SSD SMART attributes

[Franc Zabkar]

The following SMART attributes appear in the documentation for HDAT2:

ID Description
-----------------------------
170 (AAh) Reserved Block Count
171 (ABh) Program Fail Count
172 (ACh) Erase Fail Count
173 (ADh) Wear Leveling Count
174 (AEh) Unexpected Power Loss

I expect that they are SSD attributes only.

See http://www.hdat2.com/files/hdat2en_v11.pdf

- Franc Zabkar

 

[calypso]

The following SMART attributes appear in the documentation for HDAT2:

ID Description
-----------------------------
170 (AAh) Reserved Block Count
171 (ABh) Program Fail Count
172 (ACh) Erase Fail Count
173 (ADh) Wear Leveling Count
174 (AEh) Unexpected Power Loss

I expect that they are SSD attributes only.

See http://www.hdat2.com/files/hdat2en_v11.pdf

Wear leveling is SSD-only feature...

Reserved block count can be applied to both SSD and rotational drives...

Unexpected power loss? I think it can be applied to both technologies...

Erase fail count - SSD only...

Program fail count - SSD only...

--
"Stars li romobilu slusu ?" upita cokoladao pije Edipog ljubija.
"Ne znam ja nista !" rece banderao prolijece "Ja samo banderao jede lijepm !" By runf

Damir Lukic, calypso@_MAKNIOVO_fly.srk.fer.hr
http://inovator.blog.hr
http://calypso-innovations.blogspot.com/

 

[Franc Zabkar]

Wear leveling is SSD-only feature...

Reserved block count can be applied to both SSD and rotational drives...

Unexpected power loss? I think it can be applied to both technologies...

Erase fail count - SSD only...

Program fail count - SSD only...

I suspect that the above attributes were introduced as a group by one
SSD manufacturer.

The list of attributes also includes the following:

99 (63h) Average FHC (Maxtor)
100 (64h) Erase/Program cycles (Maxtor) ... ???
101 (65h) Maximum FHC (Maxtor)

FHC = flying height control ???

Other interesting attributes are ...

241 (F1h) Total LBA Write (IBM)
242 (F2h) Total LBA Read (IBM)

- Franc Zabkar

 

[Franc Zabkar]

I've found quite a few more SSD & CompactFlash SMART attributes, with
descriptions:

http://www.users.on.net/~fzabkar/SSD_CF_attribs.txt

- Franc Zabkar

 

[Arno]

I've found quite a few more SSD & CompactFlash SMART attributes, with
descriptions:

http://www.users.on.net/~fzabkar/SSD_CF_attribs.txt

Interesting!

Lets see what mine (OCZ Core 32GB) supports:

9 with a raw value of 0, i.e. meaningless
12 with a raw value of 3950, that does not seem to change, again
meaningless
194 Temperature, works
229, 232, 233, 234, 235: Still unknown

A bit disapointing. Not your fault of course.

Arno

 

[Franc Zabkar]

Interesting!

Lets see what mine (OCZ Core 32GB) supports:

9 with a raw value of 0, i.e. meaningless
12 with a raw value of 3950, that does not seem to change, again
meaningless
194 Temperature, works
229, 232, 233, 234, 235: Still unknown

A bit disapointing. Not your fault of course.

Arno

On 19 Jan 2009 21:02:28 GMT your SMART stats were:

9 Power_On_Hours 0x0032 100 100 000 - 0

12 Power_Cycle_Count 0x0032 100 100 000 - 241

194 Temperature_Celsius 0x0007 032 100 000 - 0

229 Unknown_Attribute 0x0002 100 000 000 - 259986015180268

232 Unknown_Attribute 0x0002 100 048 000 - 9028846498104

233 Unknown_Attribute 0x0002 100 000 000 - 0

234 Unknown_Attribute 0x0002 100 000 000 - 592722311424

235 Unknown_Attribute 0x0002 100 000 000 - 187423

Was that a different device?

- Franc Zabkar

 

[Franc Zabkar]

Interesting!

Lets see what mine (OCZ Core 32GB) supports:

9 with a raw value of 0, i.e. meaningless
12 with a raw value of 3950, that does not seem to change, again
meaningless
194 Temperature, works
229, 232, 233, 234, 235: Still unknown

A bit disapointing. Not your fault of course.

Arno

Try Hard Disk Sentinel for Linux:
http://www.hdsentinel.com/hdslin.php

Solid State Disk Compatibility:
http://www.hdsentinel.com/compatibility_ssd.php

- Franc Zabkar

 

[Arno]

On 25 Oct 2009 21:33:05 GMT, Arno <[email protected]> put finger to
keyboard and composed:

On 19 Jan 2009 21:02:28 GMT your SMART stats were:

9 Power_On_Hours 0x0032 100 100 000 - 0

12 Power_Cycle_Count 0x0032 100 100 000 - 241

194 Temperature_Celsius 0x0007 032 100 000 - 0

229 Unknown_Attribute 0x0002 100 000 000 - 259986015180268

232 Unknown_Attribute 0x0002 100 048 000 - 9028846498104

233 Unknown_Attribute 0x0002 100 000 000 - 0

234 Unknown_Attribute 0x0002 100 000 000 - 592722311424

235 Unknown_Attribute 0x0002 100 000 000 - 187423

Was that a different device?

Hmm. No. Let me check again.

Ok, seems temperature is stuck at cooked 32, which means it
does not actually work. My mistake. Speedfan seems to
display temerature though, and it does raise slightly after
power on and makes sense. No idea what it does or where it gets
the data.

Attribute 12 is now 3956, so it might actually be a device
reset count, not a power cycle count. I had one power-cycle
since yesterday, but a number of reboots.

The other attributes are exactly the same as January, except for
234, which is now raw 571364925191.

Arno

 

[Arno]

Hmm. No. Let me check again.

Ok, seems temperature is stuck at cooked 32, which means it
does not actually work. My mistake. Speedfan seems to
display temerature though, and it does raise slightly after
power on and makes sense. No idea what it does or where it gets
the data.

Ah, found it. Speedfan swaps HD0 and HD1 in display for some reasonm
probably a bug. This means it does not get temperature from the
SSD either.

Arno

 

[Franc Zabkar]

Hmm. No. Let me check again.

Ok, seems temperature is stuck at cooked 32, which means it
does not actually work. My mistake. Speedfan seems to
display temerature though, and it does raise slightly after
power on and makes sense. No idea what it does or where it gets
the data.

Attribute 12 is now 3956, so it might actually be a device
reset count, not a power cycle count. I had one power-cycle
since yesterday, but a number of reboots.

The other attributes are exactly the same as January, except for
234, which is now raw 571364925191.

Arno

Attribute 234 looks like it may be multibyte. For example, some bytes
have increased by 7, another has fallen by 5.

592722311424 = 0x008a0100b900
571364925191 = 0x00850800df07

Attribute 232 looks suspiciously like an ASCII string. Aside from the
leading 08, "3631303138" is an ASCII numeric string, "61018".

9028846498104 = 0x083631303138

I can't see any pattern in the following:

259986015180268 = 0xec74b614d5ec

187423 = 0x00000002dc1f

- Franc Zabkar

 

[Franc Zabkar]

Interesting!

Lets see what mine (OCZ Core 32GB) supports:

9 with a raw value of 0, i.e. meaningless
12 with a raw value of 3950, that does not seem to change, again
meaningless
194 Temperature, works
229, 232, 233, 234, 235: Still unknown

See pages 25 and 26 of the following document.

Transcend 2.5" Solid State Disk datasheet:
http://www.transcend.co.jp/support/dlcenter/datasheet/SSD25-JMI Datasheet v1.08.pdf

The attributes look very similar to yours. However, they are 8 bytes
long, not the usual 6 bytes for HD SMART attributes.

It appears that I may have been correct about the ASCII values. Do
they correspond to the firmware version or date, ie October 18, 2006?

- Franc Zabkar

 

[Franc Zabkar]

See pages 25 and 26 of the following document.

Transcend 2.5" Solid State Disk datasheet:
http://www.transcend.co.jp/support/dlcenter/datasheet/SSD25-JMI Datasheet v1.08.pdf

The attributes look very similar to yours. However, they are 8 bytes
long, not the usual 6 bytes for HD SMART attributes.

It appears that I may have been correct about the ASCII values. Do
they correspond to the firmware version or date, ie October 18, 2006?

Here are more data.

http://www.transcend.co.jp/support/dlcenter/datasheet/SSD25S-JMI Datasheet v1.05.pdf

ID(Hex) Description Reference
----------------------------------------------
0C Power Cycle Count (Supported)
09 Power On Hours Count (Not Supported)
C2 Temperature (Not Supported)
E5 Halt System ID, Flash ID (Table 1)
E8 Firmware version information (Table 2)
E9 ECC Fail Record (Table 3)
EA Average Erase Count, Max Erase Count (Table 4)
EB Good Block Count, System Block Count (Table 5)
EC-EF Reserved
F1-FF Reserved


Table 2
-------

Byte Description
----------------------------
0 Year (High Byte, ASCII)
1 Year (Low Byte, ASCII)
2 Month (High Byte, ASCII)
3 Month (Low Byte, ASCII)
4 Day (High Byte, ASCII)
5 Day (Low Byte, ASCII)
6 Channels (binary)
7 Banks (binary)

It appears that the 6 bytes returned by a typical HD SMART utility are
bytes 1 through 6.

This means your YYMMDD date code is x81016, ie October 16, 20x8.

Your SSD would then have 8 channels.

- Franc Zabkar

 

[Franc Zabkar]

Attribute 234 looks like it may be multibyte. For example, some bytes
have increased by 7, another has fallen by 5.

592722311424 = 0x008a0100b900
571364925191 = 0x00850800df07

According to table 4 of the Transcend datasheets, attribute 234 stores
the Average Erase Count and Max Erase Count.

Max Erase Count (Jan / Oct) = 0x00018A / 0x000885
Avg Erase Count (Jan / Oct) = 0x0000b9 / 0x0007DF

Attribute 232 looks suspiciously like an ASCII string. Aside from the
leading 08, "3631303138" is an ASCII numeric string, "61018".

9028846498104 = 0x083631303138

I can't see any pattern in the following:

259986015180268 = 0xec74b614d5ec

The above attribute stores the Halt System ID in byte 0 and the Flash
ID in bytes 1 through 7.

187423 = 0x00000002dc1f--

According to table 5, attribute 235 decodes as follows:

System(Free) Block Count = 0x0200
Good Block Count = 0x--1FDC

- Franc Zabkar

 

[Arno]

According to table 4 of the Transcend datasheets, attribute 234 stores
the Average Erase Count and Max Erase Count.

Max Erase Count (Jan / Oct) = 0x00018A / 0x000885
Avg Erase Count (Jan / Oct) = 0x0000b9 / 0x0007DF

The above attribute stores the Halt System ID in byte 0 and the Flash
ID in bytes 1 through 7.

According to table 5, attribute 235 decodes as follows:

System(Free) Block Count = 0x0200
Good Block Count = 0x--1FDC

Very nice, so may be something useful in these attributes.

Arno

 

[Franc Zabkar]

Very nice, so may be something useful in these attributes.

The smartmontools project will shortly be able to correctly analyse
these attributes. At the moment CrystalDiskInfo (Windows) can do it:

http://crystalmark.info/software/CrystalDiskInfo/index-e.html

In the meantime you can get a dump of the 512-byte SMART data block
(with 8-byte raw attribute values) using smartctl in debug mode:

smartctl -a -r ioctl,2


- Franc Zabkar