Community Project Proposal!

[VicBar]

So I have been getting into Data Analysis and to practice I decided something cool to study with lots of variables and order and plot, etc would be CPUs!

Remember this video?:

Well I decided to grab the spreadsheet Linus made and get to work on it but a few problems arose when I was done with it..

First of all it is quite out of date! It's 2017 and Ryzen is here. Today we started seeing the benchmarks of the 1800x and soon there will be more R7s and then the R5 and R3..

Wouldn't you like to see how these are compared to other existing processors in terms of cores, threads, price, TDP, etc?

So you can get idea what I got check out this nifty little app i got going: https://vicbar.shinyapps.io/shiny_benchmarks/

The idea is to have ALL that info in a very explorable way with the power of the visualization power of the R programming language!

 

The community project part starts here. I have the app and am actively developing it further. Its pretty bare bones still.. But most importantly the amount of info is very very small. The video above covers practically only Skylake and Broadwell CPUs, so I was thinking we could crowdsource a databank so I can we can keep up to date with this awesome time for the CPU market and have something that is ours.

 

So that is what I propose. That you guys donate some secs of free time to fill this .csv file which I can use to update app quickly! :https://docs.google.com/spreadsheets/d/1C7OkqOMGur-9S__0Vsp8_k5ABcMcu645PA0Pm8kbhbc/edit?usp=sharing

 

Check the app out, I find it pretty interesting to browse a data bank of CPUs like that, its great for some quick and easy visual comparison and benchmarking (prices and characteristics).

 

Now I'll talk about the future.

With this spreadsheet Linus also gives on about GPU overclocks, I haven't gotten to that one yet but I plan to, maybe.
What I really want is to have a comparison about the different sources we as tech enthusiasts get our info. i.e. Have one app where we can consult and visualize the different results that our favourite youtubers get from their benchmarking. I know it would be hard and maybe useless because of all the different methodologies they use, but I can help us get an idea, and a wide picture of what Ryzen (or any other future hardware) has to offer. (looking at 180ti).

So concluding, I am going to go on analyzing and playing around with data, I hope some LTT forum members come along for the ride and help in having our own data bank. Also if this goes well it can turn into larger and more interesting projects.

 

TL;DR

I'm doing a project and I could use a hand or two on gathering data on all CPUs. The columns I'm going to use are these, https://docs.google.com/spreadsheets/d/1C7OkqOMGur-9S__0Vsp8_k5ABcMcu645PA0Pm8kbhbc/edit?usp=sharing Feel free to edit!

I'm doing so to make this small app more useful and I make this project available to the LTT Community: https://vicbar.shinyapps.io/shiny_benchmarks/

 

 

[Guest]

Are the price chart their MSRP or price taken from Amazon? Cause I see loads of issue with Amazon prices 

[Guest]

Also. X99 cpu has no igpu. Not that their igpu info was missing or something so NA might not be appropriate 

[nicdepro]

This is an awesome idea!! I can add older CPU's to the spreadsheet right? 

[VicBar]

9 minutes ago, nicdepro said:

This is an awesome idea!! I can add older CPU's to the spreadsheet right? 

Yes, Please do so!!

[nicdepro]

1 minute ago, VicBar said:

Yes, Please do so!!

Awesome!! Thanks

[VicBar]

14 minutes ago, deXxterlab97 said:

Are the price chart their MSRP or price taken from Amazon? Cause I see loads of issue with Amazon prices 

I was going to use the mean from the Price.Low and Price.High and Amazon.Price from the original spreadsheet but it was too incomplete... I ended up using the Amazon Price, as you say. It is janky as hell but also no-one really respects the MSPR... It is a really good base line, but it was not in the original document so I went with Amazon.

[Guest]

Just now, VicBar said:

I was going to use the mean from the Price.Low and Price.High and Amazon.Price from the original spreadsheet but it was too incomplete... I ended up using the Amazon Price, as you say. It is janky as hell but also no-one really respects the MSPR... It is a really good base line, but it was not in the original document so I went with Amazon.

Well it's good for new cpu. older cpu could be rare and thus jack up the price. 

[Pattycake]

I just spent a chunk of time adding missing info and cleaning up the formating. It looks pretty good to me, but someone can go through and just make sure things aren't off. 

I've never seen or used shiny before, but could you show the code you used? 

[VicBar]

On 3/3/2017 at 8:51 PM, Pattycake said:

I just spent a chunk of time adding missing info and cleaning up the formating. It looks pretty good to me, but someone can go through and just make sure things aren't off. 

I've never seen or used shiny before, but could you show the code you used? 

I might even put it up on github if enough people are interested.. Shiny has a TON of tutorials and guides for its use, Rstudio has really impressed me: https://shiny.rstudio.com/

My code is messy af.. because.. well since im just constantly working it, last versions of it tend to be "What just worked". When I get home, (a work atm), I'll share some screen captures, or source if you feel like running it locally.

Also thank you very much for your contribution! 

On 3/3/2017 at 8:51 PM, Pattycake said:

I just spent a chunk of time adding missing info and cleaning up the formating. It looks pretty good to me, but someone can go through and just make sure things aren't off. 

I've never seen or used shiny before, but could you show the code you used? 

Sorry having taken so long to respond.. This little "project" didn't get the traction I had hoped for... I've been busy.. Anyway below I share an extract that shows you the inner workings of the shiny framework!

#
# This is a Shiny web application. You can run the application by clicking
# the 'Run App' button above.
#
# Find out more about building applications with Shiny here:
#
#    http://shiny.rstudio.com/
#
locale <- Sys.setlocale(category = "LC_ALL", locale = "C")
source('data-prep.R')
library(shiny)
library(ggplot2)

# Define UI for application that draws a histogram
ui <- fluidPage(
   
   # Application title
   headerPanel("LTT CPUs Spreadsheet - made interactive plot"),
   
   # Sidebar with a slider input for number of bins 
   sidebarLayout(
      sidebarPanel(
         sliderInput("range",
                    "Price Range (USD):",
                    min = 0,
                    max = 8000,
         						step = 100,
                    value = c(100,1000)),
         selectInput("yaxis", "Y-axis",
         						colnames(cpus[numerics]),
         						selected = "Base.Clock"),
         selectInput("xaxis", "X-axis",
         						colnames(cpus[numerics]),
         						selected = "Price"),
         selectInput("color", "Colour",
         						colnames(cpus[non_numerics]),
         						selected = "Architecture"),
         selectInput("size", "Size",
         						colnames(cpus[numerics]),
         						selected = "Cores")
      ),     
      mainPanel(
         plotOutput("scattPlot", click = "plot_click", dblclick = "plot_dblclick", hover = "plot_hover"),
         tableOutput("hover_info")
      )
   ),
      wellPanel(
        tableOutput("compare1"),
      	verbatimTextOutput("verb")
      ),
	   	wellPanel(
	   		tableOutput("compare2")
	   )
)


# Define server logic required to draw a histogram
server <- function(input, output) {
	
	selectedData <- reactive ({
		subset(cpus, input$range[1]<Price & Price<input$range[2])
	})
	
	output$scattPlot <- renderPlot({
			p <- ggplot(data=selectedData(),
									aes(x=sapply(selectedData()[input$xaxis], as.numeric),
											y=sapply(selectedData()[input$yaxis], as.numeric),
											color=sapply(selectedData()[input$color], as.factor),
											size=sapply(selectedData()[input$size], as.numeric))) +
				labs(x = input$xaxis, y = input$yaxis, size=input$size, color=input$color)
			
			p + geom_point() 
			
	})
		
	output$compare1 <- renderTable({
		nearPoints(selectedData(), input$plot_click,
							 xvar = input$xaxis,
							 yvar = input$yaxis,
							 threshold = 10, maxpoints = 1)
	}, spacing = "s", align = 'l', width = "90%", caption = "single-click to compare")
  		
	output$compare2 <- renderTable({
		nearPoints(selectedData(),
							 input$plot_dblclick,
							 xvar = input$xaxis,
							 yvar = input$yaxis,
							 threshold = 10, maxpoints = 1)
	}, spacing = "s", align = 'l', width = "90%", caption = "double-click to compare") 
	
	output$verb <- renderText({
		#print(str(selectedData()))
	})
		
	output$hover_info <- renderTable({
		nearPoints(selectedData()[c(-3,-4,-6,-7,-8,-9,-10,-11)],
							 input$plot_hover,
							 xvar = input$xaxis,
							 yvar = input$yaxis,
							 threshold = 5, maxpoints = 1)
	}, spacing = "s", align = 'l', width = "80%", caption = "hover-over info") 
}

# Run the application 
shinyApp(ui = ui, server = server)