---
title: The Basics of High Performance Computing in Ecology
subtitle: HPC not for dummies, but for ecologists who just haven't been exposed to it yet
description: New Post Description
date: today
categories: [workflow]
date-modified: last-modified
draft: true
bibliography: references.bib
---

# Defining some key ideas

High performance computing is the ability to process data and perform complex calculations at high speed. This is often achieved via super computers, which contain thousands of compute notes

## What is a cluster?

a large computer composed of a collection of many smaller separate servers (computers) called nodes

## What is a node?

## distinct servers, connected to one another with a fast interconnect so data can be passed between them very quickly

## What is a core?

## What is CPU?

Central Processing Unit - primary component of a computer, acts as a "control center"

## What is Derecho?

NCAR-Wyoming Supercomputing Center

All located in cheyenne not Boulder, began in 2012

home to NWSC-3 Derecho Super computer

cheyenne just retired, 4032 nodes

derecho has more nodes but also more cores per node

3.5x performance compared to cheyenne

Derecho has different types of nodes

\- CPU-only Compute Nodes (2488 total, 318,464 total cores)

\- GPU Compute nodes (82 total)

\- Login nodes (8 total)

## What is Casper?

casper = data and viusualization system

more heterogeneous node structure - not all have same tech on each node

## Key Differences between Casper and Derecho:

Derecho - run many ilands all at once

casper - can run on smaller nodes all at once

need to clarify which node you want to request

Think of derecho as something you use for big climate models

if you compile something on derecho, it may not compile on casper and vice versa

## How do you access Derecho / Casper?

Submitting a Derecho allocation

small and large allocations

will it matter that I'm not at a university

sign in with hayesk\@ucar.edu

need to submit ticket with help desk if you need to reset password \[still need to reset mine\]

SAM = Systems accounting Manager

\- user settings

\- tells you use history

can change default shell

software tools on CISL for use inside HPC environment

\- compilers (intel)

\- 10 libraries (NetCDF, PNetCDF, HDF5

analysis (r, python, julia)

etc

need to know:

\- how many nodes am i going to use

Some of my favorite resources on Ecological coding

-   [@wilson2014]
