From d0a397903a5732560ed5dd354e7dad0789eae1dc Mon Sep 17 00:00:00 2001
From: pjotr
Date: Fri, 29 May 2026 08:56:56 +0000
Subject: docker

---
 scripts/lib-pangenome-pack.sh | 65 +++++++++++++++++++++++++++++++++++++++----
 1 file changed, 60 insertions(+), 5 deletions(-)

(limited to 'scripts')

diff --git a/scripts/lib-pangenome-pack.sh b/scripts/lib-pangenome-pack.sh
index 35c50f7..efc5281 100644
--- a/scripts/lib-pangenome-pack.sh
+++ b/scripts/lib-pangenome-pack.sh
@@ -23,6 +23,58 @@ mkdir -p "$DEST_DIR"
 
 DATE=$(date +%Y%m%d)
 GB_HASH=$(git -C "$CHANNEL_DIR" rev-parse --short=8 HEAD)
+# The pack is built for the host architecture (no cross-compile).
+# Embed it in the filename so x86_64 / aarch64 / ... images cannot
+# be confused.  Optional TUNE env var passes through to
+# `guix pack --tune=...` and is appended to the arch slug.
+#
+# Only the psABI v-levels are accepted as TUNE values:
+#
+#     x86-64      baseline (any 64-bit Intel/AMD CPU)
+#     x86-64-v2   Nehalem / Bulldozer (SSE4.2 + POPCNT)
+#     x86-64-v3   Haswell / Zen 1     (AVX2 + BMI1/2 + FMA)
+#     x86-64-v4   Skylake-SP / Zen 4  (AVX-512 F/DQ/CD/BW/VL)
+#
+# Microarch names like `cascadelake`, `znver3`, `skylake-avx512`
+# are NOT accepted: Go (which the closure pulls in via odgi etc.)
+# only understands the v-levels and `guix pack --tune=cascadelake`
+# fails with "compiler go@... does not support micro-architecture
+# cascadelake".  Pick the v-level whose feature set is implied by
+# your target microarch (e.g. Cascade Lake/Zen 4 -> v4 because
+# both have AVX-512).
+ARCH=$(uname -m)
+ARCH_SLUG="$ARCH"
+TUNE="${TUNE:-}"
+if [ -n "$TUNE" ]; then
+    case "$TUNE" in
+        x86-64|x86-64-v2|x86-64-v3|x86-64-v4) : ;;
+        *)
+            cat >&2 <<EOF
+TUNE=$TUNE is not supported.  Use one of the psABI v-levels:
+
+    x86-64      baseline (any 64-bit Intel/AMD CPU)
+    x86-64-v2   Nehalem / Bulldozer (SSE4.2 + POPCNT)
+    x86-64-v3   Haswell / Zen 1     (AVX2 + BMI1/2 + FMA)
+    x86-64-v4   Skylake-SP / Zen 4  (AVX-512)
+
+Microarch names (cascadelake, znver3, ...) are rejected because Go
+in the closure only supports the v-levels.
+EOF
+            exit 2
+            ;;
+    esac
+    ARCH_SLUG="$ARCH-$TUNE"
+fi
+
+cpu_compat () {
+    case "$TUNE" in
+        "")        printf 'Generic x86_64 -- any 64-bit Intel or AMD CPU (no AVX/AVX2/AVX-512 required).' ;;
+        x86-64)    printf 'Any 64-bit Intel or AMD CPU (psABI baseline, since ~2003).' ;;
+        x86-64-v2) printf 'Intel Nehalem (1st-gen Core i7) or newer / AMD Bulldozer or newer.  Requires SSE4.2 + POPCNT.' ;;
+        x86-64-v3) printf 'Intel Haswell (4th-gen Core) or newer / AMD Excavator / Zen 1 or newer.  Requires AVX2 + BMI1/2 + FMA.' ;;
+        x86-64-v4) printf 'Intel Skylake-SP / Cascade Lake / Ice Lake / Sapphire Rapids (Xeon Scalable, Core-X 7900X+) / AMD Zen 4 / Zen 5.  Requires AVX-512 (F/DQ/CD/BW/VL).' ;;
+    esac
+}
 
 # Resolve the exact version each package contributes to the closure.
 # `guix package -A` is regex-on-name and can be ambiguous (multiple
@@ -40,13 +92,14 @@ IMPG_VER=$(resolve_version '(@ (gn packages pangenome-rust) impg)' impg)
 WFMASH_VER=$(resolve_version '(@ (gn packages pangenome) wfmash-0.14-snapshot)' wfmash)
 PGGB_VER=$(resolve_version '(@ (gn packages pangenome) pggb)' pggb)
 
-VERSION_STEM="guix-bioinformatics-$GB_HASH-impg-$IMPG_VER-wfmash-$WFMASH_VER-pggb-$PGGB_VER"
+VERSION_STEM="$ARCH_SLUG-guix-bioinformatics-$GB_HASH-impg-$IMPG_VER-wfmash-$WFMASH_VER-pggb-$PGGB_VER"
 
 pangenome_pack () {
     fmt="$1" ; ext="$2" ; label="$3"
-    echo "==> building $fmt pack from $CHANNEL_DIR"
+    echo "==> building $fmt pack from $CHANNEL_DIR${TUNE:+ (tune=$TUNE)}"
     extra=""
     [ "$fmt" = "docker" ] && extra="--entry-point=/bin/bash --image-tag=pangenome-tools:$GB_HASH"
+    [ -n "$TUNE" ] && extra="$extra --tune=$TUNE"
     # shellcheck disable=SC2086
     STORE_PATH=$(guix pack -f "$fmt" --no-offload \
                           -L "$CHANNEL_DIR" \
@@ -83,7 +136,7 @@ pangenome_write_outputs () {
 
     # Inventory is identical regardless of pack format -- name it
     # by channel hash + date only.
-    INVENTORY="$DEST_DIR/pangenome-tools-guix-bioinformatics-$GB_HASH-$DATE.md"
+    INVENTORY="$DEST_DIR/pangenome-tools-$ARCH_SLUG-guix-bioinformatics-$GB_HASH-$DATE.md"
     TOOLS_TSV=$(mktemp)
     CLEAN_TSV=$(mktemp)
     trap 'rm -f "$TMP" "$TOOLS_TSV" "$CLEAN_TSV"' EXIT
@@ -128,9 +181,11 @@ SCM
     dashes () { printf '%*s' "$1" '' | tr ' ' -; }
 
     {
-        echo "# pangenome-tools $DATE (guix-bioinformatics @ $GB_HASH)"
+        echo "# pangenome-tools $DATE ($ARCH_SLUG, guix-bioinformatics @ $GB_HASH)"
+        echo
+        echo "Built from \`mempang-workshop\` in guix-bioinformatics @ $GB_HASH for $ARCH_SLUG."
         echo
-        echo "Built from \`mempang-workshop\` in guix-bioinformatics @ $GB_HASH."
+        echo "**CPU compatibility:** $(cpu_compat)"
         echo
         printf "| %-${NAME_W}s | %-${VER_W}s | %-${DESC_W}s |\n" \
                "Tool" "Version" "Description"
-- 
cgit 1.4.1