Here is another way to look at it.
Consider the output from my machine:
Code:
[ddenardo@cylon ~]$ free -m
total used free shared buffers cached
Mem: 3964 3261 702 0 48 2158
-/+ buffers/cache: 1054 2909
Swap: 5951 0 5951
One line at a time (plus the header):
Code:
total used free shared buffers cached
Mem: 3964 3261 702 0 48 2158
I have 3964 megs of total RAM (~ 4Gigs), 3261M is "used", leaving 702M "free". The "shared" memory column is obsolete -- ignore it. The last two numbers say that I have 48M of kernel buffers, and 2158M of "Cached" stuff.
This looks like my RAM is more than 80% "full" (only 700M out of 4G of RAM is "free"). This isn't really true, and the next line explains that.
Code:
total used free shared buffers cached
-/+ buffers/cache: 1054 2909
That says that if you factor out the buffers and cache, I really have 1054M "used" and 2909M "free" -- so I'm really only using about 25% of my RAM at this point.
The last line (the one that starts with "Swap:") just shows the total amount of swap space available and the amount of swap space used.
... so you may wonder why the first line said that 80% of my RAM was being used? Well, the kernel realizes that I have gobs and gobs of unused RAM, so it does things like buffer disk reads -- which makes the computer faster. If an application needs that RAM that it is using for buffers, the kernel will discard the buffer and give the application the memory it asked for.
In your case, it looks like you only have 49M available, but in reality you have 1380M available.
Does that make sense?