Date : Wed, 14 Dec 2011 18:20:02 +0100
From : eelco@... (Eelco Huininga)
Subject: 32016 + 32082
Hi,
Interesting find, this 2nd processor. If you take a look at the Pandora
v0.61 code you will find that it reads the configuration jumpers, and
depending on wether or not bit0 and bit1 are set, it will set the FPU
and MMU flag in the configuration register of the 32016 CPU:
0007DE CE 18 A8 C0 F9 00 ...... movzbd @0xF90000, r0 / read
config jumpers
00 .
0007E5 28 A0 03 (.. andb $0x03, r0 / mask
out bit 0 and 1
0007E8 7C E7 D8 04 |... caseb 0x0007EC[r0:b] / case
jump using r0
0007EC 08 0d .. bicb r1, $<undefined> / ???
0007EE 12 17 .. ret 0x17
0007F0 0E 0B 00 ... setcfg [] / no
jumpers set
0007F3 EA 11 .. but 0x000804
0007F5 0E 0B 01 ... setcfg [f] /
jumper 0 set
0007F8 EA 0C .. but 0x000804
0007FA 0E 0B 02 ... setcfg [m] /
jumper 1 set
0007FD EA 07 .. but 0x000804
0007FF 0E 0B 03 ... setcfg [m,f] /
jumper 0 and 1 set
000802 EA 02 .. but 0x000804
.
000804 17 a0 00 02 00 00 ...... movd $0x00020000, r0
Also, if you look at the photo of your 32016 board you'll see that there
are two jumpers set on the configuration jumper block, which would mean
that both the FPU and MMU are present in the config jumpers. But... the
buffer chip, which presumably should be fitted as IC16, isn't mounted,
so the jumpers can never be read by the software!
Now, if you look at the Pandora v2.00 code, the setcfg[] instruction is
used just once, to set the FPU flag (the setcfg[] instruction is used to
set the CFG-register of the 32016). The configuration jumper register is
never read.
This might suggest that they explored the use of a MMU during the
development of Pandora v0.61 but abandonded it by the time Pandora v2.00
was developed, or that they moved MMU functionality from the firmware to
software.
Cheers,
Eelco
Op 11-12-2011 16:04, John Kortink schreef:
>
> I've just acquired a 32016 Second Processor, and I've
> noticed that this one has a 32082 MMU on board.
>
> Anyone know in what contexts this provides any benefits ?
> Because no pictures on the net seem to show one (it's just
> an empty space, not even a socket). So I suppose it is
> (optionally) useful when running some version of Panos
> (I have both 1.1 and 1.4).
>
>
> John Kortink
>