flat assembler
Message board for the users of flat assembler.

Index > Heap > SSD lifetime

Goto page 1, 2  Next
Author
Thread Post new topic Reply to topic
b1528932



Joined: 21 May 2010
Posts: 287
b1528932
Is it really worth buying laptop with SSD?
SSDs have 10k write cycles per unit, wich is horrible. Even with dodays drives, assuming write takes 1ms, it takes 10 seconds to use this limit.
And SSDs having far shorter write times would propably fall bellow 1ms..

So is it worth buying SSD, and then having 1 virus wriiting to sector 1 many thousand times effect8ivly destroying your disk by making bootsector read-only?

Does SSDs have some HARDWARE based mechanisms to randomize writes?

For example, i write 100 butes at 0. Will my drive copy 512bytes to temporary buffer, overwrite first 100b with my data, then randomize location, and write 512 data linking virtual sector 1 to specyfied address?
Or will my write go through to the same location over and over again?

I belive SSD will suck untill someone make them more durable than mechanican HDD...
Post 28 Jan 2011, 16:45
View user's profile Send private message Reply with quote
Coty



Joined: 17 May 2010
Posts: 546
Location: ␀
Coty
Quote:
Does SSDs have some HARDWARE based mechanisms to randomize writes?

Wouldn't SSDs stack data lineally for faster acsess instead of random? You think that is how it would work since 'faster' is part of there goal...

As far as I have seen some SSDs have an extra 4-8MB flash for when sectors begin to ware it will re-alocate the sector to the backup flash and warn that disk is failing. I have also seen auto defragment drives that defrag everytime file information is changed (My uncle's clames to do this, he thought it was supper awesome, but I think its a waste and a way to tear up the SSD)

Quote:
I belive SSD will suck untill someone make them more durable than mechanican HDD...

I can use them in a magnetic room without worry of my HDD being wiped on the spot, if it gets bumped it is less likely suffer HDD failuer... I am spectacle with there writing ware and tare though Confused
Post 28 Jan 2011, 17:00
View user's profile Send private message Send e-mail Visit poster's website AIM Address Reply with quote
revolution
When all else fails, read the source


Joined: 24 Aug 2004
Posts: 17344
Location: In your JS exploiting you and your system
revolution
SSDs have wear-levelling to even out the wear. Most also have internal RAM to buffer data before writing. Both of these mechanisms increase the expected lifetime of the SSD. But check before you buy if it has both of these things.

SSDs are much more expensive per byte than HDDs, but their performance is much better.

Choose any two: Fast - cheap - long life
Post 28 Jan 2011, 17:28
View user's profile Send private message Visit poster's website Reply with quote
edfed



Joined: 20 Feb 2006
Posts: 4240
Location: 2018
edfed
the only real problem with SSD is not in the device itself, but in the OS.

for example, my netbook with ssd and linux on it spend a lot of time, just to do some uneeded paging file management, save configuration each shutdown.

what is incredible with theses OSes is the systematic use of paging with swap.even on a PC with 1GB RAM, why the fuck do i need more than 1GB ram for fasm coding? video watching, and whatever things made on win98 with less than 512M of RAM + SWAP?

just explain why the fuck linux should update each week? doesn't it works fine?
i don't see any reason to always update some OS when they stull work good.

but, be quiet, end of world is soon, then, don't worry about the really bad applications made by companies with an unbalanced technology, develloped with only official science theories.
Post 28 Jan 2011, 17:52
View user's profile Send private message Visit poster's website Reply with quote
revolution
When all else fails, read the source


Joined: 24 Aug 2004
Posts: 17344
Location: In your JS exploiting you and your system
revolution
edfed: I use WinXPSP2 with 512MB RAM and no swap file.

Also, if you are concerned about lifetime then I would suggest that when using SSDs to move swap files to another drive, or disable the swap file. Swap files, even though recommended by MS, are bad for SSD wear. MS recommend to put swap files on SSDs because they are thinking about the performance, not the lifetime.
Post 28 Jan 2011, 17:59
View user's profile Send private message Visit poster's website Reply with quote
b1528932



Joined: 21 May 2010
Posts: 287
b1528932
Im not concerned about swap, because it can be shut off as i did.

im concerned if i use 'dd of=/dev/xxx if=/dev/zero bs=512 count=1'
a milion tmies will it destroy sector 1.
Post 28 Jan 2011, 18:04
View user's profile Send private message Reply with quote
revolution
When all else fails, read the source


Joined: 24 Aug 2004
Posts: 17344
Location: In your JS exploiting you and your system
revolution
b1528932 wrote:
im concerned if i use 'dd of=/dev/xxx if=/dev/zero bs=512 count=1'
a milion tmies will it destroy sector 1.
No. Wear levelling will keep switching to new sectors at the hardware level. The only thing you would notice is slower writes as blocks are erased and swapped around.

Plus check for drives with RAM buffers, this is also very good for reducing writes to FLASH.
Post 28 Jan 2011, 18:11
View user's profile Send private message Visit poster's website Reply with quote
Artlav



Joined: 23 Dec 2004
Posts: 188
Location: Moscow, Russia
Artlav
I have two SSDs on my main rig for over two years already, no signs of wear or data errors.

Estimations are good and things, but we won't know for sure until first commercial SSDs start to fail due to wear.
The things are still too new, i think.

I haven't heard of anyone having problems due to SSD wear so far.

I also wonder at which point are these wear levelling systems are?
The chips have address and data lines, with commands dealing with bytes and sectors. So, an embedded system should handle the wear levelling at OS level, right?
Thus, formatting the chip into swap and ext3 partitions would destroy it rapidly?

And, SSDs provide the wear levelling of their own in between the chips and the SATA/whatever?

b1528932 wrote:
SSDs have 10k write cycles per unit, wich is horrible. Even with dodays drives, assuming write takes 1ms, it takes 10 seconds to use this limit.
Where did you get that?
Typical figures i see are 1000000 cycles and 20 years data retention.
Post 28 Jan 2011, 18:47
View user's profile Send private message Visit poster's website Reply with quote
revolution
When all else fails, read the source


Joined: 24 Aug 2004
Posts: 17344
Location: In your JS exploiting you and your system
revolution
The SSD controller does the wear levelling, not the OS. There is no fixed translation between the interface sector 1 and the hardware sector 1.

Write endurance figures are highly dependant upon the underlying FLASH chips uses. I don't know of any current SSDs using MLC, all SSDs I have ever seen use SLC with expected endurance of 100k+ cycles.

Keep in mind that endurance figures are merely indicative, not absolute. And failures mechanisms are not always fatal. Often it just takes longer and longer to erase and reprogram bits as they wear until finally the bit cannot be flipped. Smart SSD controllers monitor these times and swap with fresh sectors before data gets lost. But eventually all blocks will wear if used enough. The wearing is only postponed by the use of levelling algorithms and switching with spare sectors. ECC is also used but generally is not very strong, just enough to correct one or two bytes per sector.

SSDs have spare blocks for when other blocks get too worn to be useful. But they are usually in limited supply.
Post 28 Jan 2011, 19:07
View user's profile Send private message Visit poster's website Reply with quote
b1528932



Joined: 21 May 2010
Posts: 287
b1528932
Do you know estimate nimber of write cycles until normal mech HDD fail?
You say SSD is like 100k before performance degradation. Ok, how about classic drive?
If its like 10x or 100x even 1000x more, its fine, SSD is worth it. But if mech can endure billions of write and still work as new, i doubt theres any reason for switching to SSD.
If you have heavy load server wich must distribute data i would suggest making volatile storage, working exactly as RAM, but being access as external device. All data has to be copy into it on poweron (wich in case of server, doesnt happen very often).

Im not considering here 'normal' usage, even swaping memory, but extreme situation like malware designed to destroy SSD.
Post 28 Jan 2011, 21:12
View user's profile Send private message Reply with quote
Artlav



Joined: 23 Dec 2004
Posts: 188
Location: Moscow, Russia
Artlav
Reasons?
How about every program staring in under a second?

Nerve cells are worth more than an SSD, and nothing is more stressful than a computer that works slower than your thoughts.
Post 28 Jan 2011, 21:39
View user's profile Send private message Visit poster's website Reply with quote
revolution
When all else fails, read the source


Joined: 24 Aug 2004
Posts: 17344
Location: In your JS exploiting you and your system
revolution
Magnetic HDDs have no known wear problem. In the magnetic domain you can write forever and still get good data. Of course in the mechanical domain the drive will physically degrade and die. Lifetimes of an HDD is dependant upon uptime, spindle wear, bearing wear, etc. Basically all mechanical failure modes due to ageing and thermal cycling.
Post 28 Jan 2011, 22:03
View user's profile Send private message Visit poster's website Reply with quote
b1528932



Joined: 21 May 2010
Posts: 287
b1528932
So today basicly SSD is too new to. In 10 years or so, mech HDDs will eventually share fate of floppies, but untill then - SSD will continue to improve, and people will use HDDs. Its my point of view, not worth investing in SSD right now, its better to wait couple of years untill twechnology will be better and more reliable.

When SSD degradation will be less than HDD, HDDs will die. Not untill then.
Post 28 Jan 2011, 23:11
View user's profile Send private message Reply with quote
JohnFound



Joined: 16 Jun 2003
Posts: 3502
Location: Bulgaria
JohnFound
SSD have only one big advantage for the user - low power consumption.
And one really big advantage for the industry - short live expectation.
You know, the industry don't have any use for everlasting hardware. Very Happy
Post 29 Jan 2011, 07:45
View user's profile Send private message Visit poster's website ICQ Number Reply with quote
b1528932



Joined: 21 May 2010
Posts: 287
b1528932
Quote:
You know, the industry don't have any use for everlasting hardware. Very Happy

True unfortunatly, but imagine a company that build its reputation based on hardware quality. When thousand persons will recommend something, tests will confirm it, i bet it will pay off for that company. How many companies make SSD? And what are the odds user choose to buy again from same company when their disk fails? He will get angry, flame on internet company that caused him to lose money, and buy from other one. As i would do.
Post 29 Jan 2011, 20:03
View user's profile Send private message Reply with quote
JohnFound



Joined: 16 Jun 2003
Posts: 3502
Location: Bulgaria
JohnFound
Well, it is well known and advertised fact that the flash memories "wears" with time.
None of the users will complain about something that is declared by the manufacturer.
Additionally, the modern, well tamed user, feels himself very happy that he should buy new peace of hardware again and again. Twisted Evil

IMHO, the things could get better if there was OSes (and application software) optimized for SSD - I mean with little or none use of swap files and decent RAM cache.
For now the common work around is to use external flash memory for swap files and to change it when it wears fully.

Paradoxical, the flash memories are very good for swap files, because of fast random access time without latency.
Post 29 Jan 2011, 21:08
View user's profile Send private message Visit poster's website ICQ Number Reply with quote
edfed



Joined: 20 Feb 2006
Posts: 4240
Location: 2018
edfed
remember that commercial mainstream software and hardware are develloped not by geeks, but busynessmens.

the case of fasm is very rare, and realativelly usefull.

i don't know anybody that needs to code asm. only there we are able to communicate, and we are not a lot.

then, our ideas are only ideas, and nothing will be made from here.

we just need a mecene, somebody who loves the good things, well made, and then, ideas will become true.
Post 30 Jan 2011, 02:15
View user's profile Send private message Visit poster's website Reply with quote
Madis731



Joined: 25 Sep 2003
Posts: 2140
Location: Estonia
Madis731
First, we wouldn't be having this talk if people would read some materials before posting. Anandtech, tomshardware, other sites - they all talk about wear leveling when testing SSDs. More accurately http://en.wikipedia.org/wiki/Write_amplification
Just one example of many: http://www.anandtech.com/show/3690
They have diagrams and what-not.

Secondly, my opinion about the SSDs:
Everybody who has them, praises them. Compilation/assembly goes faster, programs start up faster, OS starts faster. Everything is just quick.
The first generation Intel SSD (80GB I think) I tested, could run 5 HD movies from SSD. The promised 300/600GB SSDs from Intel (AFAIK they should be called G3) will last even longer because of the sheer size. Even writing a pessimistic 5000 times to each cell would guarantee 1.5-3 PB (petabytes) of consecutive writing (@ byte 0 if you like).

Here's a list why one would buy an SSD:
1) Seek speed (tens of microseconds opposed to multiple milliseconds)
2) read/write speed (250MB compared to 70MB)
3) Guaranteed life* - you know what you're paying for
*with wear-leveling SSDs can write a PB (petabyte) or more before failing. This means that even a 100GB-per-day-torrent-leaching-machine will last YEARS.
3.1) HDDs can die anywhere from arrival to many years. HDDs wear even if you don't read/write, but SSDs wear only by writing meaning you can still read.
3.2) When HDDs fail you might lose all your data. SSD failure will usually mean you can only read your data (though that is not 100% accurate i.e. controller failure).
Post 02 Feb 2011, 18:58
View user's profile Send private message Visit poster's website Yahoo Messenger MSN Messenger Reply with quote
JohnFound



Joined: 16 Jun 2003
Posts: 3502
Location: Bulgaria
JohnFound
Quote:
3.1) HDDs can die anywhere from arrival to many years. HDDs wear even if you don't read/write, but SSDs wear only by writing meaning you can still read.


It is not true. HDD are (potentially) immortal mechanically. In the latest models there is an air bearings on the main shaft and the heads flies above the disk. So, the only reason a disk can die is because of accidental reasons - mechanical break, failed electronics, etc.
And these are the same reasons SSD can fail prematurely, besides of the normal "guaranteed" life.
After the most of the industry began to use lead-free solder, the reliability of the electronics markedly diminished. Now the main reason of failures is the electronic itself.
Post 02 Feb 2011, 21:19
View user's profile Send private message Visit poster's website ICQ Number Reply with quote
asmhack



Joined: 01 Feb 2008
Posts: 431
asmhack
The advantage of ssd (beyond the speed) is the fact that it is solid (contains no mechanical parts) so that's a good reason to prefare it on a netbook for example for it's anticrash protection. (power consumption is still the same though).

ps: rip disks
Post 02 Feb 2011, 23:15
View user's profile Send private message Reply with quote
Display posts from previous:
Post new topic Reply to topic

Jump to:  
Goto page 1, 2  Next

< Last Thread | Next Thread >
Forum Rules:
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
You can attach files in this forum
You can download files in this forum


Copyright © 1999-2020, Tomasz Grysztar. Also on YouTube, Twitter.

Website powered by rwasa.