public inbox for linux-mmc@vger.kernel.org
 help / color / mirror / Atom feed
* SD 3.00 Physical layer FULL specification
@ 2012-09-01  2:27 George Spelvin
  2012-09-01 20:19 ` Arnd Bergmann
  0 siblings, 1 reply; 2+ messages in thread
From: George Spelvin @ 2012-09-01  2:27 UTC (permalink / raw)
  To: linux-mmc; +Cc: arnd.bergmann, linux

I was just trolling around the net trying to figure out what the
Pm parameter and speed class of SD cards meant, and I came across a
definitive answer in the form of a copy of Part 1 of the unredacted
v3.00 specification (April 16, 2009) at

http://www.61ic.com/code/attachment.php?aid=73871&k=cfaccbbb80c338769bd7282636c0ca7e&t=1331742726

I have of course snarfed a copy, and if anyone else on the list
would benefit from one, I figured it was worth sharing.

In particular, it documents the FAT update optimizations in SD cards
(section 4.3.1.7, p. 112).

If anyone cares, write performance is defined in terms of updates to
"allocation units" (i.e. erase blocks) 512K-4MiB in size of a card-defined
size, which are made up of "recording units" (i.e. FAT clusters).
Table 4-51 on p. 113 gives the limits.

An update to an allocation unit is a combination of writes of new data
and and moves (copies) of old data from the pre-existing AU.  Each of
these has a performance standard, Pw and Pm, respectively.

For a "Class n" card, Pw and Pr must be >= n MB/s, while Pm must be
>= n/2 MB/s.  (Exacpt, oddly, class 10 has NO lower bound on Pm.)

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: SD 3.00 Physical layer FULL specification
  2012-09-01  2:27 SD 3.00 Physical layer FULL specification George Spelvin
@ 2012-09-01 20:19 ` Arnd Bergmann
  0 siblings, 0 replies; 2+ messages in thread
From: Arnd Bergmann @ 2012-09-01 20:19 UTC (permalink / raw)
  To: George Spelvin; +Cc: linux-mmc

On Saturday 01 September 2012, George Spelvin wrote:
> In particular, it documents the FAT update optimizations in SD cards
> (section 4.3.1.7, p. 112).
> 
> If anyone cares, write performance is defined in terms of updates to
> "allocation units" (i.e. erase blocks) 512K-4MiB in size of a card-defined
> size, which are made up of "recording units" (i.e. FAT clusters).
> Table 4-51 on p. 113 gives the limits.

I believe these recommendations are often ignored anyway. A lot of the
SD cards nowadays have erase block sizes of 1.5, 3, 6, or 8 MB. None
of these can be specified with this version of the standard. Even for
the rare cards that have 1MB or 2MB erase block sizes, I've never seen
one that reports anything but 4MB.

	Arnd


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2012-09-02 18:50 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-09-01  2:27 SD 3.00 Physical layer FULL specification George Spelvin
2012-09-01 20:19 ` Arnd Bergmann

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox