Ruby takes the doubled amount of RAM it should

Question

I was trying things out with my homeserver and wrote a little ruby program that fills up the RAM by a given amount. But actually I have to halve the amount of bytes I want to put into the RAM. Am I missing something here or is this a bug?

Here the code:

class RAM

  def initialize
    @b = ''
  end

  def fill_ram(size)
    puts 'Choose if you want to set the size in bytes, megabytes or gigabytes.'
    answer = ''
    valid = ['bytes', 'megabytes', 'gigabytes']
    until valid.include?(answer)
      answer = gets.chomp.downcase
      if answer == 'bytes'
        size = size * 0.5
      elsif answer == 'megabytes'
        size = size * 1024 * 1024 * 0.5
      elsif answer == 'gigabytes'
        size = size * 1024 * 1024 * 1024 * 0.5
      else
        puts 'Please choose between bytes, megabytes or gigabyte.'
      end
    end
    size1 = size
    if @b.bytesize != 0
      size1 = size + @b.bytesize
    end
    until @b.bytesize == size1
      @b << '0' * size
    end
    size = 0
  end

  def clear_ram
    exit
  end

  def read_ram
    puts 'At the moment this program fills ' + @b.bytesize.to_s + ' bytes of RAM'
  end

end

Just imagine that the "* 0.5" at each line wouldn't be there.

I did test it in IRB and just created a new RAM object and filled it with 1000 Megabytes of data. In my case it filled the RAM actually with 2000 Megabytes of data, so I did add the times 0.5 to each line, but that can't be the solution.

Maybe from interest: http://stackoverflow.com/questions/7220896/get-current-ruby-process-memory-usage — knut, Apr 19 '14 at 21:30

knut · Accepted Answer · 2014-04-19T22:48:44.580

3

When I run it I get:

Choose if you want to set the size in bytes, megabytes or gigabytes.
bytes
At the moment this program fills 512 bytes of RAM

I think the problem is the missing check for the encoding.

I ran my test in US-ASCII (One character = 1 Byte).

If you run it in UTF-16 you have an explanation for your problem.

Can you try the following code to check your encoding:

p Encoding.default_internal 
p Encoding.default_external

After reading the comment:

The result of your script depends on the parameter of RAM.fill_ram. How do you start your script - and how often do you call RAM.fill_ram?

Please provide the full code.

I called my example with

r = RAM.new
r.fill_ram(1024)
r.read_ram

edited Apr 19 '14 at 22:48

answered Apr 19 '14 at 21:12

knut

27,320
6
84
112

Internal: nil and external: UTF-8; the weird thing is: If you remove the time 0.5 at the upper part (so you get the false program code) the actual used memory is twice as much as @b.bytesize - that is the problem and I don't understand why. – Robert McHardy Apr 19 '14 at 22:18
I took a 2nd look: You never mentioned how you call your script. Please provide the full example to see how you use your class. – knut Apr 19 '14 at 22:49
Edit will come in a minute. – Robert McHardy Apr 19 '14 at 23:55
I call it exactly like you. The program tells me that it uses 1 GiB of the RAM but I check simultaneously with htop and the RAM usage increases by 2 GiB. So basically the program takes twice as much RAM as I want to. This is why I implemented the "* 0.5" in the bytes/megabytes/gigabytes call. There has to be a variable I forgot or something which is exactly as big as @b. – Robert McHardy Apr 20 '14 at 00:11

Ruby takes the doubled amount of RAM it should

1 Answers1