Processor identification

Hi,

The little program below should return the Processor ID on the modern systems. If you have such a system, could you please run this program and let me know what the result is?
[if you want to save space, I only need the ID word, the rest can be derived from it]

REM >SimpleCPUID
REM

DIM code% 15

P% = code%
[ OPT 2

  SWI "OS_EnterOS"
  MRC CP15, 0, R0, C0, C0, 0
  SWI "OS_LeaveOS"
  MOV PC, R14
]
ID% = USR(code%)

PRINT "CPU ID is &"+STR$~(ID%)

bit19% = ((ID% >> 19) AND 1)
bits1512% = ((ID% >> 12) AND 15)

IF (bit19% = 0) THEN
  REM Old style ID
  PRINT "This is an old processor (roughly ARM6-ARM9)."
  PRINT "Run my CPUinfo program for more details."
ELSE
  REM Revised ID
  PRINT "Implementor : &"+STR$~( ((ID% >> 24) AND 255) )
  PRINT "Variant     : &"+STR$~( ((ID% >> 20) AND 15) )
  PRINT "Architecture: &"+STR$~( ((ID% >> 16) AND 15) )
  PRINT "Primary part: &"+STR$~( ((ID% >> 4) AND 4095) )
  IF (((ID% >> 4) AND 4095) = &B76) THEN
    PRINT STRING$(14, " ")+"Original Pi! ;-)"
  ENDIF
  PRINT "Revision    : &"+STR$~( (ID% AND 15) )
ENDIF

END

As far as I am aware, the usual distributions of RISC OS are for:

• IOMD (RiscPC era)
  • Too old, it’ll just say “Old processor”
• Iyonix
  • May just report “Old processor”
• OMAP3
  • Beagle (original)
  • Beagle xM &413FC082
  • IGEPv2 (likely same as one of the above)
  • Pandora console (ditto)
• OMAP4
  • PandaBoard
  • Pandaboard ES &412FC09A
• OMAP5
  • IGEPv5
  • OMAP5432EVM &412FC0F2
• Sitara AM5728
  • Titanium &412FC0F2
• iMX6
  • iMX6 &412FC09A
• ??
  • Rapido IG &412FC0F2
• And the Pi family (oh my God…)
  • Pi model B (first version) (Sony?)
  • Pi model B (tweaked first version) (Sony?)
  • Pi model B revision 2, 256MiB (Sony) – I think mine is a Sony. ID is &410FB767.
  • Pi model B revision 2, 256MiB (Qisda)
  • Pi model B revision 2, 256MiB (Egoman)
  • Pi model A, 256MiB [Sony|Qisda|Egoman]
  • Pi model B revision 2, 512MiB (Sony)
  • Pi model B revision 2, 512MiB (Qisda)
  • Pi model B revision 2, 512MiB (Egoman)
  • Pi model B+ (Sony)
  • Chris’ Pi 512M B (0F) / Pi 256M B (02) / Pi 512M B+ (10) / Pi Compute Module (11) / Pi model A &410FB767
  • Pi Compute Module (Sony)
  • Pi model A+ (Sony)
  • Pi model B+ revision 1.2 (2015 model)
  • Pi2 model B (Sony)
  • Pi2 model B (Embest)
  • Chris’ Pi2 model B (A01041) &410FC075
  • Pi Zero (Sony) &410FB767
  • Pi3 Model B (Sony) &410FD034

Have I missed any? ;-) All of the Pi1 are likely to report the same thing. Ditto the Pi2…

Thanks.

Hi this is an OMAP5432EVM cpu ID &412FC0F2

could you please run this program and let me know what the result is?

Yes.
Rapido IG 412FC0F2
Titanium 412FC0F2
ARMX6 412FC09A
Risc PC No such mnemonic
Pi model 3 410FD034
Panda ES 412FC09A
VRPC No such mnemonic
Pi Zero 410FB767
Pi 2 model B (A01041) 410FC075
Pi 256M B (04) 410FB767
Beagleboard XM 413FC082
Pi 512M B (0F) 410FB767
Pi 256M B (02) 410FB767
Pi 512M B+ (10) 410FB767
Pi Compute Module (11) 410FB767
Pi model a+ (12) can’t find it!
Pi model A 410FB767

Hope this helps.

Risc PC No such mnemonic
VRPC No such mnemonic

Interesting RPCEmu gives &4401A102 and the old processor message

Wow, Chris, that’s a heck of a collection.
You’ve certainly helped fill out the list! I’ve edited the original with the IDs.

It looks like the BCM2835 boards (Pi 1/Compute/Zero) all have an ID of &410FB767, which is what I expected.

BTW, what do you mean by the Pi Model A+ “can’t find it”?

For others reading, the devices outstanding on the list are:

• Iyonix (for completeness)
• Original Beagleboard
• IGEPv2 (to confirm)
• Pandora (to confirm)
• Pandaboard (not ES)

This means…

Just tested my Pi 1 and 3 and I get the same results. As expected, but just confirming :)

BTW, what do you mean by the Pi Model A+ “can’t find it”?

Presumably that it’s hiding in a drawer somewhere!

Presumably that it’s hiding in a drawer somewhere

Found it in this drawer

Just for completeness, RPi A+ returns the same as A/B/B+ etc.
“Model A+ V1.1 2014”

The little program below should return the Processor ID on the modern systems.

Apropros yesterday’s thread about using defined APIs rather than making assumptions about machine level detail, you probably want OS_PlatformFeatures.

This SWI insulates you from ARM’s ever changing definition of the CPU ID register, in fact ARM spilled out into lots more than just one ID register now. Gone are the days when you could say “ah, ARM7, so it has 8k caches”.

My Iyonix reported:-
“This is an old processor (roughly ARM6 – ARM9)

using defined APIs rather than making assumptions about machine level detail

I was wondering when somebody would join the dots. ;-)

Problem is, if a nice API is created tomorrow, it doesn’t help those using older versions (like those who prefer to stick to the “stable” versions). Plus, in my specific case (which IIC bus is the expansion), it would be serious feature creep for PlatformFeatures to return that.

This SWI insulates you from ARM’s ever changing definition of the CPU ID register,

I noticed. That code is a cut down version of my CPUinfo program which “should” cover ARM6 to the Cortex-A8 (and similar). The code for parsing the ID word is rather horrible.

in fact ARM spilled out into lots more than just one ID register now.

Yup. That’s the extended ID stuff. So the processor can finally report useful stuff about its behaviour. Rather necessary given the many permutations of the ARM design around these days, though thankfully in our “world” they’re all fairly similar at application level.

For people who’d like more details on their processor (tested on a Pi and emu, should work on ARM6+):

http://heyrick.co.uk/random/cpuinfo,ff8

Note that even with the shiny new tell-us-lots identification, it appears that the interpretation of some bits vary (does bit=0 mean this or…?) from one core to the next… ;-)

Still outstanding – CPU ID for the original Beagleboard (OMAP3530), and the original Pandaboard (OMAP4430).