Trevor Keller tkphd

Bottom line: use du --apparent-size to compare information content (independent of filesystems), rather than disk utilization (block-size dependent).

                *Distributed Disk Utilization*

BeeGFS	Items	Apparent Size	per Item	On Disk	per Item
`hsf_0/`	176893	162.0 MB	0.94 KB	100.2 MB	0.58 KB
`hsf_1/`	165189	152.5 MB	0.94 KB	93.9 MB	0.58 KB

Compute Node Comparison

Trevor Keller, Ph.D. • NIST MML MSED • May 15, 2018

Description

This document summarizes the side-by-side comparison of dual-socket servers:

2× [AMD EPYC 7601][_epyc] vs. 2× [Intel Xeon E5-2697Av4][_xeon].

The AMD system is a testbed, access to which was provided by [KOI Computers][_koic].

	#!/usr/bin/python3
	# -- coding: utf-8 --

	about = """
	A random card generator for The Quiet Year, which ends when 🂮 is drawn.
	"""

	fleet = """
	short game: remove 5 cards from each suit; drop 🃎 but keep 🂮
	"""

	#!/usr/bin/env python3

	import numpy as np
	import matplotlib.pyplot as plt
	from numpy.random import default_rng
	rng = default_rng()

	n = int(100_000)
	N = int(1_000_000_000)
	pop = int(983) # children born in Kiev

	# Ross is a programmer for Megabank, working on a new smartphone app that
	# helps users transfer money between accounts. He's relatively new to R,
	# the language preferred by the top brass, and is not entirely sure of
	# the difference between all the operators available for assigning
	# values to variables. Specifically, there's this "scoped assignment" thing
	# that sure looks like a harpoon to Ross. What's it for?

	# Ross opens a bank account, then uses the app to perform a series of
	# transactions. Because he's no fool, he does so in the lobby of a
	# Megabank branch, and the bank teller informs him after each step what

	HPLinpack benchmark input file
	Innovative Computing Laboratory, University of Tennessee
	HPL.out output file name (if any)
	6 device out (6=stdout,7=stderr,file)
	1 # of problems sizes (N)
	17400 Ns
	1 # of NBs
	128 NBs
	0 PMAP process mapping (0=Row-,1=Column-major)
	1 # of process grids (P x Q)

	HPLinpack benchmark input file
	Innovative Computing Laboratory, University of Tennessee
	HPL.out output file name (if any)
	6 device out (6=stdout,7=stderr,file)
	1 # of problems sizes (N)
	5040 Ns
	1 # of NBs
	128 NBs
	0 PMAP process mapping (0=Row-,1=Column-major)
	1 # of process grids (P x Q)

	/* vector_addition.cu
	* Available as a Gist at https://tinyurl.com/y34nfxgc
	*/

	#include <stdio.h>
	#include <cuda.h>

	#define N 16

	__global__ void vector_addition(int* d_u, int* d_v, int* d_w)

	% Modifications for @b0rk's excellent Bite Size Networking
	% to get a proper 2-sided booklet
	\documentclass[12pt]{article}
	\usepackage{graphicx,pdfpages}
	\usepackage[paperheight=11in,paperwidth=8.5in,margin=0in]{geometry}
	\pagestyle{empty}
	\begin{document}
	\begin{figure}\centering
	\vskip-1cm
	\includegraphics[page=24,width=8.25in,angle=180]{bite-size-networking.pdf}