Hi all, what is the best way (from Basic or machine code) to detect a 128k machine (i.e. one that supports the kl_bank_switch call) vs a 64k machine?
Thanks,
Lachlan
Hi Lachlan
I don't know if it's the best way but the below is a fast way to test if a machine has more than the first 64kb ram available.
Code below #4000 or above #8000, address #4000 in 64Kb and first byte in 16kb bank corrupted.
org &1000
di
ld bc,#7f00
out (c),c
xor a
ld (&4000),a
ld c,#c4
out (c),c
ld a,#ff
ld (&4000),a
ld c,#c0
out (c),c
ld a,(&4000)
cp 0
jr z,found_expansion
ld hl,KB64
jr pr
.found_expansion
ld hl,KB128
.pr ld a,(hl)
cp 0
ret z
call &bb5a
inc hl
jr pr
.KB64 text "No expansion ram found",0
.KB128 text "Expansion ram found",0
Quote from: lachlank on 10:07, 25 August 09
Hi all, what is the best way (from Basic or machine code) to detect a 128k machine (i.e. one that supports the kl_bank_switch call) vs a 64k machine?
Thanks,
Lachlan
Sorry a typo crept in!
Corrected!
/Ygdrazil
Quote from: Ygdrazil on 13:09, 25 August 09
Hi Lachlan
I don't know if it's the best way but the below is a fast way to test if a machine has more than the first 64kb ram available.
Code below #4000 or above #8000, address #4000 in 64Kb and first byte in 16kb bank corrupted.
org &1000
di
ld bc,#7fc0
out (c),c
xor a
ld (&4000),a
ld c,#c4
out (c),c
ld a,#ff
ld (&4000),a
ld c,#c0
out (c),c
ld a,(&4000)
cp 0
jr z,found_expansion
ld hl,KB64
jr pr
.found_expansion
ld hl,KB128
.pr ld a,(hl)
cp 0
ret z
call &bb5a
inc hl
jr pr
.KB64 text "No expansion ram found",0
.KB128 text "Expansion ram found",0
Ah yeah I wondered about that. Thanks.
Got another game in the pipeline so am experimenting with an "enhanced" 128k version :)
Quote from: lachlank on 10:07, 25 August 09
Hi all, what is the best way (from Basic or machine code) to detect a 128k machine (i.e. one that supports the kl_bank_switch call) vs a 64k machine?
Thanks,
Lachlan
Hmmm.. you probably need to check the rom version numbers.. or possibly peek the location of that function in the firmware jumpblock to ensure it's valid.
A tad late for a reply, nevertheless both RAM banks and Jumpblock checks are necessary if your program uses _KL_BANK_SWITCH.
The _KL_BANK_SWITCH function is available in Firmware v2 (CPC 664) and up. So, if you only check that the RAM banks are available (eg. using the snippet posted by Ygdrazil), your program will crash on a CPC 464 (Firmware v1, no _KL_BANK_SWITCH available) with a RAM expansion. If you only check the jumpblock, your program will crash on a bare CPC 664 (Firmware v2 but 64K RAM only) and 464Plus too (Firmware v4 and 64K RAM only).
So you should first check the jumpblock to make sure the _KL_BANK_SWITCH is there, then check if there's any extra RAM available to enable the 128K features of your program.
A jumpblock check could be done like this:
_KL_BANK_SWITCH equ &BD5B
; check jumpblock
; firmware v1 has &EF,&37,&33 located at &BD5B
; (&33 XOR &37) XOR &EF = &EB
ld hl,(_KL_BANK_SWITCH)
ld a,(_KL_BANK_SWITCH+2)
xor h
xor l
cp &EB
jr z,no_kl_bank_switch
; KL_BANK_SWITCH is available
; proceed...
Another option would be to simply drop _KL_BANK_SWITCH and access directly to the MMR register with a single OUT instruction. It's safe (to some extent) and you won't have to bother anymore about the firmware version (assuming your program doesn't call specific Firmware v2+ functions).