for.
Solid state memory has limited amount of writes for each memory location. With repeated writes on the same location, that spot will eventually be rendered unwriteable (or unreliable) and, if drive is smart, will be mapped out and eventually the size of useable memory, or worse, drive reliability, will start to suffer from that.
Available memory getting smaller would be better option than losing file integrity without warning.
This means that SSD's are ideal for storing information that needs to be accessed fast (like loading programs or operating system) but produces very small amount of writing on the drive. If you install operating system or programs on SSD, you might want to disable page file on that particular drive, and either get a smaller, more "disposable" SSD for fast page file, or just use HDD for page file.
Paging data in and out of memory would be the type of continuous writing that would wear SSD's out the fastest, so if you want to maximize the life time of your SSD, disable the pagefile on it, and use it as little as possible for saving actual data on.
However, it's worth noting that the fast read/write speeds for SSD's would also make them ideal for being used as a page file; it would certainly mean faster operations with memory paging, than with a regular HDD. So, if you have the monies to shell out for a small, say, 16-32 GB SSD just for a page file (disable page file for all other drives but that particular one, set page file to the drive's full size, and leave it otherwise empty), that might work out quite well.
Of course, the ever increasing amounts of memory means going to page file is a rare occurrence. In my case, that requires heavy render work on Blender with huuuge textures and render resolutions taking up massive amounts of memory, but I have had Blender go into page file, and naturally that slows things to a crawl.
Has there been any benchmarks on how much better performance SSD's have compared to HDD's when stuff goes into pagefile territory, and compared to RAM operations? :nervous:
-
I took the plunge about a year ago. Best upgrade I've ever made to my system. Windows starts up with no waiting. Applications loaded on to it start... with no waiting. It's just brilliantly fast. I got a Mushkin Callisto 2 60GB and it's sufficient for OS and core programs. I have a 1TB Caviar Blue to back it up for all of my things like games and music. Since last year the prices have dropped and the cost for a larger drive is quite a bit less than I paid for this one.
I figure in 10 years or so we'll see SSD's start to take over. Until then we'll see SSD's alongside traditional platter drives. There are even some hybrid models out there worth having a look at too.
-
Did the hybrid tech ever get realized? I heard about it just before SSDs went slightly mainstream. But I never heard of any finishing development.
-
Solid state memory has limited amount of writes for each memory location. With repeated writes on the same location, that spot will eventually be rendered unwriteable (or unreliable) and, if drive is smart, will be mapped out and eventually the size of useable memory, or worse, drive reliability, will start to suffer from that.
yes, but how high is that limit? low enough to present a very real drawback, or long enough to be outside an average lifecycle of the drive? long enough to even be a net benefit over mechanical? EVERYTHING ultimately has an endurance limit, but it may not be THE limiting factor.
-
It was in drives when I was looking at buying them, I don't know about now, but it's a very real drawback, SSDs are much less likely to last as long as a normal HDD.
-
Lol.
Even if you leave your machine up all the time, fewer moving parts means longer life.
granted i have hard drives from 8 years ago that still work great. but as i understand flash technology, it has long term reliability issues. limited write cycles are limited, thats just the way flash works. much of the ssd technology revolves around wear leveling to compensate the inherent issue of memory cell degradation. that said i still like the technology, and will probably switch to them once their prices are more affordable with respect to mechanical hard drives. ssds have other bonuses such as removing inductive loads (motors), and the electrical noise that comes with them, from the system, requiring lower voltages, and generally reducing overall power consumption. my next computer or two will likely have mechanical drives.
-
Just have both, mechanical for storage/less often used stuff, SSD for regularly used/system stuff.
Then it's all gravy.
What's more, if you can avoid writing to the SSD much after you've installed system stuff on it the drive lasts a significantly longer period than the 'suggested' lifetime, which I've probably exceeded on mine now.
-
In statistical terms, write access really is the exception. If you're putting largely static bits of data on your SSD, like the OS and programs, and put the stuff that gets the most write hits like the pagefile onto a normal HDD, you'll get a very good lifetime out of the disc.
Also keep in mind that SSDs have wear-levelling hardware built in that will distribute the data written to it as evenly as possible in order to maximize the lifetime, and that even in the worst case you are looking at a very gradual failure process with plenty of opportunity to back things up.
-
As reliability was my primary concern the only option was Intel (for me).
-
working with microcontrollers it was interesting to read in the datasheet that a flash cell can be written to about 10k times, while an eeprom cell can be written to 100k times. i had avoided using the eeprom because i was always concerned id wear it out and make it useless, but long before that happens it will be impossible to reprogram the flash.
-
Unless write cycles exceed the amount the SSD is specified for.
Solid state memory has limited amount of writes for each memory location. With repeated writes on the same location, that spot will eventually be rendered unwriteable (or unreliable) and, if drive is smart, will be mapped out and eventually the size of useable memory, or worse, drive reliability, will start to suffer from that.
Available memory getting smaller would be better option than losing file integrity without warning.
This means that SSD's are ideal for storing information that needs to be accessed fast (like loading programs or operating system) but produces very small amount of writing on the drive. If you install operating system or programs on SSD, you might want to disable page file on that particular drive, and either get a smaller, more "disposable" SSD for fast page file, or just use HDD for page file.
Paging data in and out of memory would be the type of continuous writing that would wear SSD's out the fastest, so if you want to maximize the life time of your SSD, disable the pagefile on it, and use it as little as possible for saving actual data on.
However, it's worth noting that the fast read/write speeds for SSD's would also make them ideal for being used as a page file; it would certainly mean faster operations with memory paging, than with a regular HDD. So, if you have the monies to shell out for a small, say, 16-32 GB SSD just for a page file (disable page file for all other drives but that particular one, set page file to the drive's full size, and leave it otherwise empty), that might work out quite well.
Of course, the ever increasing amounts of memory means going to page file is a rare occurrence. In my case, that requires heavy render work on Blender with huuuge textures and render resolutions taking up massive amounts of memory, but I have had Blender go into page file, and naturally that slows things to a crawl.
Has there been any benchmarks on how much better performance SSD's have compared to HDD's when stuff goes into pagefile territory, and compared to RAM operations? :nervous:
What's a page file?
-
Lrn2wiki (http://en.wikipedia.org/wiki/Paging).
A pagefile is used to store memory pages (that is, chunks of content from main system RAM) that have not been accessed in a while, and for which there is no room in system RAM anymore due to another program needing the active memory, but which can't be unloaded completely because the program that allocated that memory is still actively running. So these bits of memory are written out to the HDD. Most Linux distributions dedicate a small partition on the HDD to this duty exclusively, while Windows uses a hidden file on the main system drive called pagefile.sys.
-
Lol.
Even if you leave your machine up all the time, fewer moving parts means longer life.
granted i have hard drives from 8 years ago that still work great. but as i understand flash technology, it has long term reliability issues. limited write cycles are limited, thats just the way flash works. much of the ssd technology revolves around wear leveling to compensate the inherent issue of memory cell degradation. that said i still like the technology, and will probably switch to them once their prices are more affordable with respect to mechanical hard drives. ssds have other bonuses such as removing inductive loads (motors), and the electrical noise that comes with them, from the system, requiring lower voltages, and generally reducing overall power consumption. my next computer or two will likely have mechanical drives.
Don't disagree...but I suspect that my SSD will wear out about as fast as the mechanical ones with less of a catastrophic failure at the end when a drive head fails and screws up a bunch of data in the process. If I remember right the expected lifespan of the current gen SSD's are about 5 years or so given average usage. Standard hard drives last about that long too.
-
Don't disagree...but I suspect that my SSD will wear out about as fast as the mechanical ones with less of a catastrophic failure at the end when a drive head fails and screws up a bunch of data in the process. If I remember right the expected lifespan of the current gen SSD's are about 5 years or so given average usage. Standard hard drives last about that long too.
But Thats just it. Define "Average use". Intel SSDs come with a 5 year warranty (damn good for any SSD), but does that cover natural wearing to the point of unusability? Most companies will blame it on you writing too much to the disk and you're out of luck. How do you replace a controller card on an SSD to recover the data if it fries? (which I have already had happen to me with my 1st SSD which has now been RMA'd). To top this off, any HDD that lasts -only- five years is bad quality. I've seen tons of poor old 40-80GB IDE drives from Dimension 8300s that live for eight years, then continue to serve as I find uses for them in other old computers. I expect them to live that long.
If you do use an SSD, it would probably be a very good idea to disable pagefile and hibernation on it to help it last longer, and to enjoy those < 20 second boot times.
-
What i want to do is go a step further with flash storage. Buy some good thumb drive or sd memory card in a card reader, boot to ram, and user data storage would just take place on a normal hard drive. Yes boot time will suffer, but everything be fast as hell if you have enough ram for the os and open programs. It would be low cost and see how much head room i really have with 4gb's of ram.
On boot time. I usually mitigate boot time altogether by leaving the computer booted, using suspend to ram (my computer's ready to use in 5 seconds of hitting the on button) works great powering down everything but your ram for when you do power down. Of which case, i find suspend to hard disk to be no longer relevant (except possibly the few special cases it's used for god knows what by others i don't know) nowadays since boot time has improved so much on the software optimizing side of the subject (faster just to just boot a computer like normal with a modern day os, than to restore a session stored on the hard drive).
Don't disagree...but I suspect that my SSD will wear out about as fast as the mechanical ones with less of a catastrophic failure at the end when a drive head fails and screws up a bunch of data in the process. If I remember right the expected lifespan of the current gen SSD's are about 5 years or so given average usage. Standard hard drives last about that long too.
I don't disagree with your estimate of ssd wear out time either (ssd's intended as upgrade and replacement of magnetic storage work fantastically), even for normal magnetic hard drives. Although, i was able to see the longevity of normal magnetic storage for many a year.
9 years is how long an average hard drive from 2001 in a work environment will take to die. Work environment being virus scans, defrags, on 8 hours a day, surfing web, email, word processing, etc (what i consider average usage in a work environment).
5 years is a great time to replace magnetic storage. Lest trust your data with uncertainty past that. I'm not in any way recommending 9 years. **** that. People will likely scrutinize me for saying 9 years, but this 9 years of hard drive longevity is another story for another thread on a different day about bosses not wanting to listen.
-
get one of those dram drives and fill it with your unused old ram and use that for your pagefile. computers are starting to have as much ram as they did hard drive space 10 years ago. at some point these machines will go obsolete and we will need something to do with all that ram. using it as a ultra high speed page file sounds hella awesome complement to an ssd.
i really wish we could come up with some revolutionary way to make sram smaller, at least die space equivalent to dram. sram is currently only used as cpu cache and main memory in microcontrollers. its very fast, but very expensive in terms of cost and die space. but the other feature it has is a very low power consumption at idle. making it easy to power one of these things with a small battery or even a supercapacitor for very long periods of time. of course when that happens say bye bye to booting all together.
-
I've heard SSDs include the ability to basically map data usage out so that no one are of storage gets over-used, instead usage is distributed evenly throughout the drive; it's transparent so the requests are still the same (the OS doesn't see this process, the drive takes care of it by itself).
-
I've heard SSDs include the ability to basically map data usage out so that no one are of storage gets over-used, instead usage is distributed evenly throughout the drive; it's transparent so the requests are still the same (the OS doesn't see this process, the drive takes care of it by itself).
http://www.hard-light.net/forums/index.php?topic=78737.msg1557425#msg1557425
i heard that too
-
I've heard SSDs include the ability to basically map data usage out so that no one are of storage gets over-used, instead usage is distributed evenly throughout the drive; it's transparent so the requests are still the same (the OS doesn't see this process, the drive takes care of it by itself).
http://www.hard-light.net/forums/index.php?topic=78737.msg1557425#msg1557425
i heard that too
i see what you did there.
read the ****ing thread, lol.
-
I did... or rather, I skimmed it... missed the second part of the E's post. Was really only looking for a mention of wear-leveling but I missed it, so I mentioned it. :doubt: Gotta get better at speed-reading...
-
Good thing to mention next in the area of flash drives wearing out. Don't defrag them. You wont get any performance benefit out of flash storage doing this.
Defragging is meant to make a magnetic hard drive's read write head move less since it can find all of the data it needs to find in a smaller area of the disk as opposed to looking for data spread out all over it. This is why defragging really can speed up load times on magnetic hard drives.
Defragging a flash drive will add no performance benefits and will rather instead, make it go through lots and lots of reads and writes for no good reason (sounds like a good torture test for killing flash storage on purpose though). Flash drives organize data themselves for wear leveling so you wont wear out one area of the storage more than another (defragging ****s with that). I get asked this question all the time about defragging flash drives; in a nutshell, i just tell people it's a really great way to decrease the longevity of it really fast.
Defragging: only meant for magnetic hard drives...pointless for storage with no moving parts.
-
what he^ said. flash is random access, it can be fragmented to hell and back for and the speed loss is negligible (sequential read is faster than random read, because of switching times, on the order of nanoseconds).
-
Caveat: The NTFS filesystem might get its panties in a bunch if it wasn't evar defragged... IMHO defrag an SSD once a year to keep your filesystem sane.
-
You go ahead and do that.
I just wont use a journaling file system on flash storage.
-
wasnt windows 7 supposed to come with ssd-friendly features?
-
It'd be cool if the windows disk defragmenter would restrict defragmenting of flash storage.
Flash storage friendly features in windows 7? I don't know anything about that...must research :confused:
There is the usage of exfat which is a lot friendlier than journaling file systems on flash storage. The exfat driver for linux is in beta, so linux is currently lacking for the moment, but macosx and of course windows support it.
-
wasnt windows 7 supposed to come with ssd-friendly features?
It does and everything I've read suggests it works fairly well. When I installed Win 7 on my SSD it immediately figured it out and disabled some features and enabled others to take advantage of the drive.
-
Don't disagree...but I suspect that my SSD will wear out about as fast as the mechanical ones with less of a catastrophic failure at the end when a drive head fails and screws up a bunch of data in the process. If I remember right the expected lifespan of the current gen SSD's are about 5 years or so given average usage. Standard hard drives last about that long too.
But Thats just it. Define "Average use". Intel SSDs come with a 5 year warranty (damn good for any SSD), but does that cover natural wearing to the point of unusability? Most companies will blame it on you writing too much to the disk and you're out of luck. How do you replace a controller card on an SSD to recover the data if it fries? (which I have already had happen to me with my 1st SSD which has now been RMA'd). To top this off, any HDD that lasts -only- five years is bad quality. I've seen tons of poor old 40-80GB IDE drives from Dimension 8300s that live for eight years, then continue to serve as I find uses for them in other old computers. I expect them to live that long.
If you do use an SSD, it would probably be a very good idea to disable pagefile and hibernation on it to help it last longer, and to enjoy those < 20 second boot times.
Typically my systems tend to last about 5 years anyways before a power surge or some other malady happens to the system that makes it difficult to work with and ultimately time for a new one. I feel pretty good out of my last system that got 6 and a half solid years before a transformer in the nighborhood blew up (I saw the glare from the explosion) and it never worked very well after that. Never could figure out what was wrong with it either. So... I built this one. It's a year old and I figure it's got 4-5 great years ahead of it before it gets zapped by lightning or hit by a freight train :D
But in all seriousness I think 5 years is about what I trust my standard HDD's to last. Some of them have lasted far longer and the odd one has failed. In four years time when this SSD starts to reach the end of it's useful lifespan I'll start to phase it out and get a new one. Probably with something that is significantly faster, bigger and cheaper.
-
Or you could go for the 9 year attempt :drevil:
-
It'd be cool if the windows disk defragmenter would restrict defragmenting of flash storage.
It does, in a sense. If Windows 7 detects SSD drive, it unselects them by default in disk defragmenter. AFAIK you can still re-select them manually however. For W7 to detect SSD's, you need to have the latest service pack and subsequent updates as well as have AHCI enabled in both BIOS and W7. Depending on chipset, it may also be necessary to update chipset drivers. Updating BIOS and SSD firmware shouldn't be necessary for this, but you never know.
Caveat: The NTFS filesystem might get its panties in a bunch if it wasn't evar defragged... IMHO defrag an SSD once a year to keep your filesystem sane.
Don't, just don't. I repeat, do NOT defrag SSD drive even if it is once in 10 years.
-
Yeah, it'd be cool if windows restricted the defragmentation via the properties window of said flash storage drive.
At least they did something.
Now, it'd be cool if windows supported partitioning of flash storage. It's stupid that they don't unless something changed recently.
-
Caveat: The NTFS filesystem might get its panties in a bunch if it wasn't evar defragged... IMHO defrag an SSD once a year to keep your filesystem sane.
Bzzt, wrong. The natural state of an SSD is to be heavily fragmented, and if you have a defragmentation prog that attempts to defrag an SSD, you should delete it immediately. If you do it anyway, you will massively decrease your SSDs lifetime, as defragging involves A LOT of writes.
-
Now, it'd be cool if windows supported partitioning of flash storage. It's stupid that they don't unless something changed recently.
Whut? Why do you think it doesn't? I haven't seen anything on the contrary.
-
Because it doesn't, and it's ghey that windows wouldn't detect other flash drive partitions (and when was the last time anybody needed to partition a thumb drive anyway...well i would like my 16gb flash drive to have 2 partitions dammit...the needs of the few). In xp i know it does this. It used to not be this way in win98 (which will detect other flash drive partitions). It happens in such a way in xp that you know they did it on purpose because somehow ms thought that flash storage should be considered specially different.
I believe in vista and 7 wont detect other partitions on flash storage. But, when i first found this out, i only had an xp machine sitting around to try it out on. Yes disk management console will always detect the other partitions on flash storage, and it'll make you think you can assign a drive letter too. Only to restart, and find that other partitions on that flash storage didn't get assigned a drive letter, let alone detected by my computer.
EDIT: seems newer than xp windows like vista and 7 can't either by default. There is one way though that i have found that will windows support partitions on flash drives after googling. It all matters if it's removable media or not (http://www.ghacks.net/2009/04/17/partition-usb-flash-drives/). And i think it's totally bogus that you'd have to do this at all when other operating systems can support partitioned thumb drives without using that utility first in the link.
-
Oh, you're talking about removable flash storage such as USB sticks. Well, SSD's are different and partitioning should work just fine on them. Googling at least reveals no limitations in partitioning of SSD's on Windows 7. Don't know about WinXP, but why would you use obsolete OS with SSD's anyway.
-
Caveat: The NTFS filesystem might get its panties in a bunch if it wasn't evar defragged... IMHO defrag an SSD once a year to keep your filesystem sane.
Bzzt, wrong. The natural state of an SSD is to be heavily fragmented, and if you have a defragmentation prog that attempts to defrag an SSD, you should delete it immediately. If you do it anyway, you will massively decrease your SSDs lifetime, as defragging involves A LOT of writes.
My program doesn't defrag SSDs unless told otherwise. However, how much fragmentation can the NTFS filesystem take before it becomes slow / corrupt? I mean, I imagine it would be years, but I don't know... Isn't it true that for every file fragment, there must be a notation somewhere as to where it is (well, virtually for an SSD) located? So if you have thousands of fragments... I've seen drives that have been in use for 2-3 years that have 10,000+ fragments, now if that was an SSD and you add another 2, 3, 4, or more years... what kind of a hellish mess are you going to be dealing with in the Master File Table? Not to mention, fragmentation in the MFT itself, the system logs (although IIRC Windoze tries not to go so crazy on some of the logs it keeps with SSDs, or at least some manufacturers disable those options for SSD-equipped models), etc, etc... everything will be a stupendously huge spaghetti ball of mess. Granted, with an SSD, it can read any sector with equal ease.. the issue isn't the drive, the issue is the filesystem. Do I dare trust NTFS to keeps it's beans accounted for? I'm thinking of what can happen to an Outlook .PST when it gets over-sized. Apples and Oranges, I know. Same farmer selling them though.
Thoughts?
-
maybe it IS time to consider a SSD. HDD prices just SKYROCKETED, right in time for my new build. <.<
-
However, how much fragmentation can the NTFS filesystem take before it becomes slow / corrupt? I mean, I imagine it would be years, but I don't know... Isn't it true that for every file fragment, there must be a notation somewhere as to where it is (well, virtually for an SSD) located? So if you have thousands of fragments... I've seen drives that have been in use for 2-3 years that have 10,000+ fragments, now if that was an SSD and you add another 2, 3, 4, or more years... what kind of a hellish mess are you going to be dealing with in the Master File Table? Not to mention, fragmentation in the MFT itself, the system logs (although IIRC Windoze tries not to go so crazy on some of the logs it keeps with SSDs, or at least some manufacturers disable those options for SSD-equipped models), etc, etc... everything will be a stupendously huge spaghetti ball of mess. Granted, with an SSD, it can read any sector with equal ease.. the issue isn't the drive, the issue is the filesystem. Do I dare trust NTFS to keeps it's beans accounted for? I'm thinking of what can happen to an Outlook .PST when it gets over-sized. Apples and Oranges, I know. Same farmer selling them though.
Thoughts?
The only relevant MS Knowledgebase article I could find is this: http://support.microsoft.com/kb/967351
I also found random bits about there having been an issue in Windows XP pre-SP2 where too fragmented MFT broke and prevented OS from booting. This was supposedly fixed in a hotfix and Service Pack 2. Google didn't find me a KB article on this though.
If there was actually an issue in regards to fragmentation and SSD's, even with MFT, it would be known issue already. This is not the case, so there should be zero need to defragment anything on SSD's.
-
flash is random access. it doesnt care if its a convoluted mess. there is no head to move and no platter to rotate, just a few logic gates that need to be twiddled to change address. the master file table entry for any given files does not contain any file data, if anything merely the file header, file size, etc. as well as a pointer to the first allocation unit that the file is stored in. the allocation unit has a pointer to the next unit (perhaps also pointers to the previous unit and the file table entry) and so on.
fragmentation is bad because of the mechanical motions the drive must go through if you got bits of file everywhere, it has to stop reading to find the next allocation unit. its a lot of stop and go. when this happens on the flash drive its all done with transistor-transistor logic, which aside from a tiny propagation delay (were talking nanoseconds here) is instantaneous. read (or write) command goes something like this. address is written to the address bus by the controller, then the read (or write) enable pin goes active and for read i/o pins produce the value stored at the address, or for write, the i/o pins accept the new value and store it at the address.
i believe the flash chips used support block operations, where once an address is set and a read operation begins, it will read/write a byte sequentially each clock until the read/write enable pin changes logic level, without needing to key in new addresses each cycle. since the file system uses allocation units, and data stored in those units is guaranteed to be sequential, you could block read an entire allocation unit without needing to change the address. same if all your allocation units are in a row. youd need to change address if your allocation units are non-sequebntial, but this operation is several orders of magnitude faster than on a mechanical hard drive and the performance loss would be negligible.
-
My program doesn't defrag SSDs unless told otherwise. However, how much fragmentation can the NTFS filesystem take before it becomes slow / corrupt? I mean, I imagine it would be years, but I don't know... Isn't it true that for every file fragment, there must be a notation somewhere as to where it is (well, virtually for an SSD) located? So if you have thousands of fragments... I've seen drives that have been in use for 2-3 years that have 10,000+ fragments, now if that was an SSD and you add another 2, 3, 4, or more years... what kind of a hellish mess are you going to be dealing with in the Master File Table? Not to mention, fragmentation in the MFT itself, the system logs (although IIRC Windoze tries not to go so crazy on some of the logs it keeps with SSDs, or at least some manufacturers disable those options for SSD-equipped models), etc, etc... everything will be a stupendously huge spaghetti ball of mess. Granted, with an SSD, it can read any sector with equal ease.. the issue isn't the drive, the issue is the filesystem. Do I dare trust NTFS to keeps it's beans accounted for? I'm thinking of what can happen to an Outlook .PST when it gets over-sized. Apples and Oranges, I know. Same farmer selling them though.
Thoughts?
Yes. None of the issues you speak of have been reported as actual real-life problems with actual, real-life SSDs. So it sounds more like you're afraid of things that aren't an issue because developers at MS and the companies making SSDs have been thinking about them and come up with solutions before the first SSDs hit the open market.
-
i will not be using ntfs on any ssds i may get in the near future. it sounds like its not suited at all for the technology. granted i trust ntfs to hell and back on my mechanical drives, have never lost an ntfs partition, ever. but unless its is revised with ssd features in mind, its just not suited to the task.
-
But what other options are there when using Windows?
-
isn't exfat supported?
of course i figure by the time i have the cashflow to start using ssds, either an extension to ntfs will support more ssd friendly features, or a new file system will be developed. next rig i build will use mechanical drives, from then on, depends on price. could always get it as an upgrade too if they become affordable in the near future. the price is projected to drop 50% every 2 years so i figure in 5 years ssd will be the superior option for me.
-
Windows 8 has new file system, code name "Protogon". Not much is known about it so far.
-
after doing some reading on wikipedia it seems that the controllers on an ssd handle a majority of the requirements that ssds would have, so the proper file system to use would be one that reduces unnecessary disk writing. ssd drives also seem to have an internal file system to handle the flash specific details, so your file system can be rather arbitrary.
-
Oh, you're talking about removable flash storage such as USB sticks.
Yes i was. I didn't quite make the distinction in my post. I still think it's retarded that windows doesn't support partitioning of usb sticks. In reality, the only difference i can find is that one goes in a usb slot, and the other goes inside your computer (it's all flash storage anyway, regardless of high or low quality). This is why by default i don't make the distinction. It is cool however that windows 7 automatically adapts to ssd's when detected that it's installed on one (as nuke mentioned).
isn't exfat supported?
Yes there's exfat. Exfat is not a journaling file system. Sounds great for limited reads and writes compared to a journaling file system without the 4gb file size limitation.
I would also like to note, there's plenty of other file systems out there. Many designed specifically for flash storage. Microsoft however, conveniently doesn't include compatibility in windows for other file systems than it's own.
Windows 8 has new file system, code name "Protogon". Not much is known about it so far.
Did some reading. Sounds it's intended to be the default file system for the drive windows is installed on (i could be wrong).
-
designed specifically for flash storage.
could you name a few?
-
Ubifs, logfs, and yaffs2. There is jffs2. But, the first 3 i mentioned replace jffs2. These are linux specific file systems. It'd be great if they weren't so linux specific which is what i am getting after.
Fat32 was nice, but i do come around with the 4gb file size limitation more often than not. Then there's ntfs, which is great, but has journaling. You can disable journaling with ntfs (http://blog.zachsaw.com/2009/06/ntfs-journaling-file-system.html) at least.
The main hurdle i see with exfat currently, is device support since it's new. I'm talking about stuff like routers that let you plug in external storage, hd tv's with sd card readers, car audio dash mounts with usb slots, etc. Exfat support in this area will take some time.
-
yeah, I'm on Linux, using an SSD, so it perked my interest.
-
You can disable journaling with ntfs (http://blog.zachsaw.com/2009/06/ntfs-journaling-file-system.html) at least.
You linked to a guide on how to enable USN journalling on XP, which does not have it enabled by default, unlike Vista (and probably 7).
enable:
fsutil usn createjournal m=1000 a=100 C:
check status:
fsutil usn queryjournal C:
EDIT:
disable:
fsutil usn deletejournal /d C:
-
Thank you. Quite frankly i didn't know you could disable journaling in ntfs until i googled it while making that part of the post (where it said how do i enable, i thought it said disable in a quick glance of the article i linked). It's still cool to know, i thought it was one of the many thing's you couldn't do with windows.
While those are great flash file systems bobbau. When i did have an ssd. I routinely just used ext4 with the noatime and nodiratime fstab options for that drive to disable any kind of journaling and lessen reads and writes. But, definitely look for any fstab options for the flash file systems i listed if you're still interested in them.
-
To disable (just Googled it): run the following with Admin privileges:
fsutil usn deletejournal /d C:
-
isn't exfat supported?
Nope. Vista and later (which includes Server 2008 and later) must be installed on a file system that supports hard links (for winSxS (Windows Side By Side) among other things). I also believe that the root file system that the windows directory is installed on has to have transaction support and have the appropriate hooks for Volume Shadow Services.