#!/usr/bin/env bash
# benchCyclesCalypsi.sh — measure per-call cycles for each benchmark in
# benchmarks/ compiled with Calypsi cc65816 5.16 (--speed -O 2).
# Mirrors benchCyclesPrecise.sh but for Calypsi.  Output: markdown
# table; per-call cycles via MAME emu.time().

set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
PROJECT_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
BENCH_DIR="$PROJECT_ROOT/benchmarks"

CC="$PROJECT_ROOT/tools/calypsi/usr/local/lib/calypsi-65816-5.16/bin/cc65816"
LN="$PROJECT_ROOT/tools/calypsi/usr/local/lib/calypsi-65816-5.16/bin/ln65816"
RUNNER="$PROJECT_ROOT/scripts/runInMameCyclesCalypsi.sh"

LINKER_SCM=$(mktemp --suffix=.scm)
trap 'rm -f "$LINKER_SCM"' EXIT
cat > "$LINKER_SCM" <<EOF
(define memories
  '((memory IIgsCode (address (#x1000 . #x9FFF))
            (section code farcode cdata idata switch data_init_table))
    (memory IIgsBSS  (address (#xA000 . #xBFFF))
            (section stack data zdata heap))
    (memory IIgsNear (address (#x020000 . #x02FFFF))
            (section znear))
    (memory IIgsDP   (address (#x0000 . #x00FF))
            (section (registers ztiny)))
    (memory IIgsVec  (address (#xFF00 . #xFFFF))
            (section (reset #xFFFC)))
    (block stack (size #x200))
    (block heap  (size #x100))
    (base-address _DirectPageStart IIgsDP   0)
    (base-address _NearBaseAddress IIgsNear 0)))
EOF

# Mirror bench definitions from benchCyclesPrecise.sh
benchInputs() {
    case "$1" in
        sumOfSquares) echo 'sumOfSquares(50)';;
        fib)          echo 'fib(10)';;
        strcpy)       echo 'mystrcpy(dst, "hello world!")';;
        memcmp)       echo 'mymemcmp("hello", "hello", 5)';;
        bsearch)      echo 'bsearch(arr, 8, 5)';;
        dotProduct)   echo 'dotProduct(va, vb, 4)';;
        popcount)     echo 'popcount(0x12345678UL)';;
        crc32)        echo 'crc32((const unsigned char *)"hello", 5)';;
        bubbleSort)   echo '(bubbleSort(bsBuf, 16), 0)';;
        strLen)       echo 'strLen("The quick brown fox jumps over the lazy dog!")';;
        djb2Hash)     echo 'djb2Hash("hello world")';;
        *)            echo "/* unknown */";;
    esac
}

benchExtern() {
    case "$1" in
        sumOfSquares) echo 'extern unsigned long sumOfSquares(unsigned short n);';;
        fib)          echo 'extern unsigned short fib(unsigned short n);';;
        strcpy)       echo 'extern char *mystrcpy(char *d, const char *s); static char dst[16];';;
        memcmp)       echo 'extern int mymemcmp(const void *a, const void *b, unsigned int n);';;
        bsearch)      echo 'extern int bsearch(const int *arr, int n, int key); static const int arr[] = {1,2,3,4,5,6,7,8};';;
        dotProduct)   echo 'extern long dotProduct(const short *a, const short *b, unsigned int n); static const short va[] = {1,2,3,4}; static const short vb[] = {5,6,7,8};';;
        popcount)     echo 'extern int popcount(unsigned long x);';;
        crc32)        echo 'extern unsigned long crc32(const unsigned char *p, unsigned int n);';;
        bubbleSort)   echo 'extern void bubbleSort(short *a, unsigned short n); static short bsBuf[16] = {7,3,1,9,4,5,8,2,6,0,15,11,13,10,14,12};';;
        strLen)       echo 'extern unsigned short strLen(const char *s);';;
        djb2Hash)     echo 'extern unsigned long djb2Hash(const char *s);';;
        *)            echo '';;
    esac
}

benchIters() {
    case "$1" in
        sumOfSquares) echo 50;;
        fib)          echo 100;;
        strcpy)       echo 200;;
        memcmp)       echo 500;;
        bsearch)      echo 200;;
        dotProduct)   echo 200;;
        popcount)     echo 100;;
        crc32)        echo 100;;
        bubbleSort)   echo 10;;
        strLen)       echo 200;;
        djb2Hash)     echo 200;;
        *)            echo 50;;
    esac
}

runOneCalypsiBench() {
    local name="$1"
    local extern_decl call_expr iters
    extern_decl=$(benchExtern "$name")
    call_expr=$(benchInputs "$name")
    iters=$(benchIters "$name")
    [ -z "$extern_decl" ] && { echo "skip"; return; }

    local cwrap obench bin
    cwrap=$(mktemp --suffix=.c)
    owrap=$(mktemp --suffix=.o)
    obench=$(mktemp --suffix=.o)
    bin=$(mktemp --suffix=.bin)
    bin_raw="${bin%.bin}.raw"
    lst=$(mktemp --suffix=.lst)
    trap 'rm -f "$cwrap" "$owrap" "$obench" "$bin" "$bin_raw" "$lst"' RETURN

    cat > "$cwrap" <<EOF
$extern_decl
volatile unsigned long sink;
__task int main(void) {
    for (int w = 0; w < 5; w++) sink = (unsigned long)($call_expr);
    *((volatile unsigned long __far *)0x025000) = 0xa1a1UL;
    for (int i = 0; i < $iters; i++) sink = (unsigned long)($call_expr);
    *((volatile unsigned long __far *)0x025002) = 0xa2a2UL;
    while (1) {}
}
EOF
    "$CC" -O 2 --speed --code-model=small -c "$cwrap" -o "$owrap" 2>/dev/null \
        || { echo "compile-fail-wrap"; return; }
    "$CC" -O 2 --speed --code-model=small -c "$BENCH_DIR/$name.c" -o "$obench" 2>/dev/null \
        || { echo "compile-fail-bench"; return; }
    # ln65816 emits the raw alongside the bin; one per memory
    "$LN" "$LINKER_SCM" "$owrap" "$obench" -o "$bin" \
        --output-format raw --raw-multiple-memories --rtattr exit=simplified \
        --list-file "$lst" clib-sc-sd.a 2>/dev/null \
        || { echo "link-fail"; return; }

    local entry
    entry=$(grep "__program_start =" "$lst" | head -1 | awk '{print substr($NF,3)}')
    [ -z "$entry" ] && { echo "no-entry"; return; }

    local val
    val=$(bash "$RUNNER" "$bin_raw" "$entry" "$iters" 2>&1 | grep -oE 'cyc_per_call=[0-9.]+' | head -1 | sed 's/cyc_per_call=//')
    if [ -z "$val" ]; then
        echo "(no read)"
    else
        printf '%.0f cyc/call' "$val"
    fi
}

printf '| Benchmark | Calypsi (cyc) |\n'
printf '|-----------|--------------:|\n'
for src in "$BENCH_DIR"/*.c; do
    name=$(basename "$src" .c)
    extern_decl=$(benchExtern "$name")
    [ -z "$extern_decl" ] && continue
    result=$(runOneCalypsiBench "$name")
    printf '| %s | %s |\n' "$name" "$result"
done