2019-09-03

Moore's Law

This week's data is from Wikipedia - by way of the Data Is Beautiful Subreddit.

Additional info and graphics can be found at Our World in Data.

Moore's Law: Transistors per microprocessor

• The observation that the number of transistors in a dense integrated circuit doubles approximately every two years.

Get the data!

cpu <- readr::read_csv("https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2019/2019-09-03/cpu.csv")

gpu <- readr::read_csv("https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2019/2019-09-03/gpu.csv")

ram <- readr::read_csv("https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2019/2019-09-03/ram.csv")

Data Dictionary

cpu.csv

variable	class	description
processor	character	Processor name
transistor_count	double	Number of transitors
date_of_introduction	double	year introduced
designer	character	Designer
process	double	Size of manufacturing process (in nanometers)
area	double	Area of chip in square millimeters

gpu.csv

variable	class	description
processor	character	Processor name
transistor_count	double	Transistor count
date_of_introduction	double	year introduced
designer_s	character	designer
manufacturer_s	character	manufacturer
process	double	size of manufacturing process (nanometers)
area	double	area of chip in square millimeters
ref	character	reference

ram.csv

variable	class	description
chip_name	character	Chip name
capacity_bits	character	capacity bits - essentially how many units of information it can work on, units in next column
bit_units	character	Units for bit capacity (bits < kb < Mb < Gb)
ram_type	character	Ram type
transistor_count	double	Transistor count
date_of_introduction	double	Year introduced
manufacturer_s	character	Manufactured
process	double	Size of manufacturing process (nanometers)
area	double	Area of chip in square millimeters
ref	character	reference

Cleaning Script

library(tidyverse)
library(rvest)

url <- "https://en.wikipedia.org/wiki/Transistor_count"

tables <- url %>% 
  read_html() %>% 
  html_table(fill = TRUE)

df1 <- tables %>% chuck(1) %>% 
  janitor::clean_names() %>% 
  as_tibble()

df1_clean <- df1 %>% 
  mutate(
    # transistor_count = gsub("\\[[^\\]]*\\]", "", transistor_count, perl=TRUE),
    transistor_count = str_remove(transistor_count, "\\[[^\\]]*\\]"),
    transistor_count = str_remove(transistor_count, "[:punct:]+"),
    transistor_count = str_remove(transistor_count, "\\[[^\\]]*\\]"),
    transistor_count = str_remove(transistor_count, "[:punct:]+"),
    transistor_count = str_remove(transistor_count, "[:punct:]+"),
    transistor_count = str_extract(transistor_count, "[:digit:]+"),
    date_of_introduction = str_sub(date_of_introduction, 1, 4),
    process = str_remove(process, ","),
    process = str_extract(process, "[:digit:]+"),
    area = str_extract(area, "[:digit:]+")
    ) %>% 
  mutate_at(.vars = vars(transistor_count:date_of_introduction, process:area), as.double)


df1_clean %>%
  mutate() 
df2 <- tables %>% chuck(2) %>% 
  janitor::clean_names() %>% 
  as_tibble()

df2_clean <- df2 %>% 
  mutate(
    # transistor_count = gsub("\\[[^\\]]*\\]", "", transistor_count, perl=TRUE),
    transistor_count = str_remove(transistor_count, "\\[[^\\]]*\\]"),
    transistor_count = str_remove(transistor_count, "[:punct:]+"),
    transistor_count = str_remove(transistor_count, "\\[[^\\]]*\\]"),
    transistor_count = str_remove(transistor_count, "[:punct:]+"),
    transistor_count = str_remove(transistor_count, "[:punct:]+"),
    transistor_count = str_extract(transistor_count, "[:digit:]+"),
    process = str_remove(process, ","),
    process = str_extract(process, "[:digit:]+"),
    area = str_extract(area, "[:digit:]+")
  ) %>% 
  mutate_at(.vars = vars(transistor_count:date_of_introduction, process:area), as.double)

df3 <- tables %>% chuck(4) %>% 
  janitor::clean_names() %>% 
  as_tibble()

df3

df3_clean <- df3 %>% 
  mutate(
    # transistor_count = gsub("\\[[^\\]]*\\]", "", transistor_count, perl=TRUE),
    transistor_count = str_remove(transistor_count, "\\[[^\\]]*\\]"),
    transistor_count = str_remove(transistor_count, "[:punct:]+"),
    transistor_count = str_remove(transistor_count, "\\[[^\\]]*\\]"),
    transistor_count = str_remove(transistor_count, "[:punct:]+"),
    transistor_count = str_remove(transistor_count, "[:punct:]+"),
    transistor_count = str_extract(transistor_count, "[:digit:]+"),
    date_of_introduction = if_else(
      str_length(date_of_introduction) >= 5,
      str_sub(date_of_introduction, -4),
      str_sub(date_of_introduction, 1, 4)),
    process = str_remove(process, ","),
    process = str_extract(process, "[:digit:]+"),
    area = str_extract(area, "[:digit:]+"),
    bit_units = case_when(
      str_detect(capacity_bits, "bit") ~ "Bits",
      str_detect(capacity_bits, "kb") ~ "kb",
      str_detect(capacity_bits, "Mb") ~ "Mb",
      str_detect(capacity_bits, "Gb") ~ "Gb",
      TRUE ~ ""
                 )
  ) %>% 
  mutate_at(.vars = vars(transistor_count:date_of_introduction, process:area), as.double) %>% 
  select(chip_name, capacity_bits, bit_units, everything()) %>% 
  mutate(capacity_bits = str_extract(capacity_bits, "[:digit:]+"))

df3_clean

write_csv(df1_clean, here::here("2019", "2019-09-03", "cpu.csv"))
write_csv(df2_clean, here::here("2019", "2019-09-03", "gpu.csv"))
write_csv(df3_clean, here::here("2019", "2019-09-03", "ram.csv"))