| |
| Author: |
|
Zydor Elite rating: Competent
Joined: 01 Nov 2010 Total posts: 328 Location: Hampshire Gender: Male Posts per day: 0.34
|
Posted: Mon Aug 20, 2012 12:55 am Post subject: Optimising SSD Drives |
| |
A pointer for anyone who has or is about to convert to a SSD drive. The link below is highly recommended, I checked it out with come of my Technician Department collegues, and its good info highly recommended. It will get your SSD flying as it should be, and minimise the number of cache writes - the latter being key to prolonguing SSD life.
The list takes 5-10 mins, very easy to do, just go through it slowly and methodicaly and you'll be fine.
Dont forget if you have more than one SSD drive to apply the fixes to each drive.
Highly recommended if you have SSD Drive(s), its essential stuff for performance and drive longevity - the latter being particularly important as they have a theoretical life of 10,000hrs and anything that can be done to reduced uneeded disc writes is a (very) good thing.
http://www.auslogics.com/en/articles/ssd-tweaks-to-increase-ssd-performance/
Regards
Zy
|
|
| Back to top |
|
| Author: |
|
UBT - Founder Site Admin
Joined: 13 Mar 2006 Total posts: 3114 Posts per day: 1.17
|
Posted: Mon Aug 20, 2012 5:54 pm Post subject: |
| |
Hi Zydor,
That's really useful info - the main thrust seems be that for a SSD, don't allow the OS to make *any* writes to the device, unless you are actually saving some data......
And with only 10,000 writes "per cell" (before the cell itself becomes damaged/useless) so minimising the number of writes of either temporary files, indexing files or just defragging files, won't do you any good, as it'll shorten the useful life of the SSD.
One assumes then that for BOINC Manager, you should switch off "Tasks checkpoint to disk" as that will also be writing additional files to the SSD, and which isn't really necessary....
regards
Tim
_________________ http://www.cambspsa.org.uk/ - Website of my local Prostate Cancer Support charity.
Last edited by UBT - Founder on Thu Aug 30, 2012 12:25 pm; edited 2 times in total |
|
| Back to top |
|
| Author: |
|
Zydor Elite rating: Competent
Joined: 01 Nov 2010 Total posts: 328 Location: Hampshire Gender: Male Posts per day: 0.34
|
Posted: Tue Aug 21, 2012 10:36 am Post subject: |
| |
The thing to bare in mind with SSD drives is we have all become brainwashed over the decades with our hard disk drive activities. SSDs rearly are a totally different beast and individuals need to move out of their decades long "hard disk comfort zone". SSDs are different, and need to be treated that way.
The key difference, although obvious on the face of it - its what dictates the different treatment of the drives - is that there is no moving parts. The data therefore moves around at electronic speeds, not limited by mechanical moving parts. All the mechanisms and tricks and traps of using traditional hard disks are mostly mute - some can even be detrimental to SSDs by shortening life (eg regular defragging - its not needed [at all] on SSDs, and can be detrimental by ramping up the disc access totals and shortening drive life).
The key to it all is for a user to bare in mind that everything inside the SSD drive happens (as such) at the speed of light - electronics through and through. All the old tricks of hard disk useage, defragging, blah blah blah, do NOT apply as there is no mechanical drive arm limiting access speed etc. Fragmentation is a non issue as the search for the next needed byte is at the (essentially) speed of light, therefore you dont give a Rats As* about fragmented files. The latter is the key to the rationale of their useage and how to look after them.
The current accepted "norm" for SSDs of a 10,000 Hour drive life is in reality only an educated guess tilted on the side of caution. The long term reality will not be less, and vertually guaranteed to be much much longer - no one really knows for sure, and wont for a few years until the first start to fail. However, its already clear that their life will easily go beyond the accepted life of PC, so there is not an issue as long as they are treated as solid state drives, not a "super fast hard drive".
The latter phrase means take the time to do a one time effort of tuning up the drive and change traditional computing habits to meet the new capabilities. The actions in the first post all tune the drive's useage to take advantage of solid state speeds, and reduce to a minimum the access to the drive.
What should be avoided like the plague is an over reaction and live in fear of drive accesses. The latter really are NOT a problem for practical life span as long as the drive is tuned for use as a solid state drive, and not a mechanical Arm driven drive. Once the changes are made, just - quite litterally - forget about the drive, its maintenance free, leave it alone ...... dont fiddle with it rofl If you never touched it again after tuning it as in post one, until the day it died of old age, even that would be too soon
The biggest change (once the tweeks above are done) is in the human mind set. We are all so used to maintaining hard drives by fiddling around with them, that there is a real urge to "fiddle" with them to make them run faster - arrrrrrgh! - dont
Bottom line ..... once the tweeks are done in post one - leave the damn thing alone, it dont need you - you are now redundent - live with the latter and find something else to fiddle with rofl
Regards
Zy
Last edited by Zydor on Sun Aug 26, 2012 12:47 am; edited 2 times in total |
|
| Back to top |
|
| Author: |
|
Jeffers Elite rating: Above Average
Joined: 24 Jul 2006 Total posts: 220 Location: Halifax, UK Gender: Male Posts per day: 0.09
|
|
| Back to top |
|
| Author: |
|
Jeffers Elite rating: Above Average
Joined: 24 Jul 2006 Total posts: 220 Location: Halifax, UK Gender: Male Posts per day: 0.09
|
Posted: Fri Aug 24, 2012 11:42 am Post subject: |
| |
| UBT - Founder wrote: |
One assumes then that for BOINC Manager, you should switch off "Tasks checkpoint to disk" as that will also be writing additional files to the SSD, and which isn't really necessary....
regards
Tim |
Where do I find that option? I've had a trawl round and can't spot it.
_________________ Regards,
Jeff
|
|
| Back to top |
|
| Author: |
|
UBT - Founder Site Admin
Joined: 13 Mar 2006 Total posts: 3114 Posts per day: 1.17
|
Posted: Fri Aug 24, 2012 12:02 pm Post subject: |
| |
| Jeffers wrote: | | Where do I find that option? I've had a trawl round and can't spot it. |
Try Tools > Computing preferences... > disk and memory usage "tab"
> then the 4th item down the list is the one you want and change it to 0 (which I assumes means it doesn't then checkpoint stuff, as it's switched "off"
You could also try changing the value to something crazy, like 100,000 (this is about 27.77 hours - which is long enough for most WU's not to checkpoint), which BOINC Manager seems to accept....though whether it takes any notice is another thing!
see here: http://boinc.berkeley.edu/wiki/Local_preferences
where they say
Tasks checkpoint to disk at most every X seconds: A suggested interval between disk accesses. Useful on laptops where the disk may be spun down for long periods. Default is 60 seconds, maximum 999 seconds. The project application does not have to follow this suggestion!
regards,
Tim
_________________ http://www.cambspsa.org.uk/ - Website of my local Prostate Cancer Support charity.
Last edited by UBT - Founder on Thu Aug 30, 2012 12:25 pm; edited 1 time in total |
|
| Back to top |
|
| Author: |
|
Jeffers Elite rating: Above Average
Joined: 24 Jul 2006 Total posts: 220 Location: Halifax, UK Gender: Male Posts per day: 0.09
|
Posted: Fri Aug 24, 2012 1:55 pm Post subject: |
| |
Cheers Tim, I'd seen that one and wondered if that might be the one to play with. As you say, it seems logical that setting it to zero would disable it, I'll give that a try, although I'm not sure how to check if it's having the desired effect?
_________________ Regards,
Jeff
|
|
| Back to top |
|
| Author: |
|
Jeffers Elite rating: Above Average
Joined: 24 Jul 2006 Total posts: 220 Location: Halifax, UK Gender: Male Posts per day: 0.09
|
Posted: Fri Aug 24, 2012 2:17 pm Post subject: SSD terminology |
| |
I do think that some of the confusion that arises from SSDs is the terminology, as there isn't a disk anywhere in their assembly. I know that historically they were known as Solid State Disks because, certainly in the mainframe environment where I worked, they emulated traditional spinning drives. The operating system didn't know the difference, they just looked like conventional disk volumes at the end of the I/O channel.
Even nowadays, when most disk cabinets contain dozens, if not hundreds, of individual small disks (in size not capacity!), the operating system (at least in the IBM world) still sees them as strings of old-style big platter disks.
Perhaps a better name for them would be SSS, or Solid State Storage, but I think that the SSD label has gotten too entrenched to be changed now, and confusion will continue.....
_________________ Regards,
Jeff
|
|
| Back to top |
|
| Author: |
|
Zydor Elite rating: Competent
Joined: 01 Nov 2010 Total posts: 328 Location: Hampshire Gender: Male Posts per day: 0.34
|
Posted: Sat Aug 25, 2012 6:11 pm Post subject: |
| |
The terminology is "Solid State Drive"
Regards
Zy
|
|
| Back to top |
|
| Author: |
|
Jeffers Elite rating: Above Average
Joined: 24 Jul 2006 Total posts: 220 Location: Halifax, UK Gender: Male Posts per day: 0.09
|
Posted: Sun Aug 26, 2012 11:10 am Post subject: |
| |
| Zydor wrote: | The terminology is "Solid State Drive"
Regards
Zy |
Maybe that's true now, but when they were first introduced where I worked in the mainframe environment, they were definitely Solid State Disks.... we're talking about 20-odd years ago here!
_________________ Regards,
Jeff
|
|
| Back to top |
|
| Author: |
|
Zydor Elite rating: Competent
Joined: 01 Nov 2010 Total posts: 328 Location: Hampshire Gender: Male Posts per day: 0.34
|
Posted: Thu Aug 30, 2012 7:46 am Post subject: |
| |
One aspect of Diskeeper I avoided on this thread initially was the treatment of SSD drives, until I had more solid fact to go on. Whilst I have used Diskeeper since its inception back in the Dark Ages (!), the advent of solid disk drives and the latest Diskeeper version warrented a pause until I had solid data, as the change both to Solid State Drives and the latest version of Diskeeper creates a whole new ball park.
Its remarkable what they have achieved. I installed the latest version of Diskeeper on 18 Aug, halfway through a three week holiday break. I had installed SSDs at the start of August. For the last three weeks the stats have been building up, and its remarkable, and very very significant for SSDs. Bare in mind that the overall "enemy" of SSDs is disk accesses, and whilst an overeaction to "any" disk access should be avoided - they are after all built for that (!) - nontheless, their inherent relative disk life compared to traditional Hard Disks warrents attention and habit changes to prolong SSD life. Stats for the last 24 hours on my 3960 Box (running the PC on BOINC - traditionally heavy on disk accesses as such - for about 70% of the 24 hour period):
Disk Accesses Prevented by Intelliwrite: 46,291
Disk Access Eliminated by Instant Defrag (defrag "on the fly" before it hits the disk) 23
Disk Accesses Eliminated by Traditional Defragmentation: 82
Cumulative Disk IO Acesses Saved (therefore speeding up PC reponse times) 7,171,971
That is remarkable ...... and very very good news for SSDs whose great enemy is disk accesses. To have 46,291 disk accesses stopped before they got any where near the disk, 23 stopped "at the front door", and 82 dealt with "after they sat down", is stunning
Overall it has got to the stage with Diskeeper that they no loner produce pretty Fragmentation graphs blinking away during Defrag, there is no point - there are so little fragmented On-Disk packets to deal with rofl
With SSDs the trick is to turn off On-Disk defrag as its pointless (dont defrag on-disk with SSDs - baaaaad idea, it hits overall life span considerably) and leave the Diskeeper IntelliWrite facility turned on. The latter stops Fragmented packets hitting the disk in the first place, all good news for SSDs with finite disk access life, and equally good for non-SSDs response times. Even traditional Hard Disks, there is vertually zero defragging is needed due to the IntelliWrite facility stopping fragmented packets hitting the disk at all.
The point to bare in mind, is that it also means that Diskeeper can handle a mixed Hard Disk / SSD environment, optimising its activities for the type of disc it comes across. So those with (say) 1xSSD and 1xHard Disk drive as their main drives get the best of all worlds and able to tune the Defrag activity to their pecise needs - mixing techniques to suit SSD or Hard Disk.
Cost is peanuts frankly, its a one off fee, not annual. Dont bother with the "Home" addition, get the Pro version - its £41.11, and that covers up to three PCs. So at a cost of about £14 a PC lasting for 18-24 months before a new version comes along, its a No-Brainer.
I was a great fan of Diskeeper before SSDs, I am an even greater fan now .... a remarkable collection of software - run, dont walk, and get it for a - fast - hassle free SSD life
http://www.condusiv.com/
Regards
Zy
|
|
| Back to top |
|
| Author: |
|
UBT - PiezPiedPy Elite rating: Dangerous
Joined: 23 Mar 2006 Total posts: 566 Location: Liverpool - UK Age: 42 Gender: Male Posts per day: 0.21
|
Posted: Sun Sep 30, 2012 1:05 pm Post subject: |
| |
I was planning on gettting an SSD to replace my OS drive but dont think i'll bother now, since they only have 10,000 hrs use i'm going to assume its size will decrease after about 1.5yrs
_________________
|
|
| Back to top |
|
| Author: |
|
Zydor Elite rating: Competent
Joined: 01 Nov 2010 Total posts: 328 Location: Hampshire Gender: Male Posts per day: 0.34
|
Posted: Sun Sep 30, 2012 8:52 pm Post subject: |
| |
Dont hesitate - go for it. The calculation is far more complex than a simple figure of 10,000.
A vary good analysis is at: http://forums.anandtech.com/showpost.php?s=1210cd60c12370157ca84c62e35a611d&p=27852736&postcount=11
The calculation is hugely complex, however a common mistake made by many is to assume 10,000 hrs or 10,000 writes (pick your figure frankly) per drive. Thats no the case, the high level figure refers to individual cells as if they were accessed 7x24 - which of course is way way from reality. Even if it was 10,000 hrs for the whole drive (based on 7x24) that would still give a drive life of 6-8 years, as hard drives are not continualy accessed 7x24. They are on average in use around 8 hours a day, and on top of that are not continuously written - overwhelmingly most cells are read not written to. The drive software will level out cell useage to prolong life, just leave it to get on with it and dont mess it about with traditional "disc optimisers" - the latter are redundent with SSD drives.
The latter is way from the whole story as the 10,000 "figure" refers to each individual cell, not all cells are written to at the same time. Bare in mind that the example in the reference above was laid out over three years ago. There have yet to be reported instances of solid state drives failing in large numbers, and they have been around for over 5 years now.
You will run out of space on the drives before they fail. A little like the 250Gb drives that seemed so huge a few years back, but now are sidelined by Multi-Tb drives as the needs for multi-media kick in. At present solid state drives are perfect for mainstream C: and D: drives, then have a traditional Hard Drive as your back-up drive, and data drive. Or if drive bays are limited, go for C: as solid state, and D: as muilti Tb hard drive backup.
I would not use solid state drives as long term backup storage - use traditional hard drives for that. But the use of 256/512Gb solid state drives as the main C: drive (and D: drive) is now a no brainer in technology terms, cost is the only limiter. They have been around long enough to prove the reliability of the technoloy - just ensure that backups are done to a traditional drive for now.
Everyone is still waiting for main SSD failures from the first wave deployments circa 5 years ago - not happened yet
Dont hesitate - go for it
Regards
Zy
|
|
| Back to top |
|
| Author: |
|
UBT - PiezPiedPy Elite rating: Dangerous
Joined: 23 Mar 2006 Total posts: 566 Location: Liverpool - UK Age: 42 Gender: Male Posts per day: 0.21
|
Posted: Mon Oct 01, 2012 11:09 pm Post subject: |
| |
Nice1 Zydor, makes sence now.
I'll be getting myself one in the near future
All my data is on seperate drives I hate my work mixed in with the OS, its a pain when reinstalling or updating, so I'll only be using the SSD for the OS.
_________________
|
|
| Back to top |
|
| Author: |
|
Zydor Elite rating: Competent
Joined: 01 Nov 2010 Total posts: 328 Location: Hampshire Gender: Male Posts per day: 0.34
|
Posted: Tue Oct 02, 2012 1:15 am Post subject: |
| |
In that case get 120Gb SSD minimum as the main C: drive. If you look at the reference I gave and the calculations, he based it around a 60Gb drive (which was the common size then). By taking on a 120Gb C: Drive, you give the internal drive software more room to breath and less chance of repeated writes to the same cell. Basicly by doubling the size from 60Gb to 120Gb, the drive life is quadrupled in a broad sense not merely double for the same size of data set as the drive software has more cells it can play with.
A 120Gb drive as the OS drive will last until doomsday frankly for the average home user, and way over 10 years even for BOINC crunchers like us, especially if care is taken to do the actual BOINC crunching based on the Traditional Disc Drive.
If an OS fills a 120Gb drive at any time in the next 10 uears we are all in trouble, and SSD capacity will be the least of our worries
Regards
Zy
|
|
| Back to top |
|
| Choose Display Order |
|
| User Permissions |
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|
 |