Pierluca D'Oro proceduralia

System Design Cheatsheet

Picking the right architecture = Picking the right battles + Managing trade-offs

Basic Steps

Clarify and agree on the scope of the system

User cases (description of sequences of events that, taken together, lead to a system doing something useful)
- Who is going to use it?
- How are they going to use it?

In this article, I will share some of my experience on installing NVIDIA driver and CUDA on Linux OS. Here I mainly use Ubuntu as example. Comments for CentOS/Fedora are also provided as much as I can.

Installing the NVIDIA driver, CUDA and cuDNN on Linux (Ubuntu 20.04)

This is a companion piece to my instructions on building TensorFlow from source. In particular, the aim is to install the following pieces of software

NVIDIA graphics card driver (v450.57)
CUDA (v11.0.2)
cuDNN (v8.0.2.39)

on an Ubuntu Linux system, in particular Ubuntu 20.04.

	Latency Comparison Numbers (~2012)
	----------------------------------
	L1 cache reference 0.5 ns
	Branch mispredict 5 ns
	L2 cache reference 7 ns 14x L1 cache
	Mutex lock/unlock 25 ns
	Main memory reference 100 ns 20x L2 cache, 200x L1 cache
	Compress 1K bytes with Zippy 3,000 ns 3 us
	Send 1K bytes over 1 Gbps network 10,000 ns 10 us
	Read 4K randomly from SSD* 150,000 ns 150 us ~1GB/sec SSD

	pager.c:344 pline("Cannot open data file!")
	pager.c:467 pline("? Seek error on 'data' file!")
	pager.c:482 pline("I don't have any information on those things.")
	pager.c:578 pline("Please move the cursor to %s.", what_is_an_unknown_object)
	pager.c:581 pline("Pick an object.")
	pager.c:815 pline("%s", out_str)
	pager.c:825 pline("I've never heard of such things.")
	pager.c:865 pline("That is %s%s%s.", an(defsyms[trap_to_defsym(tt)].explanation), !trap->madeby_u ? "" : (tt == WEB) ? " woven" : <x> <x> <x> (tt == HOLE \|\| tt == PIT) ? " dug" : " set", !trap->madeby_u ? "" : " by you")
	pager.c:876 pline("I can't see a trap there.")
	pager.c:891 pline("Cannot open data file!")

	from keras.models import Sequential
	from keras.layers import Dense
	from keras.utils.io_utils import HDF5Matrix
	import numpy as np

	def create_dataset():
	import h5py
	X = np.random.randn(200,10).astype('float32')
	y = np.random.randint(0, 2, size=(200,1))
	f = h5py.File('test.h5', 'w')

	def get_jacobian(net, x, noutputs):
	x = x.squeeze()
	n = x.size()[0]
	x = x.repeat(noutputs, 1)
	x.requires_grad_(True)
	y = net(x)
	y.backward(torch.eye(noutputs))
	return x.grad.data

	import jax
	from jax import core
	from graphviz import Digraph
	import itertools


	styles = {
	'const': dict(style='filled', color='goldenrod1'),
	'invar': dict(color='mediumspringgreen', style='filled'),
	'outvar': dict(style='filled,dashed', fillcolor='indianred1', color='black'),

	llmc() {
	local system_prompt='Output a command that I can run in a ZSH terminal on macOS to accomplish the following task. Try to make the command self-documenting, using the long version of flags where possible. Output the command first enclosed in a "```zsh" codeblock followed by a concise explanation of how it accomplishes it.'
	local temp_file=$(mktemp)
	local capturing=true
	local command_buffer=""
	local first_line=true
	local cleaned_up=false # Flag to indicate whether cleanup has been run

	cleanup() {
	# Only run cleanup if it hasn't been done yet

Pierluca D'Oro proceduralia

System Design Cheatsheet

Basic Steps

Table of Contents

Installing the NVIDIA driver, CUDA and cuDNN on Linux (Ubuntu 20.04)