跳转至

应用性能测试

集群主要有2种CPU,arm 128核;intel,64核,型号为 8358P。

hifiasm

基因组组装

https://github.com/chhylp123/hifiasm

hifi+ont+hic 组装

$ module load arm/hifiasm/0.19.9
$ time hifiasm -o ath.asm -t 128 --ul CRR302667.fastq.gz --ul-cut 500 --h1 CRR302669_R1.fastq.gz --h2 CRR302669_R2.fastq.gz CRR302668.fastq.gz
CPU核数运行时间(sec)cputime最大内存(MB)
arm128(整节点)59796614457377027
intel 8358P64(整节点)115400662042865546

hifi 组装

$ module load arm/hifiasm/0.19.9
$ time hifiasm -o ath.asm -t 128  CRR302668.fastq.gz
CPU核数运行时间(sec)cputime最大内存(MB)
arm162849743020241591
intel 8358P162668943005491667

原生 ARM 和 repack 包

$ export REPACK_EXTRA_OPTS='--bind  /path/to/ath_data/ /data'
$ ./repack_ARM/hifiasm-0.19.8-aarch64-conda.rpk -- hifiasm -o ath4.asm -t 16  /data/CRR302668/CRR302668.fastq.gz
软件核数运行时间(sec)cputime最大内存(MB)
hifiasm162849743020241591
hifiasm-0.19.8-aarch64-conda.rpk162858440096391667

BWA

短序列比对

https://github.com/lh3/bwa

$ module load arm/bwa/0.7.18 arm/samtools/1.21
$ bwa mem -t 128 -R '@RG\tID:test146\tPL:illumina\tLB:library\tSM:humen146' hg38.fa ERR194146_1.fastq.gz ERR194146_2.fastq.gz |samtools sort -@ 128 -o $ERR194146_srt.bam"
CPU核数运行运行时间(sec)cputime最大内存(MB)
arm163408251978966291
arm128(整节点)7136602206259437
intel 8358P162899945831666132
intel 8358P64(整节点)10123510047129309

原生 ARM 和 repack 包

$ export REPACK_EXTRA_OPTS='--bind  /path/to/WGS/ /data'
$ /share/home/software/tmp/repack_ARM/bwa-0.7.17-aarch64-openEuler20.03.rpk -- bwa mem -t 128  /data/hg38.fa /data/ERR194146_1.fastq.gz /data/ERR194146_2.fastq.gz > ERR194146.sam
软件核数运行时间(sec)cputime最大内存(MB)
hifiasm64(整节点)6557586457151514
hifiasm-0.19.8-aarch64-conda.rpk64(整节点)6655597742151157

hisat2

RNASeq短序列比对

https://github.com/DaehwanKimLab/hisat2

$ module load  arm/hisat/2.1.0 arm/samtools/1.21
$ time hisat2 -p ${thread} -x genome -1 sample_1.fq.gz -2 sample_2.fq.gz | samtools sort -@${thread} -o sample.sorted.bam
CPU核数运行时间(sec)cputime最大内存(MB)
arm128(整节点)378359723351
arm16371349917031
intel 8358P64(整节点)4992916423153
intel 8358P16288426217070

GATK HaplotypeCaller

变异检测

https://github.com/broadinstitute/gatk

$ module load arm/gatk/4.6.0.0
$ gatk HaplotypeCaller --native-pair-hmm-threads $thread -R hg38.fa -L chr20 -I ERR194146_sort_redup.bam -O ERR194146_chr20_arm.vcf.gz
CPU核数运行时间(sec)cputime最大内存(MB)
arm128(整节点)14458145031938
arm(GKL优化)1281476101241938
intel 8358P64(整节点)1181262891851

vg

泛基因组比对和变异检测等

https://github.com/vgteam/vg

$ module load arm/vg/1.60.0
$ time vg giraffe -p -t $thread -Z ref.giraffe.gbz -d ref.dist -m ref.min -f sampleid_clean_1.fq.gz -f sampleid_clean_2.fq.gz > sampleid.gam
CPU核数运行时间(sec)cputime最大内存(MB)
arm128(整节点)222258899771
arm161243202845313
intel 8358P64(整节点)327196877018
intel 8358P16996162605312

原生 ARM 和 repack 包

软件核数运行时间(sec)cputime最大内存(MB)
vg(1.60.0)128222258899771
vg-1.56.0-aarch64-conda.rpk1281942217911437

STAR

RNASeq 短序列比对

https://github.com/alexdobin/STAR

$ module load  arm/star/2.7.11b arm/samtools/1.21
$ STAR     --genomeDir star_index/      --runThreadN 128    --readFilesIn CRR232282_f1.fastq.gz CRR232282_r2.fastq.gz     --readFilesCommand zcat     --outFileNamePrefix CRR232282     --outSAMtype BAM SortedByCoordinate     --outBAMsortingThreadN $128
CPU核数运行时间(sec)cputime最大内存(MB)
arm128(整节点)219442439745
arm16280417623199
intel 8358P64(整节点)154485230225
intel 8358P16314476823141

megahit

宏基因组组装

https://github.com/voutcn/megahit

$ module load arm/megahit/1.2.9
$ megahit -1 sampleid_clean_R1.fq.gz -2 sampleid_clean_R2.fq.gz -o out  --presets meta-large --no-mercy --kmin-1pass -t 128
CPU核数运行时间(sec)cputime最大内存(MB)
arm128(整节点)48623457526369953
intel 8358P64(整节点)61315336280570090

minimap2

三代长序列比对

https://github.com/lh3/minimap2

$ module load arm/minimap2/2.28  arm/samtools/1.21 
$ minimap2 -ax map-pb -t 128 MH63.fa MH63WGS_all_subreads.fasta.gz |samtools sort -@128 -o MH63WGS_all_subreads.bam
CPU核数运行时间(sec)cputime最大内存(MB)
arm128(整节点)3713222289353
arm1619743165423143
intel 8358P64(整节点)4522526672995
intel 8358P1614082271323402

lammps

分子动力学模拟

https://github.com/lammps/lammps

参考 https://docs.hpc.sjtu.edu.cn/app/engineeringscience/lammps.html

运行时间越短越好

CPU核数运行时间(sec)cputime(sys)最大内存(MB)
arm64(单节点)19812410(169)1012
arm128(单节点)9812083(265)1360
arm128(4节点)19410985(3883)1360
arm256(2节点)1038587(4172)1114
arm512(多节点)965067(3272)642

gromacs

分子动力学模拟

https://github.com/gromacs/gromacs

参考 https://docs.hpc.sjtu.edu.cn/app/engineeringscience/gromacs.html

单位:ns/day,越高越好

核数64(单节点)128(单节点)256(2节点)512
Performance10.16018.74516.256
本文阅读量  次
本站总访问量  次