svaba_jemalloc

#!/usr/bin/env bash
#
# svaba_jemalloc — run svaba with jemalloc forced as the allocator via
# LD_PRELOAD. Recommended for Linux workloads at high thread counts
# (~12+ threads): jemalloc's thread-local arenas relieve lock
# contention the glibc allocator suffers under heavy malloc/free
# traffic from the per-thread assembly pipeline. Typical wall-time
# win: 10–20 % at -p 16 on a WGS region; less at low -p, sometimes a
# wash under -p 4.
#
# *Do not use this on macOS.* Apple's libmalloc (nanomalloc fast path)
# plus DYLD_INSERT_LIBRARIES overhead has empirically run 5–10× SLOWER
# than the default allocator on the same BAM. Stick with system malloc
# there.
#
# Usage:
#   svaba_jemalloc run -t tumor.bam -n normal.bam -G ref.fa ...
#     # or any other svaba subcommand — args are passed through.
#
# Optional environment overrides:
#   JEMALLOC_LIB=/path/to/libjemalloc.so.2    # skip auto-detection
#   SVABA=/path/to/svaba                      # skip $PATH lookup
#
# MALLOC_CONF is passed through if the caller sets it; a reasonable
# default for high-concurrency svaba runs is:
#   MALLOC_CONF=background_thread:true,narenas:24,dirty_decay_ms:10000
# Tune `narenas` to match -p if you go above 24 threads.

set -euo pipefail

# Refuse to run on macOS — jemalloc loses badly there.
case "$(uname -s)" in
  Darwin)
    echo "svaba_jemalloc: Linux only. On macOS, Apple's libmalloc is" >&2
    echo "                faster than jemalloc for this workload." >&2
    echo "                Just run svaba directly." >&2
    exit 1
    ;;
esac

# 1) Resolve libjemalloc path.
JEMALLOC_LIB=${JEMALLOC_LIB:-}
if [[ -z "$JEMALLOC_LIB" ]]; then
  for cand in \
      /usr/lib/x86_64-linux-gnu/libjemalloc.so.2 \
      /usr/lib64/libjemalloc.so.2 \
      /usr/lib/libjemalloc.so.2 \
      /usr/local/lib/libjemalloc.so.2 \
      /opt/homebrew/lib/libjemalloc.so.2
  do
    if [[ -f "$cand" ]]; then JEMALLOC_LIB=$cand; break; fi
  done
fi
if [[ -z "$JEMALLOC_LIB" ]]; then
  # last-ditch: ask ldconfig for a cached match
  JEMALLOC_LIB=$(ldconfig -p 2>/dev/null | awk '/libjemalloc\.so\.2/ {print $NF; exit}')
fi
if [[ -z "$JEMALLOC_LIB" || ! -f "$JEMALLOC_LIB" ]]; then
  echo "svaba_jemalloc: could not find libjemalloc.so.2." >&2
  echo "  Install it (e.g. 'apt install libjemalloc2' or 'yum install jemalloc')" >&2
  echo "  or set JEMALLOC_LIB=/path/to/libjemalloc.so.2." >&2
  exit 1
fi

# 2) Resolve svaba binary.
SVABA=${SVABA:-}
if [[ -z "$SVABA" ]]; then
  # Prefer $PATH; fall back to a build/svaba next to this script.
  if command -v svaba >/dev/null 2>&1; then
    SVABA=$(command -v svaba)
  else
    here=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
    for cand in "$here/build/svaba" "$here/bin/svaba"; do
      if [[ -x "$cand" ]]; then SVABA=$cand; break; fi
    done
  fi
fi
if [[ -z "$SVABA" || ! -x "$SVABA" ]]; then
  echo "svaba_jemalloc: could not find the svaba binary." >&2
  echo "  Put svaba on \$PATH, or set SVABA=/path/to/svaba." >&2
  exit 1
fi

exec env LD_PRELOAD="$JEMALLOC_LIB" "$SVABA" "$@"