pixi.toml

[workspace]
authors = ["Seth Stadick <seth_stadick@bio-rad.com>"]
channels = [
    #"https://conda.modular.com/max-nightly",
    "https://prefix.dev/conda-forge",
    "https://conda.modular.com/max",
    "https://repo.prefix.dev/modular-community",
]
platforms = ["osx-arm64", "linux-64", "linux-aarch64"]
preview = ["pixi-build"]

[package]
name = "ish"
version = "1.4.0"
license = "Apache-2.0"

[package.build]
backend = { name = "pixi-build-mojo", version = "0.*" }
channels = [
    "https://prefix.dev/pixi-build-backends",
    "https://prefix.dev/conda-forge",
]

[package.host-dependencies]
mojo-compiler = "=0.25.6"

[package.build-dependencies]
extramojo = "=0.18.0"
mojo-compiler = "=0.25.6"
zlib = ">=1.3.1,<2"

[package.run-dependencies]
mojo-compiler = "=0.25.6"
zlib = ">=1.3.1,<2"


[tasks]
build = "mojo build -D ISH_LOG_LEVEL=info --march native -o ish main.mojo"
build_generic = "mojo build -D ISH_SIMD_TARGET=baseline -D ISH_LOG_LEVEL=info -o ish main.mojo"
debug = "mojo build -D ISH_LOG_LEVEL=debug --debug-level line-tables --march native -o ish main.mojo"
pkglib = "mojo package ./ishlib"
testlib = "mojo test tests -I . --co && pixi run mojo test tests -I ."

# Bench the gpu fun in ish using Bench
build-bench-gpu = "mojo build -D ISH_LOG_LEVEL=timing --march native -I . -o bench-gpu ./benchmarking/ish_bench_aligner_gpu_only.mojo"

# Bench any algorithm using internal timings
build-bench = "mojo build -D SIMD_MOD='none' --march native  -I . -o ish-aligner ./benchmarking/ish_bench_aligner.mojo"
build-bench-128 = "mojo build -D SIMD_MOD='sse' --march native  -I . -o ish-aligner-128 ./benchmarking/ish_bench_aligner.mojo"
build-bench-256 = "mojo build -D SIMD_MOD='avx2' --march native  -I . -o ish-aligner-256 ./benchmarking/ish_bench_aligner.mojo"
build-bench-512 = "mojo build -D SIMD_MOD='avx512' --march native  -I . -o ish-aligner-512 ./benchmarking/ish_bench_aligner.mojo"
build-bench-gpu-all = "mojo build -D ISH_LOG_LEVEL=timing -D SIMD_MOD='none' --march native  -I . -o ish-aligner-gpu ./benchmarking/ish_bench_aligner_gpu.mojo"


# Below here all all tasks for running benchmarks

[tasks.get-bench-data]
cmd = """
mkdir -p bench/data && 
cd bench/data && 
git clone https://github.com/jeffdaily/parasail && 
mkdir refdata && 
cd refdata && 
curl https://ftp.uniprot.org/pub/databases/uniprot/previous_releases/release-2015_11/knowledgebase/uniprot_sprot-only2015_11.tar.gz --output uniprot_sprot-only2015_11.tar.gz && tar -xvzf uniprot_sprot-only2015_11.tar.gz && 
gunzip uniprot_sprot.fasta.gz &&
awk 'BEGIN {RS=">"} NR > 1  && NR < 12 { printf ">%s", $0 }' uniprot_sprot.fasta > uniprot_sprot.small.fasta && 
cat uniprot_sprot.fasta uniprot_sprot.fasta uniprot_sprot.fasta uniprot_sprot.fasta uniprot_sprot.fasta > uniprot_sprot_5x.fasta &&
cat uniprot_sprot_5x.fasta uniprot_sprot_5x.fasta > uniprot_sprot_10x.fasta 
"""
outputs = ["bench/data/**/*.fasta"]

[tasks.compile-parasail]
depends-on = ["get-bench-data"]
cmd = """
cd bench/data/parasail &&
autoreconf -fi &&
./configure &&
make -j $(nproc)
"""
outputs = ["bench/data/parasail/apps/parasail_aligner"]

[tasks.run-bench-sw]
depends-on = [
    "compile-parasail",
    "build-bench-128",
    "build-bench-256",
    "build-bench-512",
]
cmd = """
bash -c '
  mkdir -p bench_results/sw/plots &&
  ARCH=$(uname -m) &&
  ARCH_INFO="${ARCH_INFO:-$ARCH: $(lscpu | grep \"Model name\" | cut -d: -f2 | xargs || echo \"Unknown CPU\")}" &&
  if [[ "$ARCH" == "x86_64" ]]; then
      PARASAIL_INSTS="sse41_128 avx2_256"
      ISH_BINS="$(readlink -f ish-aligner-128) $(readlink -f ish-aligner-256) $(readlink -f ish-aligner-512)"
  elif [[ "$ARCH" == "aarch64" || "$ARCH" == "arm64" ]]; then
      PARASAIL_INSTS="neon_128"
      ISH_BINS="$(readlink -f ish-aligner-128)"
  else
      echo "Unsupported architecture: $ARCH" >&2
      exit 1
  fi &&
  python benchmarking/bench_scripts/bench.py \
      --ish-binaries $ISH_BINS \
      --ish-algorithm striped-local \
      --parasail-aligner-binary $(readlink -f ./bench/data/parasail/apps/parasail_aligner) \
      --parasail-algorithm sw \
      --score-size adaptive \
      --query-seqs-base-dir ./bench/data/parasail/data/ \
      --ref-data bench/data/refdata/uniprot_sprot.fasta \
      --parasail-instruction-set $PARASAIL_INSTS \
      --output-dir ./bench_results/sw/ \
      > ./bench_results/sw/benchmark_results.csv &&
  python benchmarking/bench_scripts/plot.py \
      ./bench_results/sw/benchmark_results.csv \
      "$ARCH_INFO" \
      ./bench_results/sw
'
"""
outputs = [
    "./bench_results/sw/*.csv",
    "./bench_results/sw/benchmark_results.csv",
]

[tasks.run-bench-sg]
depends-on = [
    "compile-parasail",
    "build-bench-128",
    "build-bench-256",
    "build-bench-512",
]
cmd = """
bash -c '
  mkdir -p bench_results/sg/plots &&
  ARCH=$(uname -m) &&
  ARCH_INFO="${ARCH_INFO:-$ARCH: $(lscpu | grep \"Model name\" | cut -d: -f2 | xargs || echo \"Unknown CPU\")}" &&
  if [[ "$ARCH" == "x86_64" ]]; then
      PARASAIL_INSTS="sse41_128 avx2_256"
      ISH_BINS="$(readlink -f ish-aligner-128) $(readlink -f ish-aligner-256) $(readlink -f ish-aligner-512)"
  elif [[ "$ARCH" == "aarch64" || "$ARCH" == "arm64" ]]; then
      PARASAIL_INSTS="neon_128"
      ISH_BINS="$(readlink -f ish-aligner-128)"
  else
      echo "Unsupported architecture: $ARCH" >&2
      exit 1
  fi &&
  python benchmarking/bench_scripts/bench.py \
      --ish-binaries $ISH_BINS \
      --ish-algorithm striped-semi-global \
      --parasail-aligner-binary $(readlink -f ./bench/data/parasail/apps/parasail_aligner) \
      --parasail-algorithm sg \
      --score-size word \
      --query-seqs-base-dir ./bench/data/parasail/data/ \
      --ref-data bench/data/refdata/uniprot_sprot.fasta \
      --parasail-instruction-set $PARASAIL_INSTS \
      --output-dir ./bench_results/sg/ \
      > ./bench_results/sg/benchmark_results.csv &&
  python benchmarking/bench_scripts/plot.py \
      ./bench_results/sg/benchmark_results.csv \
      "$ARCH_INFO" \
      ./bench_results/sg
'
"""
outputs = [
    "./bench_results/sg/*.csv",
    "./bench_results/sg/benchmark_results.csv",
]


[tasks.run-bench-gpu]
depends-on = ["get-bench-data", "build-bench-gpu-all"]
cmd = """
bash -c '
  mkdir -p bench_results/gpu/plots &&
  if command -v nvidia-smi &> /dev/null; then
      GPU_INFO=$(nvidia-smi --query-gpu=name --format=csv,noheader | head -n1 | sed "s/NVIDIA //;s/GeForce //;s/-.*//" | xargs)
  elif command -v rocm-smi &> /dev/null; then
      GPU_INFO=$(rocm-smi --showproductname | grep "GPU" | head -n1 | awk "{print $NF}" | sed "s/AMD //;s/Radeon //" | xargs)
  elif command -v lspci &> /dev/null; then
      GPU_INFO=$(lspci | grep -ioE "(NVIDIA|Radeon|GeForce|RTX|GTX|Quadro|Tesla|RX)[^[]*" | head -n1 | sed "s/NVIDIA //;s/GeForce //;s/Radeon //;s/Corporation //" | xargs)
  else
      echo "No suported GPU Found" && exit 1
  fi &&
  python benchmarking/bench_scripts/bench.py \
      --ish-binaries $(readlink -f ./ish-aligner-gpu) \
      --ish-algorithm basic-semi-global-gpu-parallel \
      --score-size word \
      --query-seqs-base-dir ./bench/data/parasail/data/ \
      --ref-data bench/data/refdata/uniprot_sprot_5x.fasta \
      --devices 4 \
      --iterations 1 \
      --output-dir ./bench_results/gpu/ \
      > ./bench_results/gpu/benchmark_results.csv &&
  python benchmarking/bench_scripts/plot.py \
      --is-gpu \
      ./bench_results/gpu/benchmark_results.csv \
      "$GPU_INFO" \
      ./bench_results/gpu
'
"""
outputs = [
    "./bench_results/sg/*.csv",
    "./bench_results/sg/benchmark_results.csv",
]

[tasks.get-chm13v2]
args = [{ arg = "download_location", default = "./bench/refdata/" }]
cmd = """
bash -c '
DOWNLOAD_LOC="{{ download_location }}"
mkdir -p $DOWNLOAD_LOC &&
curl -L https://s3-us-west-2.amazonaws.com/human-pangenomics/T2T/CHM13/assemblies/analysis_set/chm13v2.0.fa.gz | gunzip > $DOWNLOAD_LOC/chm13v2.0.fa
'
"""

[tasks.run-whole-genome-search]
depends-on = [
    "build",
    { task = "get-chm13v2", args = [
        "{{download_location}}",
    ] },
]
args = [{ arg = "download_location", default = "./bench/refdata/" }]
cmd = """
bash -c '
mkdir -p ./bench_results/whole_genome/
DOWNLOAD_LOC="{{ download_location }}"
SIZE=$(stat -c%s $DOWNLOAD_LOC/chm13v2.0.fa) &&
echo "Searching through CHM13 reference ($SIZE bytes)..." &&
START=$(date +%s.%N) &&
./ish --scoring-matrix actgn --record-type fastx --threads 16 ACCTTTTGGTTATATCATTCTTACATAAAGGACACTGTGAAGGCCCTTTCTTCTGGTTGAGA $DOWNLOAD_LOC/chm13v2.0.fa > /dev/null &&
END=$(date +%s.%N) &&
ELAPSED=$(echo "$END - $START" | bc) &&
BYTES_PER_SEC=$(echo "scale=2; $SIZE / $ELAPSED" | bc) &&
MB_PER_SEC=$(echo "scale=2; $BYTES_PER_SEC / 1048576" | bc) &&
echo "Search complete: $SIZE bytes in $ELAPSED seconds" &&
echo "Speed: $BYTES_PER_SEC bytes/sec ($MB_PER_SEC MB/sec)" > ./bench_results/whole_genome/speed.txt
'
"""
outputs = ["./bench_results/whole_genome/speed.txt"]

[tasks.bench-all-cpu]
depends-on = ["run-bench-sw", "run-bench-sg", "run-whole-genome-search"]

[tasks.bench-all-gpu]
depends-on = ["run-bench-gpu"]

[tasks.clean-benchmarks]
cmd = """
rm -rf ./bench_results
"""

[dependencies]
mojo = "=0.25.6"
# extramojo = { git = "https://github.com/ExtraMojo/ExtraMojo.git" }
extramojo = "=0.18.0"
defopt = ">=7.0.0,<8"
pandas = ">=2.3.3,<3"
matplotlib = ">=3.10.7,<4"
numpy = ">=2.3.4,<3"