Merge branch 'app' into 'master'

App See merge request devops/docs.hpc.sjtu.edu.cn!57
SJTU-HPC · Jul 12, 2021 · b54491e · b54491e
2 parents f68ec9b + ecb8f02
commit b54491e
Show file tree

Hide file tree

Showing 5 changed files with 61 additions and 40 deletions.
diff --git a/docs/app/hyper-mpi.rst b/docs/app/hyper-mpi.rst
@@ -0,0 +1,25 @@
+*********
+Hyper-MPI
+*********
+
+.. TODO: Guangchao
+
+Hyper-MPI是
+
+Hyper-MPI使用方式
+-----------------
+
+-  首先一定要用 \ ``ssh``\ 登录ARM节点
+
+.. code:: bash
+
+   $ ssh -p 18022 username@202.120.58.248
+
+-  使用 \ ``module``\ 导入应用命令
+
+.. code:: bash
+
+   $ module load hmpi/4.0.3-gcc-9.3.0
+
+参考资料
+========
diff --git a/docs/app/index.rst b/docs/app/index.rst
@@ -22,13 +22,15 @@
 
 MPI库
 
-+-----------+------------------------------+----------+-------------+
-| 软件      | 描述                         | 可用版本 | 平台        |
-+===========+==============================+==========+=============+
-| OpenMPI   | OpenMPI是一个开源MPI实现     | 3.1.5    | |cpu| |arm| |
-+-----------+------------------------------+----------+-------------+
-| Intel MPI | Intel MPI是一个高性能MPI实现 | 2019.4   | |cpu|       |
-+-----------+------------------------------+----------+-------------+
++------------------+------------------------------+----------+-------------+
+| 软件             | 描述                         | 可用版本 | 平台        |
++==================+==============================+==========+=============+
+| OpenMPI          | OpenMPI是一个开源MPI实现     | 3.1.5    | |cpu| |arm| |
++------------------+------------------------------+----------+-------------+
+| Intel MPI        | Intel MPI是一个高性能MPI实现 | 2019.4   | |cpu|       |
++------------------+------------------------------+----------+-------------+
+| :doc:`hyper-mpi` |                              |          | |arm|       |
++------------------+------------------------------+----------+-------------+
 
 基准测试
 
@@ -299,6 +301,7 @@ AI 计算软件
    gsl
    eigen
    intel
+   hyper-mpi
    keras
    lammps
    mrbayes

diff --git a/docs/job/jobsample1.rst b/docs/job/jobsample1.rst
@@ -39,7 +39,7 @@
 π 集群队列介绍
 --------------
 
-π 集群上现有 small, cpu, huge, 192c6t 和 dgx2 队列。
+π 集群上现有 small, cpu, huge, 192c6t, dgx2 和 arm128c256g 队列。
 
 ``scontrol show partition`` 查看集群队列介绍
 
@@ -81,6 +81,11 @@
 |               | 96 核，1.45T 内存，16>块 32G      |
 |               | 显存的 V100卡                     |
 +---------------+-----------------------------------+
+| arm128c256g   |                                   |
+|               |                                   |
+|               |                                   |
+|               |                                   |
++---------------+-----------------------------------+
 
 small, cpu, dgx2 队列允许的作业运行最长时间为 7 天。huge 和 192c6t 为 2天。
 
@@ -207,7 +212,25 @@ dgx2 队列 slurm 脚本示例：单节点，分配 2 块 GPU，GPU:CPU 配比 1
    #SBATCH --output=%j.out 
    #SBATCH --error=%j.err
 
+arm128c256g
+~~~~~~~~~~~
+
+arm128c256g 队列 slurm 脚本示例：单节点60核
+
+.. code:: bash
+
+    #!/bin/bash
+
+    #SBATCH --job-name=test
+    #SBATCH --partition=arm128c256g
+    #SBATCH -N 1
+    #SBATCH --ntasks-per-node=60
+    #SBATCH --output=%j.out
+    #SBATCH --error=%j.err
+
+    source /lustre/share/singularity/commercial-app/vasp/activate arm
 
+    mpirun -n $SLURM_NTASKS vasp_std
 
 常用软件作业示例
 ----------------

diff --git a/docs/system/filesystem.rst b/docs/system/filesystem.rst
@@ -128,36 +128,4 @@ Lustre最常用于高性能计算HPC，世界超级计算机TOP 10中的70%、TO
     --VALIDATION_STRINGENCY LENIENT \
     --CREATE_INDEX true
 
-    gatk BaseRecalibrator \
-    -R ${REFDIR}/ucsc.hg19.fasta \
-    --tmp-dir ${TMPDIR} \
-    -L ${REFDIR}/hg19.interval_list.bed \
-    --known-sites ${REFDIR}/1000G_phase3_v4_20130502.hg19.sites.indels.vcf \
-    --known-sites ${REFDIR}/Mills_and_1000G_gold_standard.indels.hg19.sites.vcf \
-    -I ${SampleID}_mem.sorted.reorder.rmdup.bam \
-    -O ${SampleID}_mem.sorted.reorder.rmdup.pre.table
-
-    gatk ApplyBQSR \
-    -R ${REFDIR}/ucsc.hg19.fasta \
-    --tmp-dir ${TMPDIR} \
-    -L ${REFDIR}/hg19.interval_list.bed \
-    -I ${SampleID}_mem.sorted.reorder.rmdup.bam \
-    -bqsr ${SampleID}_mem.sorted.reorder.rmdup.pre.table \
-    -O ${SampleID}_mem.sorted.reorder.rmdup.bqsr.bam
-
-    gatk BaseRecalibrator \
-    -R ${REFDIR}/ucsc.hg19.fasta \
-    --tmp-dir ${TMPDIR} \
-    --known-sites ${REFDIR}/1000G_phase3_v4_20130502.hg19.sites.indels.vcf \
-    --known-sites ${REFDIR}/Mills_and_1000G_gold_standard.indels.hg19.sites.vcf \
-    -I ${SampleID}_mem.sorted.reorder.rmdup.bqsr.bam \
-    -O ${SampleID}_mem.sorted.reorder.rmdup.post.table
-
-    gatk PrintReads \
-    -R ${REFDIR}/ucsc.hg19.fasta \
-    --tmp-dir ${TMPDIR} \
-    -L ${REFDIR}/hg19.interval_list.bed \
-    -I ${SampleID}_mem.sorted.reorder.rmdup.bqsr.bam \
-    -O ${SampleID}_mem.sorted.reorder.rmdup.recal.grp.bam
-
 过程中，会产生许多中间文件和临时文件。因此，可利用 $SCRATCH 作为临时目录，加快分析过程。只需要把脚本中的 ``WORKDIR=$HOME/WES_TEST`` 修改为 ``WORKDIR=$SCRATCH/WES_TEST`` 即可。
diff --git a/requirements.txt b/requirements.txt
@@ -17,4 +17,6 @@ sphinx>=2.0
 sphinx_markdown_tables
 sphinx_material
 Sphinx-Substitution-Extensions
+sphinxcontrib-serializinghtml
+sphinxcontrib.htmlhelp
 jieba