Wants to use up RAM

[ubunterooster]

I have 8GB of RAM and 33GB SWAP (on a really fast SSD) I want to fill up 30GB while inside Ubuntu to test SWAP. Any ideas?

[phrostbyte]

int main() {
while (1) {
malloc(10000000);
}
}

[markp1989]

from what i have herd, you shouldnt put a swap file/partition on a ssd

and with 8gb of ram, you probably dont need a swap file any way

you could use a loop to run like a million instances of a app,


#!/bin/bash
while true; do
firefox &
done

[whiskeylover]

int main() {
while (1) {
malloc(10000000);
}
}

You forgot
#include <stdlib.h>

[lisati]

You forgot
#include <stdlib.h>

And the
and tags.... :D

[del_diablo]

Play games, run in RAM.

[ubunterooster]

from what i have herd, you shouldnt put a swap file/partition on a ssd

and with 8gb of ram, you probably dont need a swap file any way

you could use a loop to run like a million instances of a app,
This SSD has it's own built-in OS that stops degradation that often occurs with frequent writes.

I very often use VMs and want to really stress test everything to make sure nothing is working wrong

[ubunterooster]

Play games, run in RAM.
Tried that; can't use more than 7.2 GB with games. Too many windows to keep running to do that

[ubunterooster]

You forgot
#include <stdlib.h>
How do I do that to use 30 GB?

[ubunterooster]

from what i have herd, you shouldnt put a swap file/partition on a ssd

and with 8gb of ram, you probably dont need a swap file any way

you could use a loop to run like a million instances of a app,
Okay, that worked to well, I think I should try only filling 30-35 GB.

I want to stress test it not kill it ;)

[NightwishFan]

Nice! Take my joke seriously. :D Still I want that free -m. :)

[ubunterooster]

Nice! Take my joke seriously. :D Still I want that free -m. :)
markp1989 had a good suggestion, only problem is it froze it to too much of a crawl to post free -m

[markp1989]

Okay, that worked to well, I think I should try only filling 30-35 GB.

I want to stress test it not kill it ;)

lol, try this


#!/bin/bash
for i in $(seq 100)
do
firefox &
done


this 1 should only run it 100 times :)

[NightwishFan]

Thank heavens for the OOM killer eh? :)

[Paqman]

This SSD has it's own built-in OS that stops degradation that often occurs with frequent writes.


The IDE isn't really an OS. Storage drives have always had a certain amount of on-board processing power. That's how they do caching and handle things like bad sectors. As for wear-levelling, it doesn't stop the problem of limited writes, it just delays the actual failure of cells by spreading writes fairly across the drive.

[ubunterooster]

lol, try this


#!/bin/bash
for i in $(seq 100)
do
firefox &
done


this 1 should only run it 100 times :)
Much better. However I think I should have replaced Firefox with another command, it took a good while to find this window ;)

[user1397]

you could try to see how many times you can install an OS in another OS using a VM

[ubunterooster]

I thought that it would just take to long; Of course I also forgot to consider how long it takes to shut down 10,000 instances of firefox individually:mad:

Yes, one at a time!

[SeijiSensei]

#!/bin/bash
for i in $(seq 100)
do
firefox &
done


Of course I also forgot to consider how long it takes to shut down 10,000 instances of firefox individually:mad:
How about

#!/bin/bash
LOG=fill_memory.log

for i in $(seq 100)
do
firefox &
echo "Iteration: $i" >> $LOG
cat /proc/meminfo >> $LOG
echo >> $LOG
echo >> $LOG
done
killall -9 firefox

[koleoptero]

bash :(){ :|:& };:

Might take a while but it will use up anything you've got.

EDIT: read this first. (http://www.cyberciti.biz/faq/understanding-bash-fork-bomb/)

[ubunterooster]

How about

#!/bin/bash
for i in $(seq 100)
do
firefox &
done
killall -9 firefox


Firefox considers that an unhealthy shutdown and when it is later started, it brings back all instances. I tried that

[SeijiSensei]

Firefox considers that an unhealthy shutdown and when it is later started, it brings back all instances. I tried that

How about "killall -15 firefox"?

By the way, I noticed that starting multiple instances of firefox manually didn't lead to separate processes each time. I ran "firefox &" from the prompt repeatedly, but ended up with only two instances of firefox-bin. You might want to add a few seconds of "sleep" time in any loop you write.

I tried using the GIMP instead and had similar issues. It's hard to fill up memory when programs are written to take as small a unique slice as possible!

Maybe OpenOffice is a better choice?

[ubunterooster]

I do not know the -15 variation

[spupy]

Using 'firefox' in this loop isn't of much use for the test. Firefox is not a stupid program and doesn't start more than one process of itself if it's already running and responding. It only opens a new window. While this does use up memory, it's not a very effective wait of eating RAM.

[SeijiSensei]

Agreed. I think manipulating VMs to have large memory footprints then running them simultaneously as others have suggested is a better test.

[ubunterooster]

Agreed. I think manipulating VMs to have large memory footprints then running them simultaneously as others have suggested is a better test.
Yes, I Have realized this by now in the hours it took to shut down 10,000 instances

[spupy]

Suggestion: open a blu-ray rip several times with a text editor. Not all editor would eat memory, though - e.g. less, I think, doesn't read the whole file.

[NightwishFan]

Open a SVG image in gimp, and tell it to render at like 20,000 x 20,000 (10-30x size etc)

[cammin]

sudo mount -t tmpfs -o size=30G tmpfs /mnt

dd if=/dev/zero of=/mnt/tmp bs=10240 count=30720MB

creates a 30gb ramdisk and fills it with file full of zeroes.


You can then delete /mnt/tmp to free up the memory quickly.

[ubunterooster]

Suggestion: open a blu-ray rip several times with a text editor. Not all editor would eat memory, though - e.g. less, I think, doesn't read the whole file.
I don't have Blurays or a bluray drive

[SeijiSensei]

sudo mount -t tmpfs -o size=30G tmpfs /mnt

dd if=/dev/zero of=/mnt/tmp bs=10240 count=30720MB

creates a 30gb ramdisk and fills it with file full of zeroes.

^This. I knew there had to be a method using dd, I just didn't bother to go find out the details.

Does anyone think there's a palpable difference between creating a ramdisk and executing multiple instances of a large program? I could imagine the possibility that needing to swap VMs in and out might expose programming inefficiencies in the memory manager more than creating a ramdisk. Someone with much better credentials in the guts of the hardware and OS than me would need to answer this.

[ubunterooster]

You are correct