SATA AN addendum, draft

David Burg daviburg at windows.microsoft.com
Tue Oct 9 15:48:20 PDT 2007


* From the T10 Reflector (t10 at t10.org), posted by:
* David Burg <daviburg at windows.microsoft.com>
*
Hello again,
I'd like to follow-up on the question 2: we do not concur with Pioneer's
numbers. We actually observed important further power saving with SATA AN
added to devices that already have PM capability. Also, not only the device
saves additional power with AN + PM from PM only, also the host saves
additional power.
I am unable however at this time to release the exact metrics.
Best regards,
David Burg.
-----Original Message-----
From: owner-mtfuji5 at avc-pioneer.com [mailto:owner-mtfuji5 at avc-pioneer.com] On
Behalf Of David Burg
Sent: Tuesday, October 09, 2007 8:40 AM
To: mtfuji5 at avc-pioneer.com
Cc: T10 Reflector
Subject: RE: SATA AN addendum, draft
Hi Katata-san,
Ok, looks like we agree now on question 1.
For the question 2 power saving issue, let me follow-up internally with the
SATA experts who claim the power saving.
Best regards,
David Burg.
-----Original Message-----
From: owner-mtfuji5 at avc-pioneer.com [mailto:owner-mtfuji5 at avc-pioneer.com] On
Behalf Of keiji_katata at post.pioneer.co.jp
Sent: Monday, October 08, 2007 10:12 PM
To: mtfuji5 at avc-pioneer.com
Cc: T10 Reflector
Subject: RE: SATA AN addendum, draft
Hi David,
About Question 1, I made a misunderstanding of SATA specification. To notify
a
SATA-AN, drive wake SATA IF up from PM (Partial/Slumber) state to active
state
by COMWAKE. Then drive sends Notify_host to host.
Drive will clear NotifyPending flag in drive memory when any command is
received
|from host.
About Question 2, I think SATA-AN is not significant to reduce power
consumption. When the next command from host is not received, drive will
change
the SATA IF state into PM (partial/slumber) state. During not ready
condition,
host will send a polling command / 1 sec. SATA IF will be active state then
the
polling command will be carried out. After 5msec, drive will change the SATA
IF
into PM again.
The difference between SATA-AN enabled and not enabled is 5/1000. Slumber
mode
may reduce about 20mA of power consumption (I just tested my one sample
drive).
So SATA-AN may reduce 0.1mA/sec (20*5/1000). This is 36mA/Hour.
Most important item to reduce power consumption is that host and drive
support
PM (partial/slumber) state correctly.
And it depends on the PC system requirement. I think Desk top PC may not mind
the 20mA at usual operation condition.
Best regards,
Keiji Katata
PIONEER CORP.
David Burg <daviburg at windows.microsoft.com>@avc-pioneer.com on 2007/10/05
07:26:41
mtfuji5 at avc-pioneer.comに返信してください
送信者:     owner-mtfuji5 at avc-pioneer.com
宛先:  "mtfuji5 at avc-pioneer.com" <mtfuji5 at avc-pioneer.com>
cc:    T10 Reflector <t10 at t10.org>
bcc:
件名:  RE: SATA AN addendum, draft
Hi Katata-san and all,
Question No 1:
I do not understand this question. More exactly, I do not understand what the
verb "to assert" mean in this SATA Asynchronous Notification context. I have
carefully reviewed the SATA specification 2.5 and do not see this verb used
anywhere close to Asynchronous Notification description (while it is used for
other parts of SATA). Unlike PATA that uses lines, SATA uses packets.
Probably
what you are looking in "assert" is a state machine. The state in SATA AN for
the logical unit is the NotifyPending variable. This variable is described in
detail in SATA and I further repeated the information in the appendix
proposal.
Question No 2:
SATA AN is critical to reduce power consumption. Without SATA AN, each second
the CPU is awaken to generate the GESN, the host controller is waken up, the
SATA bus is waken up, and the device is waken up to answer the command. i.e
SATA
consumes about as more power as PATA. With SATA AN, all of CPU, controller
and
bus can be asleep until the device generate the Set Device FIS AN. On a
typical
use of a laptop, this could keep the bus asleep for several hours, sometime
for
the entire time the laptop is on battery.
The requirement for the device to support SATA AN does not break the
compatibility of the device with older host controller chips not supporting
SATA
AN. This is not a breaking change. However, systems with older host
controller
chips will indeed likely not be able to pass future Microsoft WHQL once we
add
the verification test (because of the host controller chip, not the optical
device).
Best regards,
David Burg.
-----Original Message-----
From: owner-mtfuji5 at avc-pioneer.com [mailto:owner-mtfuji5 at avc-pioneer.com] On
Behalf Of keiji_katata at post.pioneer.co.jp
Sent: Monday, October 01, 2007 6:50 PM
To: mtfuji5 at avc-pioneer.com
Cc: T10 Reflector
Subject: RE: SATA AN addendum, draft
Hi David and all,
I have two comments and a question.
Comment No 1:
I made a comment in the Sep meeting that "I do not know the functionality of
current Pioneer SATA IF drive. So if SATA AN is mandatory function and/or if
No.2 is requested, Pioneer requests to postpone the voting. Because I need to
check the functionality of the current Pioneer SATA drive."
--------
Question: which is correct description?
1:When new event is generated in drive, if AN-SATA is not asserted, drive
shall
assert AN-SATA. In this case, host must de-assert the AN-SATA before issuing
GESN command to cover the small window.
2: When new event is generated and some events remains in drive, if AN-SATA
is
not asserted or is de-asserted, drive shall assert AN-SATA. In this case,
de-assertion must be controlled by drive CPU.
--------
At the discussion, No.1 is confirmed. So your proposal may have No.1
description
as a drive behavior.
Comment No 2:
Now I confirmed Pioneer SATA drive functionality. Pioneer uses several Chips
for
SATA drive models. Some of the Chips do not have FIS register for SATA AN.
Question:
Is the SATA AN so important? Why current PC system can work without SATA AN?
Should Pioneer and its partner stop current business? (Note: for normal DVD
ROM/Writer drive, Fuji7 Rev.0.91 does not request any change.)
Best regards,
Keiji Katata
PIONEER CORP.
David Burg <daviburg at windows.microsoft.com>@avc-pioneer.com on 2007/10/02
08:29:35
mtfuji5 at avc-pioneer.comに返信してください
送信者:     owner-mtfuji5 at avc-pioneer.com
宛先:  "mtfuji5 at avc-pioneer.com" <mtfuji5 at avc-pioneer.com>
cc:
bcc:
件名:  RE: SATA AN addendum, draft
Hi again,
With Dave Walp we reviewed the voted proposal. By mistake the mandatory
support
wording was lost when the text about the feature bit was removed in favor of
the
SATA identify packet command response. Accordingly we think it would be fair
to
vote this point at the next meeting.
Best regards,
David Burg.
-----Original Message-----
From: owner-mtfuji5 at avc-pioneer.com [mailto:owner-mtfuji5 at avc-pioneer.com] On
Behalf Of David Burg
Sent: Monday, October 01, 2007 4:10 PM
To: mtfuji5 at avc-pioneer.com
Subject: RE: SATA AN addendum, draft
Hi Katata-san,
1. PDF version of proposal
Please find the PDF version attached, compressed in a zip archive.
2. Not voted item
There might be some misunderstanding. It was Microsoft's initial proposal to
make the use of Asynchronous Notification mandatory for optical devices on
SATA.
Hence the voted item might have mismatched the proposal if Pioneer understood
the voting item differently.
Notice that this requirement is specific to optical devices as probably the
most
popular PC device with removable media. For these removable media devices
also
the implementation details of the Asynchronous Notification can be specified
-
while for other type of devices it might not be clear what Asynchronous
Notification would be used for. Hence we believe the requirement belongs to
the
SFF specification, while other SATA devices might not have use of
Asynchronous
Notification thus it should remain optional for other SATA devices.
Does this make sense?
Best regards,
David Burg.
-----Original Message-----
From: owner-mtfuji5 at avc-pioneer.com [mailto:owner-mtfuji5 at avc-pioneer.com] On
Behalf Of keiji_katata at post.pioneer.co.jp
Sent: Sunday, September 30, 2007 11:02 PM
To: mtfuji5 at avc-pioneer.com
Subject: Re: SATA AN addendum, draft
Hi David,
I have two items to consult with you.
1. PDF version of proposal
My PC software (MS-Word) may not show your DOCX correctly. Could you prepare
PDF
version?
2. Not voted item
We have never discussed about following sentence of your document.
------
B-13 Asynchronous Notification
Peripheral Device Type 5 Logical unit using Serial ATA Revision 2.5 or later
compliant to this SFF pecification shall implement the SATA Asynchronous
Notification. In other words, IDENTIFY PACKET DEVICE information Word 78 bit
5
shall be set to 1 to indicate the support of Asynchronous Notification. See
Serial ATA specification for further details about IDENTIFY PACKET DEVICE
command.
------
 According to "Serial ATA Revision 2.6 15-February-2007" Asynchronous
Notification is optional. So is this new voting item? Then I will not include
all of "Asynchronous Notification" into the next Fuji revision.
Best regards,
Keiji Katata
PIONEER CORP.
David Burg <daviburg at windows.microsoft.com>@avc-pioneer.com on 2007/10/01
11:51:14
mtfuji5 at avc-pioneer.comに返信してください
送信者:     owner-mtfuji5 at avc-pioneer.com
宛先:  "mtfuji5 at avc-pioneer.com" <mtfuji5 at avc-pioneer.com>
cc:
bcc:
件名:  SATA AN addendum, draft
Dear all,
I have quickly put together the definition and appendix requested by the Fuji
committee to Microsoft. The examples at the end while helpful might be
currently
a little difficult to read. I’ll work on an easier to read revision as time
permits.
Best regards,
David Burg.
<Mt Fuji proposal - SATA AN for MMC - Definiton and Appendix - 1.docx>
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org



More information about the T10 mailing list