(SNES) Why is this function even here.

[Sarah Shinespark]

I see this code all the time at the start of text display code in Arcana. This is more me being exasperated at it than asking help, but can anyone make sense of all this multiplication and addition? Does 4216 affect the stack or something? Code before this is the main event handler so it wouldn't have pushed something recently.

Code Select Expand

$87/9797 22 39 8A 80 JSL $808A39[$80:8A39]   A:0000 X:0008 Y:9540 P:envmxdIzC
$80/8A39 20 3D 8A    JSR $8A3D  [$80:8A3D]   A:0000 X:0008 Y:9540 P:envmxdIzC

$80/8A3D DA          PHX                     A:0000 X:0008 Y:9540 P:envmxdIzC
$80/8A3E 48          PHA                     A:0000 X:0008 Y:9540 P:envmxdIzC
$80/8A3F DA          PHX                     A:0000 X:0008 Y:9540 P:envmxdIzC
$80/8A40 64 02       STZ $02    [$00:1E02]   A:0000 X:0008 Y:9540 P:envmxdIzC
$80/8A42 E2 10       SEP #$10                A:0000 X:0008 Y:9540 P:envmxdIzC
$80/8A44 AA          TAX                     A:0000 X:0008 Y:0040 P:envmXdIzC
$80/8A45 8E 02 42    STX $4202  [$80:4202]   A:0000 X:0000 Y:0040 P:envmXdIZC
$80/8A48 FA          PLX                     A:0000 X:0000 Y:0040 P:envmXdIZC
$80/8A49 8E 03 42    STX $4203  [$80:4203]   A:0000 X:0008 Y:0040 P:envmXdIzC
$80/8A4C EA          NOP                     A:0000 X:0008 Y:0040 P:envmXdIzC
$80/8A4D EA          NOP                     A:0000 X:0008 Y:0040 P:envmXdIzC
$80/8A4E AF 16 42 80 LDA $804216[$80:4216]   A:0000 X:0008 Y:0040 P:envmXdIzC
$80/8A52 85 00       STA $00    [$00:1E00]   A:0000 X:0008 Y:0040 P:envmXdIZC
$80/8A54 FA          PLX                     A:0000 X:0008 Y:0040 P:envmXdIZC
$80/8A55 8E 03 42    STX $4203  [$80:4203]   A:0000 X:0000 Y:0040 P:envmXdIZC
$80/8A58 A5 01       LDA $01    [$00:1E01]   A:0000 X:0000 Y:0040 P:envmXdIZC
$80/8A5A 18          CLC                     A:0000 X:0000 Y:0040 P:envmXdIZC
$80/8A5B 6D 16 42    ADC $4216  [$80:4216]   A:0000 X:0000 Y:0040 P:envmXdIZc
$80/8A5E FA          PLX                     A:0000 X:0000 Y:0040 P:envmXdIZc
$80/8A5F FA          PLX                     A:0000 X:0000 Y:0040 P:envmXdIZc
$80/8A60 8E 02 42    STX $4202  [$80:4202]   A:0000 X:0000 Y:0040 P:envmXdIZc
$80/8A63 FA          PLX                     A:0000 X:0000 Y:0040 P:envmXdIZc
$80/8A64 8E 03 42    STX $4203  [$80:4203]   A:0000 X:0008 Y:0040 P:envmXdIzc
$80/8A67 EA          NOP                     A:0000 X:0008 Y:0040 P:envmXdIzc
$80/8A68 EA          NOP                     A:0000 X:0008 Y:0040 P:envmXdIzc
$80/8A69 6F 16 42 80 ADC $804216[$80:4216]   A:0000 X:0008 Y:0040 P:envmXdIzc
$80/8A6D 85 01       STA $01    [$00:1E01]   A:0000 X:0008 Y:0040 P:envmXdIZc
$80/8A6F 90 04       BCC $04    [$8A75]      A:0000 X:0008 Y:0040 P:envmXdIZc
$80/8A75 FA          PLX                     A:0000 X:0008 Y:0040 P:envmXdIZc
$80/8A76 8E 03 42    STX $4203  [$80:4203]   A:0000 X:0000 Y:0040 P:envmXdIZc
$80/8A79 A5 02       LDA $02    [$00:1E02]   A:0000 X:0000 Y:0040 P:envmXdIZc
$80/8A7B 18          CLC                     A:0000 X:0000 Y:0040 P:envmXdIZc
$80/8A7C 6D 16 42    ADC $4216  [$80:4216]   A:0000 X:0000 Y:0040 P:envmXdIZc
$80/8A7F 85 02       STA $02    [$00:1E02]   A:0000 X:0000 Y:0040 P:envmXdIZc
$80/8A81 C2 10       REP #$10                A:0000 X:0000 Y:0040 P:envmXdIZc
$80/8A83 60          RTS                     A:0000 X:0000 Y:0040 P:envmxdIZc
$80/8A3C 6B          RTL                     A:0000 X:0000 Y:0040 P:envmxdIZc


[Cyneprepou4uk]

https://en.wikibooks.org/wiki/Super_NES_Programming/SNES_Hardware_Registers

[Sarah Shinespark]

I already know putting two values in 4202 and 4203 will produce the product in 4216, it's all the PHX that are confusing me.

[Cyneprepou4uk]

This code just stores 2 copies of X in the stack, no big deal.

What's confusing is that it pushes 3 bytes at the beginning, but pulls 6 bytes during execution. These 3 additional pulls must be pulling bytes of the return address. Is that what you were asking?

Can your logger log stack pointer as well? That should make things clear.

[Vehek]

I was confused by that part (the differing number of pushes and pulls) too when I read the log, but given the SEP #10, it should have gone from a 16-bit X register to an 8-bit register before it starts pulling from the stack, if I'm reading the processor flag documentation correctly. Both X and A were 16-bit, so 6 bytes were pushed then.

[Sarah Shinespark]

Here's without Squelch:

Code Select Expand

$87/9797 22 39 8A 80 JSL $808A39[$80:8A39]   A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FF3 P:envmxdIzC HC:0216 VC:249 FC:13 I:00
$80/8A39 20 3D 8A    JSR $8A3D  [$80:8A3D]   A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FF0 P:envmxdIzC HC:0270 VC:249 FC:13 I:00

$80/8A3D DA          PHX                     A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FEE P:envmxdIzC HC:0316 VC:249 FC:13 I:00
$80/8A3E 48          PHA                     A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FEC P:envmxdIzC HC:0350 VC:249 FC:13 I:00
$80/8A3F DA          PHX                     A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FEA P:envmxdIzC HC:0384 VC:249 FC:13 I:00
$80/8A40 64 02       STZ $02    [$00:1E02]   A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FE8 P:envmxdIzC HC:0418 VC:249 FC:13 I:00
$80/8A42 E2 10       SEP #$10                A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FE8 P:envmxdIzC HC:0452 VC:249 FC:13 I:00
$80/8A44 AA          TAX                     A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FE8 P:envmXdIzC HC:0476 VC:249 FC:13 I:00
$80/8A45 8E 02 42    STX $4202  [$80:4202]   A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FE8 P:envmXdIZC HC:0494 VC:249 FC:13 I:00
$80/8A48 FA          PLX                     A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FE8 P:envmXdIZC HC:0524 VC:249 FC:13 I:00
$80/8A49 8E 03 42    STX $4203  [$80:4203]   A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FE9 P:envmXdIzC HC:0596 VC:249 FC:13 I:00
$80/8A4C EA          NOP                     A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FE9 P:envmXdIzC HC:0626 VC:249 FC:13 I:00
$80/8A4D EA          NOP                     A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FE9 P:envmXdIzC HC:0644 VC:249 FC:13 I:00
$80/8A4E AF 16 42 80 LDA $804216[$80:4216]   A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FE9 P:envmXdIzC HC:0662 VC:249 FC:13 I:00
$80/8A52 85 00       STA $00    [$00:1E00]   A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FE9 P:envmXdIZC HC:0704 VC:249 FC:13 I:00
$80/8A54 FA          PLX                     A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FE9 P:envmXdIZC HC:0738 VC:249 FC:13 I:00
$80/8A55 8E 03 42    STX $4203  [$80:4203]   A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEA P:envmXdIZC HC:0770 VC:249 FC:13 I:00
$80/8A58 A5 01       LDA $01    [$00:1E01]   A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEA P:envmXdIZC HC:0800 VC:249 FC:13 I:00
$80/8A5A 18          CLC                     A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEA P:envmXdIZC HC:0834 VC:249 FC:13 I:00
$80/8A5B 6D 16 42    ADC $4216  [$80:4216]   A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEA P:envmXdIZc HC:0852 VC:249 FC:13 I:00
$80/8A5E FA          PLX                     A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEA P:envmXdIZc HC:0888 VC:249 FC:13 I:00
$80/8A5F FA          PLX                     A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEB P:envmXdIZc HC:0920 VC:249 FC:13 I:00
$80/8A60 8E 02 42    STX $4202  [$80:4202]   A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEC P:envmXdIZc HC:0952 VC:249 FC:13 I:00
$80/8A63 FA          PLX                     A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEC P:envmXdIZc HC:0982 VC:249 FC:13 I:00
$80/8A64 8E 03 42    STX $4203  [$80:4203]   A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FED P:envmXdIzc HC:1014 VC:249 FC:13 I:00
$80/8A67 EA          NOP                     A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FED P:envmXdIzc HC:1044 VC:249 FC:13 I:00
$80/8A68 EA          NOP                     A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FED P:envmXdIzc HC:1062 VC:249 FC:13 I:00
$80/8A69 6F 16 42 80 ADC $804216[$80:4216]   A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FED P:envmXdIzc HC:1080 VC:249 FC:13 I:00
$80/8A6D 85 01       STA $01    [$00:1E01]   A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FED P:envmXdIZc HC:1122 VC:249 FC:13 I:00
$80/8A6F 90 04       BCC $04    [$8A75]      A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FED P:envmXdIZc HC:1156 VC:249 FC:13 I:00
$80/8A75 FA          PLX                     A:0000 X:0008 Y:0002 D:1E00 DB:80 S:1FED P:envmXdIZc HC:1180 VC:249 FC:13 I:00
$80/8A76 8E 03 42    STX $4203  [$80:4203]   A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEE P:envmXdIZc HC:1212 VC:249 FC:13 I:00
$80/8A79 A5 02       LDA $02    [$00:1E02]   A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEE P:envmXdIZc HC:1242 VC:249 FC:13 I:00
$80/8A7B 18          CLC                     A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEE P:envmXdIZc HC:1276 VC:249 FC:13 I:00
$80/8A7C 6D 16 42    ADC $4216  [$80:4216]   A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEE P:envmXdIZc HC:1294 VC:249 FC:13 I:00
$80/8A7F 85 02       STA $02    [$00:1E02]   A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEE P:envmXdIZc HC:1330 VC:249 FC:13 I:00
$80/8A81 C2 10       REP #$10                A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEE P:envmXdIZc HC:1364 VC:249 FC:13 I:00
$80/8A83 60          RTS                     A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FEE P:envmxdIZc HC:0024 VC:250 FC:13 I:00
$80/8A3C 6B          RTL                     A:0000 X:0000 Y:0002 D:1E00 DB:80 S:1FF0 P:envmxdIZc HC:0070 VC:250 FC:13 I:00


[Raeven0]

If I've done my math correctly, this routine is multiplying the 16-bit A and X inputs, and storing the 32-bit result at $1e00.

[Sarah Shinespark]

All that for 0 x 8?? What a waste of time...

[Sarah Shinespark]

They're basically the same thing with different connotations. Coding is writing code. Programming is... writing programs.
October 22, 2020, 04:01:23 PM - (Auto Merged - Double Posts are not allowed before 7 days.)

If I've done my math correctly, this routine is multiplying the 16-bit A and X inputs, and storing the 32-bit result at $1e00.

Thanks again btw, knowing that helped out a LOT. What I thought was unrelated code after the end of interesting code, was just calculating the offset of the text to read next.

Also turns out all the text in the status screen is hard-coded addresses in a long list rather than referencing the text tables with a pointer table. What a waste of space...

[KingMike]

I already know putting two values in 4202 and 4203 will produce the product in 4216, it's all the PHX that are confusing me.

I would say it's simply a delay because the real CPU needs some time to calculate the result (looks to be about 8 cycles, from code I've seen in other games), but it's doing that BEFORE writing the values.

The ACTUAL needed delay, I'm not sure if it is known.
I do know early in emulation days, it doesn't seem like it was known, since it wasn't mentioned. So that's one source of ROM hacks possibly being incompatible with real hardware, but even then I don't think it was until the late 2000s that byuu tried to figure it out and wrote probably one of the first emulators to support that limitation (certainly earlier SNES9x and ZSNES would not emulate the necessary delay for properly-calculated results allow $4216 to give the correct result instantly).

[Raeven0]

The ACTUAL needed delay, I'm not sure if it is known.

1 CPU cycle per significant bit of the $4202 input, totaling 8 to use the entire input. The delay in the posted code cleverly uses two NOPs (4 cycles) followed by LDA $804216, the latter being 4 bytes long and therefore costing 4 cycles to read before it can be executed.

[Aaendi]

All that for 0 x 8?? What a waste of time...

Sorry for the bump, but THIS right here is a perfect example of what is wrong with the vast majority of SNES games.  It really gets on my nerves when people keep claiming that "SNES games had slowdown because of tight deadlines, and they didn't have enough time to optimize code", when they had enough time to type out all these random opcodes for no rhyme or reason.

[Sarah Shinespark]

Sorry for the bump, but THIS right here is a perfect example of what is wrong with the vast majority of SNES games.  It really gets on my nerves when people keep claiming that "SNES games had slowdown because of tight deadlines, and they didn't have enough time to optimize code", when they had enough time to type out all these random opcodes for no rhyme or reason.

Well I'm sure they had higher level language dev environments that converted to ASM for production. (What I would give to take a look at one!) Could be the original code, or inefficiencies in their compilers.

[FAST6191]

Well I'm sure they had higher level language dev environments that converted to ASM for production. (What I would give to take a look at one!) Could be the original code, or inefficiencies in their compilers.

I don't know if the dev tools extended as far back as the SNES or if it was just games (there are definitely dev tools for later systems) but might have had something in the gigaleaks.
Also for the gameboy but contemporary at least to the SNES
https://www.pagetable.com/?p=28

All that said I have never seen devs talk about high level stuff for the SNES and whatnot. Indeed many talk* about how the PS1 and later N64 was a big shift in allowing C for the games themselves (with manglement also noting it as not having to hire greybeards and super turbo nerds that still know/would learn some assembly to use in anger). Some of the talk we have seen is about low level stuff, or indeed how music making types were also getting down and dirty with assembly, but I readily accept that "rule of cool" is in play for that one. Similarly disassemblies we have seen rarely have markers/fingerprints of compilers of the day, much less cross compilers** as that was still a distinction some made, and it would be years before all the people having real fun with compiler theory and giving us intrinsics and all the fun optimisations that might make such things less blindingly obvious. The leaked SDKs that have been around for years also go into serious depth with the hardware which if everybody had nice libraries (as they do on later systems, which we also saw correspondingly less useful info in) would be redundant at best.


*gdc talks ( https://www.youtube.com/watch?v=aMcJ1Jvtef0 ), other conference talks, gamasutra has a bunch of stuff (not really relevant here but https://www.gamasutra.com/view/feature/131781/the_internet_sucks_or_what_i_.php is one of my favourite articles), general interviews with games sites, ars technica "war stories", various developer commentaries (for the N64 but I really like conker king https://www.youtube.com/channel/UCbRUl7dkRVPKlYiUq4TmEaQ/videos ), especially about 10 years back when the indie devs rediscovered 2d gaming and classic styles such that the old hands still alive were dragged in to do things or teach people.

**Most consoles having chips used for "serious business" purposes, x86 having not quite won the battle yet, so having something there means there were some desires to have an actual compiler.