[Beowulf] Re: ECC support on motherboards?

[Jim Lux]

At 08:29 AM 5/21/2008, Robert G. Brown wrote:
>On Wed, 21 May 2008, Eugen Leitl wrote:
>
>>On Wed, May 21, 2008 at 03:36:24PM +0100, Kozin, I (Igor) wrote:
>>>
>>>>b) Laptops.  I'd REALLY like a flash-driven laptop.
>>>
>>>Who wouldn't? But what's the typical lifespan of flash memory?
>>
>>Current SSDs claim >1 Mh MTBF, largely due to wear-leveling.
>
>A Mh is, lessee, 100 years?  So one failure per hundred systems per
>year?  Or if it is truly due to wear, perhaps it is zero failures per
>hundred systems for 50 years, then an increasing failure rate out to
>100+?
>
>If it is wear, however, lifetimes may depend on usage, and using it as a
>root fs and/or home directory may significantly shorten lifetime.




Actually, not a big deal. The wearout is with erases/writes, not 
reads.  What they do is not use the same physical location for a 
given block.  That is, when you read/change/write a block back, it 
gets written to a different location.   There's a systematic way to 
keep all this straight, but the net result is that writes are evenly 
distributed across the device.

Clever cacheing with delayed writes is another aspect of this.  By 
the way, erases have to be done in blocks, but writes can be done in 
little bits, so a journalling filesystem is a typical way to do 
things with flash. (See, e.g., JFFS2 or YAFFS)

Another significant feature of Flash RAM is that writes/erases are 
MUCH slower than reads (even for the NAND flash that everyone uses 
these days) But, still, blindingly fast compared to waiting for a 
disk to rotate under the read/write head, so you can keep stuff in a 
cache, and blast it out to the disk when you shutdown, before the 
power supply goes away. (Most power supplies, for instance, can hold 
up for 8 milliseconds.. that's an eternity when your write time is 
measured in nanoseconds)