Vista's ReadyBoost feature and flash drives

rgratto2

2[H]4U
Joined
Jul 22, 2005
Messages
2,319
For starters, I'm not quite sure where to appropriately post this, because the topic could fall under disk storage, operating systems, or general software, so i opted for generality.

That being said, I was reading up on some of the features of Vista (Here), notably ReadyBoost and ReadyDrive. These features exploit the negligible read times of these devices to essentially use them similarly to RAM (or are they just being used to store the prefetch cache data?). The article also mentions that there are soon to be hard drives that are a kind of hybrid drive with both normal disks and flash storage. But, don't flash drives have a limited number of reads and writes? Wouldn't these features effectively negatively impact this limitation? Also, wouldn't these aforementioned hard drives have a shorter lifetime due to this fact?
 
flash drives have a limit of reads and writes?!??!?!!?!?
 
The people that designed ReadyBoost (and ReadyDrive's basic workings) knew that people would immediately pounce on Flash-RAM's limited write-cycle lifetime, meaning they can only handle so many write operations.

Mind you, Vista had been in process for nearly 5 years before it hit store shelves, so a lot of work was done on what could be done and what was possible based on the technology of 5-6 years ago.

A lot has changed since then.

Yes, Flash-RAM products still have a limited lifespan with respects to the number of write-cycles before "damage" takes place and the stick or chip turns into something that simply doesn't work anymore. One big change has come from designing memory controllers that actually "spread" the data around. It uses write-leveraging and other fancy named schemes (sorry, it's late, or early depending on your POV) that basically doesn't continuously write to the same exact memory locations each time a write occurs.

If you put one single file on a Flash-RAM stick, say a 4KB text file, and then you erased it, and then recopied it back to the memory, it would not be in the same exact place it was before. It'll still be in the memory, but it won't occupy the same spot - hence the controller is aware that the memory from the first copy was used, and it won't write the same data there again like magnetic media would. Flash-RAM has a limited write amount; the reading of data from Flash-RAM can happen basically forever with little ill effects. But writing data is where the "damage" comes from.

Put a single file on a blank floppy drive and it'll occupy the first sectors for holding data - erase the file, or format the disc, and recopy the file a second time, and it's put in the same place or it should be based on how magnetic media operates. Flash-RAM used to operate like that and because of that the lifespan was dramatically reduced.

But the write-leveraging nowadays extends the lifespan of USB sticks and other Flash-RAM based products like 50-fold or more. Tests show the write-leveraging (I read this in a magazine for tech heads like me, sorry I can't quote an online source but Google should track down something if you word the search right) extends the usable write limit by a factor of 45 so, the days of dead USB sticks just because you wrote stuff to it a few hundred times are long gone - that is unless you fill the entire stick of Flash-RAM with each write operation, erase it all, then redo the entire stick over and over constantly, a few thousand times a minute, for YEARS.

ReadyBoost is used to increase the access time for data retrieval since Flash-RAM is still noticeably slower in most situations when it comes to actual reading and writing data. It's in the access times where it shines. Most modern hard drives have roughly 10ms access times for any data on the platters; high end SCSI drives and the vaunted Raptor tend to have 5-7ms access times accounting for rotational latency.

Compare that with nanosecond access times for Flash-RAM devices and you can easily see where the benefits are quite tangible, even for Joe Average, consumer that happens to own a computer with Vista on it.

So, hopefully that covers a little ground on why ReadyBoost won't kill off USB sticks if you choose to use them to enhance Vista's SuperPrefetch and RAM management. The ReadyBoost team put out some information around Vista's launch that showed their own in-house burn in testing showed most USB sticks used for ReadyBoost would last 10 years or more when used specifically for the type of random access "boosting" they do when used with Vista.

If anyone is still using the same USB stick 10 years from now, I'll shoot 'em in the head myself. :)

Hope this helps...
 
thanks for the read; holy crap i wasnt ever aware of flash drives having a limit.
 
Haha, that more than helps, that entirely answers my question. Thanks a lot for the excellent read. Otherwise..I'm of the opinion that this time of day is late at night ;)
 
nice, bbz hit the nail on the head, I was ready to write out that artical myself. dodged the bullet :cool:

and exactly, the same zone on flash memory won't be written to twice until one complete revolution has occured.
 
If I remember correctly, ReadyBoost has a negligible effect on performance for systems with 2GB of RAM or more, so if you have that much in a system running Vista then there really isn't much of a point anyways. (I could be wrong though)
 
Oh, ya i wasnt really concerned about running ReadyBoost myself (may even move up to 4gb), I was just generally curious. More so in regards to the hybrid hard drives too.
 
Back
Top