tpoterba

#!/bin/sh
printf "\nspark.driver.memory=85g\n" >> /etc/spark/conf/spark-defaults.conf

tpoterba

#!/opt/conda/default/bin/python3
import json
import os
import subprocess as sp
import sys
import errno
from subprocess import check_output

assert sys.version_info > (3, 0), sys.version_info

tpoterba

import hail as hl
import pytest

mt = hl.read_matrix_table('joined.mt')
rs = mt.rows()
cs = mt.cols()
es = mt.entries()

"""
Schema of `mt`:

tpoterba

import hail as hl
import pytest

mt = hl.read_matrix_table('joined.mt')
rs = mt.rows()
cs = mt.cols()
es = mt.entries()

"""
Schema of `mt`:

tpoterba

Failed benchmarks in run 1:
    pc_relate_big
Failed benchmarks in run 2:
    pc_relate_big
                                    Benchmark Name      Ratio     Time 1     Time 2
                                    --------------      -----     ------     ------
                 matrix_table_call_stats_star_star     130.0%      7.493      9.743
                  table_aggregate_downsample_dense     121.0%     54.725     66.242
       variant_and_sample_qc_nested_with_filters_4     118.2%     39.312     46.450
                          group_by_collect_per_row     116.4%      3.221      3.748

tpoterba

$ hail-bench compare 0.2.36-release.json 0.2.37-df59d9ba6a79-release-candidate.json
Failed benchmarks in run 1:
    block_matrix_nested_multiply
    pc_relate_big
Failed benchmarks in run 2:
    block_matrix_nested_multiply
    pc_relate_big
                                    Benchmark Name      Ratio     Time 1     Time 2
                                    --------------      -----     ------     ------
                                   union_p100_p100     168.5%     14.295     24.086

tpoterba

Failed benchmarks in run 1:
    block_matrix_nested_multiply
    pc_relate_big
Failed benchmarks in run 2:
    pc_relate_big
    large_range_matrix_table_sum
                                    Benchmark Name      Ratio     Time 1     Time 2
                                    --------------      -----     ------     ------
                  ndarray_matmul_float64_benchmark     286.0%      3.273      9.361
                             read_force_count_p100     127.8%      2.037      2.603

tpoterba

$ hail-bench compare 0.2.33-v33-release.json 0.2.33-4a663d2893e7-v34-release-candidate.json
Failed benchmarks in run 1:
    table_big_aggregate_compile_and_execute
    block_matrix_nested_multiply
    pc_relate_big
    table_range_array_range_force_count
    compile_2k_merge
    table_big_aggregate_compilation
Failed benchmarks in run 2:
    block_matrix_nested_multiply

tpoterba

Found non-overlapping benchmarks:
    matrix_table_scan_count_cols
    matrix_table_scan_count_rows
Failed benchmarks in run 1:
    group_by_collect_per_row
    read_with_index_p1000
    matrix_table_show
    gnomad_coverage_stats_optimized
    group_by_take_rekey
    export_range_matrix_table_entry_field_p100

tpoterba

Table._same: rows differ:

  1L Struct(vep=Struct(assembly_name='GRCh37', allele_string='A/T', ancestral=None, colocated_variants=None, context=None, end=55523002, id='1_55523002_A/T', input='1\t55523002\t.\tA\tT\t.\t.\tGT', intergenic_consequences=None, most_severe_consequence='splice_acceptor_variant', motif_feature_consequences=None, regulatory_feature_consequences=None, seq_region_name='1', start=55523002, strand=1, transcript_consequences=[Struct(allele_num=1, amino_acids=None, biotype='protein_coding', canonical=1, ccds='CCDS603.1', cdna_start=None, cdna_end=None, cds_end=None, cds_start=None, codons=None, consequence_terms=['splice_acceptor_variant'], distance=None, domains=None, exon=None, gene_id='ENSG00000169174', gene_pheno=1, gene_symbol='PCSK9', gene_symbol_source='HGNC', hgnc_id='20001', hgvsc='ENST00000302118.5:c.997-2A>T', hgvsp=None, hgvs_offset=None, impact='HIGH', intron='6/11', lof='LC', lof_flags=None, lof_filter='ANC_ALLELE', minimised=1, polyphen_prediction=None, polyphen_score=None, pr