add erase code microbenchmark (#312)

Summary:
Pull Request resolved: #312

This diff add Reed-solomon erase code to WDL. Since our own prod implementation is not open-source yet, we use the implementation from isa-l as a proxy for now.

also contains a couple of other minor fixes.

Reviewed By: charles-typ, mcfi

Differential Revision: D87477309

fbshipit-source-id: c7e510af94a6d9546fb9181522954ef8fd111aad

Author: YifanYuan3

packages/wdl_bench/convert.py

@@ -30,6 +30,8 @@
         parse_line.parse_line_xxhash_benchmark(f, sum_c)
     elif sys.argv[1] == "container_hash_maps_bench":
         parse_line.parse_line_container_hash_maps_bench(f, sum_c)
+    elif sys.argv[1] == "erasure_code_perf":
+        parse_line.parse_line_erasure_code_perf(f, sum_c)
     else:
         parse_line.parse_line(f, sum_c)
 

packages/wdl_bench/install_wdl_bench.sh

@@ -18,6 +18,7 @@ declare -A REPOS=(
     ['libaegis']='https://github.com/aegis-aead/libaegis.git'
     ['xxhash']='https://github.com/Cyan4973/xxHash.git'
     ['glibc']='https://sourceware.org/git/glibc.git'
+    ['isa-l']='https://github.com/intel/isa-l.git'
 )
 
 declare -A TAGS=(
@@ -29,6 +30,7 @@ declare -A TAGS=(
     ['libaegis']='0.4.2'
     ['xxhash']='136cc1f8fe4d5ea62a7c16c8424d4fa5158f6d68'
     ['glibc']="glibc-${GLIBC_VERSION}"
+    ['isa-l']='d36de972efc18f2e85ca182a8b6758ecc7da512b'
 )
 
 declare -A DATASETS=(
@@ -51,12 +53,12 @@ LINUX_DIST_ID="$(awk -F "=" '/^ID=/ {print $2}' /etc/os-release | tr -d '"')"
 
 if [ "$LINUX_DIST_ID" = "ubuntu" ]; then
   apt install -y cmake autoconf automake flex bison \
-    nasm clang patch git libssl-dev \
+    nasm clang patch git libssl-dev libc6-dev\
     tar unzip perl openssl python3-dev gawk
 
 elif [ "$LINUX_DIST_ID" = "centos" ]; then
   dnf install -y cmake autoconf automake flex bison \
-    meson nasm clang patch \
+    meson nasm clang patch glibc-static\
     git tar unzip perl openssl-devel python3-devel gawk
 fi
 
@@ -71,7 +73,7 @@ fi
 
 ##################### BUILD AND INSTALL FUNCTIONS #########################
 
-folly_benchmark_list="concurrency_concurrent_hash_map_bench hash_hash_benchmark container_hash_maps_bench stats_digest_builder_benchmark fibers_fibers_benchmark crypto_lt_hash_benchmark memcpy_benchmark memset_benchmark io_async_event_base_benchmark io_iobuf_benchmark function_benchmark random_benchmark synchronization_small_locks_benchmark synchronization_lifo_sem_bench range_find_benchmark"
+folly_benchmark_list="concurrency_concurrent_hash_map_bench hash_hash_benchmark container_hash_maps_bench stats_digest_builder_benchmark fibers_fibers_benchmark crypto_lt_hash_benchmark memcpy_benchmark memset_benchmark io_async_event_base_benchmark io_iobuf_benchmark function_benchmark random_benchmark synchronization_small_locks_benchmark synchronization_lifo_sem_bench range_find_benchmark hash_checksum_benchmark"
 
 fbthrift_benchmark_list="ProtocolBench VarintUtilsBench"
 
@@ -146,7 +148,7 @@ build_lzbench()
     pushd "${WDL_SOURCE}"
     clone $lib || echo "Failed to clone $lib"
     cd "$lib" || exit
-    make -j
+    make BUILD_STATIC=1 -j
     cp ./lzbench "${WDL_ROOT}/" || exit
 
     download_dataset 'silesia'
@@ -235,6 +237,20 @@ build_glibc()
     popd || exit
 }
 
+build_isa_l()
+{
+    lib='isa-l'
+    pushd "${WDL_SOURCE}"
+    clone $lib || echo "Failed to clone $lib"
+    cd "$lib" || exit
+    ./autogen.sh
+    ./configure
+    make perfs -j
+    cp ./erasure_code/erasure_code_perf "${WDL_ROOT}/" || exit
+
+    popd || exit
+}
+
 
 ##################### BUILD AND INSTALL #########################
 
@@ -248,6 +264,7 @@ build_vdso
 build_libaegis
 build_xxhash
 build_glibc
+build_isa_l
 
 cp "${BPKGS_WDL_ROOT}/run.sh" ./
 cp "${BPKGS_WDL_ROOT}/run_prod.sh" ./

packages/wdl_bench/parse_line.py

@@ -190,3 +190,12 @@ def parse_line_container_hash_maps_bench(f, sum_c):
     for k, v in data.items():
         if re.search("^(Find)|(Insert)|(InsertSqBr)|(Erase)|(Iter)", k):
             sum_c[k] = v
+
+
+def parse_line_erasure_code_perf(f, sum_c):
+    for line in f:
+        elements = line.split()
+        if re.search("warm", elements[0]):
+            name = elements[0]
+            value = float(elements[-2])
+            sum_c[name + ": MB/s"] = value

packages/wdl_bench/run.sh

@@ -122,6 +122,7 @@ main() {
 
     if [ "$run_type" = "prod" ]; then
         bash "${WDL_ROOT}/run_prod.sh"
+        exit 0
     fi
 
     set -u  # Enable unbound variables check from here onwards

packages/wdl_bench/run_prod.sh

@@ -30,27 +30,39 @@ Usage: ${0##*/} [-h] [--type single_core|all_core|multi_thread]
 EOF
 }
 
-prod_benchmark_list_mem="memcpy_benchmark bench-memcmp"
+prod_benchmark_list_mem="memcpy_benchmark bench-memcmp memset_benchmark"
 prod_benchmark_list_hash="hash_hash_benchmark xxhash_benchmark"
 prod_benchmark_compression="lzbench"
 prod_benchmark_crypto="openssl libaegis_benchmark"
-prod_benchmark_checksum="hash_checksum_benchmark"
+prod_benchmark_checksum="hash_checksum_benchmark erasure_code_perf"
 prod_benchmark_rng="random_benchmark"
 prod_benchmark_chm="concurrency_concurrent_hash_map_bench"
 prod_benchmark_thrift="ProtocolBench VarintUtilsBench"
 prod_benchmark_f14="container_hash_maps_bench"
 prod_benchmark_lock="synchronization_small_locks_benchmark synchronization_lifo_sem_bench"
 prod_benchmark_vdso="vdso_bench"
 
-prod_benchmarks="memcpy_benchmark bench-memcmp hash_hash_benchmark xxhash_benchmark lzbench openssl libaegis_benchmark hash_checksum_benchmark random_benchmark concurrency_concurrent_hash_map_bench ProtocolBench VarintUtilsBench container_hash_maps_bench synchronization_small_locks_benchmark synchronization_lifo_sem_bench vdso_bench"
+prod_benchmarks="memcpy_benchmark memset_benchmark bench-memcmp hash_hash_benchmark xxhash_benchmark lzbench openssl libaegis_benchmark hash_checksum_benchmark  erasure_code_perf random_benchmark concurrency_concurrent_hash_map_bench ProtocolBench VarintUtilsBench container_hash_maps_bench synchronization_small_locks_benchmark synchronization_lifo_sem_bench vdso_bench"
 
-benchmark_non_json_list=("openssl" "libaegis_benchmark" "lzbench" "vdso_bench" "xxhash_benchmark" "concurrency_concurrent_hash_map_bench" "container_hash_maps_bench")
+benchmark_non_json_list=("openssl" "libaegis_benchmark" "lzbench" "vdso_bench" "xxhash_benchmark" "concurrency_concurrent_hash_map_bench" "container_hash_maps_bench" "erasure_code_perf")
+
+exec_non_json() {
+  local input="$1"
+  for item in "${benchmark_non_json_list[@]}"; do
+    if [[ "$item" == "$input" ]]; then
+      return 0
+    fi
+  done
+
+  return 1
+}
 
 run_list=""
 
 declare -A prod_benchmark_config=(
     ['random_benchmark']="--bm_regex=xoshiro --json"
     ['memcpy_benchmark']="--json"
+    ['memset_benchmark']="--json"
     ['hash_hash_benchmark']="--bm_regex=RapidHash --json"
     ['hash_checksum_benchmark']="--json"
     ['synchronization_lifo_sem_bench']="--bm_min_iters=1000000 --json"
@@ -59,12 +71,13 @@ declare -A prod_benchmark_config=(
     ['ProtocolBench']="--bm_regex=\"(^Binary)|(^Compact)Protocol\" --json"
     ['VarintUtilsBench']=" --json"
     ['concurrency_concurrent_hash_map_bench']=""
-    ['lzbench']="-v -ezstd1,3 ${WDL_DATASETS}/${dataset}"
+    ['lzbench']="-v -ezstd,1,3 ${WDL_DATASETS}/silesia.tar"
     ['openssl']="speed -seconds 20 -evp aes-256-gcm"
     ['vdso_bench']="-t 10 -p 20"
     ['libaegis_benchmark']=""
     ['xxhash_benchmark']="xxh3"
     ['bench-memcmp']=""
+    ['erase_code_perf']=""
 )
 
 main() {
@@ -123,12 +136,12 @@ main() {
             ldconfig
         fi
         out_file=""
-        if [[ " $benchmark_non_json_list{[*]} " =~ " ${benchmark} " ]]; then
+        if exec_non_json "${benchmark}"; then
             out_file="out_${benchmark}.txt"
         else
             out_file="out_${benchmark}.json"
         fi
-         "./${benchmark}" "${prod_benchmark_config[$benchmark]}" 2>&1 | tee -a "${out_file}"
+        bash -c "./${benchmark} ${prod_benchmark_config[$benchmark]}" 2>&1 | tee -a "${out_file}"
         if [ "$benchmark" = "openssl" ]; then
             unset LD_LIBRARY_PATH
             ldconfig
@@ -142,7 +155,7 @@ main() {
     fi
 
     for benchmark in $run_list; do
-        if [[ " $benchmark_non_json_list{[*]} " =~ " ${benchmark} " ]]; then
+        if exec_non_json "${benchmark}"; then
             python3 ./convert.py "$benchmark"
         fi
     done