1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136
#![allow(clippy::integer_arithmetic)] use clap::{crate_description, crate_name, value_t, App, Arg}; use solana_ledger::entry::{self, create_ticks, init_poh, EntrySlice, VerifyRecyclers}; use solana_measure::measure::Measure; use solana_perf::perf_libs; use solana_sdk::hash::hash; fn main() { solana_logger::setup(); let matches = App::new(crate_name!()) .about(crate_description!()) .version(solana_version::version!()) .arg( Arg::with_name("max_num_entries") .long("max-num-entries") .takes_value(true) .value_name("SIZE") .help("Number of entries."), ) .arg( Arg::with_name("start_num_entries") .long("start-num-entries") .takes_value(true) .value_name("SIZE") .help("Packets per chunk"), ) .arg( Arg::with_name("hashes_per_tick") .long("hashes-per-tick") .takes_value(true) .value_name("SIZE") .help("hashes per tick"), ) .arg( Arg::with_name("num_transactions_per_entry") .long("num-transactions-per-entry") .takes_value(true) .value_name("NUM") .help("Skip transaction sanity execution"), ) .arg( Arg::with_name("iterations") .long("iterations") .takes_value(true) .help("Number of iterations"), ) .arg( Arg::with_name("num_threads") .long("num-threads") .takes_value(true) .help("Number of threads"), ) .arg( Arg::with_name("cuda") .long("cuda") .takes_value(false) .help("Use cuda"), ) .get_matches(); let max_num_entries = value_t!(matches, "max_num_entries", u64).unwrap_or(64); let start_num_entries = value_t!(matches, "start_num_entries", u64).unwrap_or(max_num_entries); let iterations = value_t!(matches, "iterations", usize).unwrap_or(10); let hashes_per_tick = value_t!(matches, "hashes_per_tick", u64).unwrap_or(10_000); let start_hash = hash(&[1, 2, 3, 4]); let ticks = create_ticks(max_num_entries, hashes_per_tick, start_hash); let mut num_entries = start_num_entries as usize; if matches.is_present("cuda") { perf_libs::init_cuda(); } init_poh(); while num_entries <= max_num_entries as usize { let mut time = Measure::start("time"); for _ in 0..iterations { assert!(ticks[..num_entries] .verify_cpu_generic(&start_hash) .finish_verify(&ticks[..num_entries])); } time.stop(); println!( "{},cpu_generic,{}", num_entries, time.as_us() / iterations as u64 ); if is_x86_feature_detected!("avx2") && entry::api().is_some() { let mut time = Measure::start("time"); for _ in 0..iterations { assert!(ticks[..num_entries] .verify_cpu_x86_simd(&start_hash, 8) .finish_verify(&ticks[..num_entries])); } time.stop(); println!( "{},cpu_simd_avx2,{}", num_entries, time.as_us() / iterations as u64 ); } if is_x86_feature_detected!("avx512f") && entry::api().is_some() { let mut time = Measure::start("time"); for _ in 0..iterations { assert!(ticks[..num_entries] .verify_cpu_x86_simd(&start_hash, 16) .finish_verify(&ticks[..num_entries])); } time.stop(); println!( "{},cpu_simd_avx512,{}", num_entries, time.as_us() / iterations as u64 ); } if perf_libs::api().is_some() { let mut time = Measure::start("time"); let recyclers = VerifyRecyclers::default(); for _ in 0..iterations { assert!(ticks[..num_entries] .start_verify(&start_hash, recyclers.clone(), true) .finish_verify(&ticks[..num_entries])); } time.stop(); println!( "{},gpu_cuda,{}", num_entries, time.as_us() / iterations as u64 ); } println!(); num_entries *= 2; } }