M
Mon May 12 00:43:12 PDT 2008
"M.I.5¾" <no.one@no.where.NO_SPAM.co.uk> wrote in message
news:4827ecc3$1_1@glkas0286.greenlnk.net...
>
> "Bill in Co." <not_really_here@earthlink.net> wrote in message
> news:%23M7umMhsIHA.1768@TK2MSFTNGP03.phx.gbl...
>> M.I.5¾ wrote:
>>> "Bill in Co." <not_really_here@earthlink.net> wrote in message
>>> news:OIyM5SOsIHA.1236@TK2MSFTNGP02.phx.gbl...
>>>> Updated below...
>>>>
>>>> Bill in Co. wrote:
>>>>> M.I.5¾ wrote:
>>>>>> "Michael Lanier" <jamaica@charter.net> wrote in message
>>>>>> news:0a85f12c8e9f42d79f10e0b727be1849@newspe.com...
>>>>>>> XP allows me to defragment drives other than my hard drive. I am
>>>>>>> unable
>>>>>>> to find that capability on my Vista. Is it available and how would
>>>>>>> I
>>>>>>> access it.
>>>>>>> Thanks.
>>>>>>>
>>>>>>> url:
http://www.ureader.com/gp/1279-1.aspx
>>>>>>
>>>>>> Under any operating system, I would be reluctant to defragment any
>>>>>> FLASH
>>>>>> memory drive for 2 reasons.
>>>>>>
>>>>>> 1. The multiple file fragment shifts that are required during any
>>>>>> defragment
>>>>>> operation are going to severely tax the limited life of the FLASH
>>>>>> memory
>>>>>> blocks (limited to an *average* of 1000 erase/write cycles).
>>>>>
>>>>> Is that documented somewhere? Or does it vary all over the place?
>>>>
>>>> Actually, I just spent a bit of time looking this lifetime thing up
>>>> now,
>>>> and I'm seeing articles mentioning 10,000 - 100,000 erase/write cycles,
>>>> not 1000. So while it's true that defragging will tap into this a
>>>> bit,
>>>> occasionally doing it probably wouldn't be all that harmful (although
>>>> there isn't much point).
>>>
>>> There is confusion because there are 2 lives associated with FLASH
>>> memory
>>> cells. The regular memory blocks (the ones that actually hold the data)
>>> are
>>> lifed for an average of 1000 erase/write cycles.
>>
>> Do you have a site for this low figure today? I'm not finding it.
>>
>>> These blocks are utilised
>>> on a rotating basis to ensure that the least recently erased and written
>>> block is the next one to be erased and written (this is called 'wear
>>> levelling' and is designed to extend the life as far as possible).
>>
>> Right. I've read that. But again, from the user's point of view,
>> I don't think it comes down to 1000 erase/write cycles. (The figures
>> I've been seeing are 10 to 100 times that, at least as I recall, even for
>> this - 10,000 - 100,000).
>>
>>> However,
>>> every FLASH memory device also has a block that holds the houskeeping
>>> data
>>> such as the FATs and in which block the data is actually held. This
>>> block
>>> has to have a much greater life because it has to be completely erased
>>> and
>>> rewritten every time the memory is altered. It is this block that is
>>> lifed
>>> for 100,000 erase write cycles.
>>
>> Or was, in the past?
>>
>>> The architecture of FLASH memory cells is
>>> designed to keep the cost down as much as possible, and it is a fact of
>>> these devices that if one block fails (be it data or housekeeping), the
>>> whole
>>> memory device is rendered useless.
>>
>> But I think they have some recovery algorithms built into the UFD to
>> handle some of these potential contingencies.
>>
>>> Fortunately it is only the erase/write that is life limited. If the
>>> memory
>>> can be written and verified, then the read life is, thankfully,
>>> relatively
>>> infinite. That makes this type of memory ideal for holding device
>>> firmware
>>> that may be updated relatively infrequently, but its use in USB thumb
>>> drives, though not ideal, is considered acceptable because the low cost
>>> of
>>> the memory means that the drives can be cheaply replaced when a failure
>>> does
>>> occur. I have got through several.
>>>
>>> If you have ever watched the graphic representation of file fragment
>>> movements that occur when a drive is defragmented, you will appreciate
>>> that
>>> a FLASH drive will get one hell of workout as the defragmentation
>>> proceeds.
>>
>
> We use these memories to hold firmware in avionic equipment. The life
> data is freely available in the manufacturer's reliability data. Airborne
> approved devices don't have any longer life than the more normal FLASH
> memory devices. The life doesn't cause any problem on normal service
> equipment because we have to verify the load after writing. However, on
> development equipment the FLASH device can often fail with the frequent
> rewrites that occur (often the whole device as firmware changes).
>
> I was going to link to some data or information, but the internet is down
> here at present.
>
Just looked it up in a paper catalogue and the wording from NEC is rather
more vague. They guarantee the life of block 0 (the houskeeping block) for
100,000 erase write cycles. However, they specifically state that they make
no guarantee as to the life of any other blocks but that they have a typical
average life of 1000 erase/write cycles.