How is it possible to compress so much memory in RAM?

Memory compression is parts of an app, not the entire app. App data in memory often times has lots of zeros. For example, 64-bit pointers often have the first few bytes always zero or some other constant value that is ideal for compression. Partially filled buffers may have zeros in the unused parts. Buffers containing text is highly compressible.

Also, you are seeing megabytes, not gigabytes of compression.

As I said earlier, the really big files tend to be music, pictures, and movies, all of which are already highly compressed, generally using compression models tailored to the kind of data and repeat patterns often found in the media.

There seems to be some confusion about what the compressed memory is - it is application data that hasn't been used recently, and is compressed rather than swapped out by the VM. Without knowing exactly what the application data is (caches, data structures, etc), it will be hard to determine what kind of compression percentages apply. I think in your screenshot it is just coincidental that some of the memory (application memory) and compressed memory (application data) columns are similar - looking at other processes, there are many with no compressed memory, and some with compressed memory much larger than the application memory, so I think it just depends on the application and what it stores in memory.

From the webpage pointer provided above it says compression cuts the size in half. And the compression algorithm is WKdm.

...In Mavericks, the OS has one more option before it has to resort to swapping: compressed memory. Mavericks will find the least-recently-used data in memory and compress it, usually to about half its original size. Et voilà, more free memory.

Like the HFS+ compression feature introduced in Snow Leopard, compressed memory trades relatively abundant CPU cycles for decreased disk I/O. The compression and decompression process is extremely fast, using the WKdm compression algorithm, and it will run in parallel, taking advantage of multiple CPU cores...

My assumption is that lsd is using 269.9MB of memory and that of that memory 268.5MB is compressed. That is to say 258.5MB is compressed code and data, and 1.4MB of that memory is uncompressed. That tells you nothing about how much memory lsd would be using if it was all uncompressed. You also do not know if all of the lsd code is loaded into memory, nor how much memory lsd has allocated for data and stack space.

To put it another way, I have a "Jack-in-the-Box" with a compressed spring. I know how big the box is (resident memory), but until I open the box (uncompress it), I do not know how much space the expanded Jack-in-the-Box spring will occupy.

You and I are both drawing conclusions based on partial information.

I do not think you and I are seeing the data the same way.

I see a bucket filled with 1% Orange Juice and 99% Orange Juice Concentrate.

From the information presented I do not know how big a bucket will be needed after the Orange Juice Concentrate is turned back into Orange Juice, as nothing in the data available says what the original size was.

I think the poster wants an explanation of the difference in the numbers between the memory and compressed mem. What does a zero mean? I assume it means no compression was done. What does a memory number of 269.9 vs. 268.5 mean? Is it that 269.9mg got compressed into 268.5?

I think the poster wants an explanation of the difference in the numbers between the memory and compressed mem. What does a zero mean? I assume it means no compression was done. What does a memory number of 269.9 vs. 268.5 mean? Is it that 269.9mg got compressed into 268.5?

Since my system has all zeros for the Compressed Memory column, I would assume that 0 means nothing was compressed.

In the examples given by Student_Learner and rccharles, I'm going to guess that the Memory column is the total resident memory the process is consuming (both uncompressed and compressed), and the Compressed Memory column is the subset of that resident memory containing compressed code and data. This is a guess, but in the examples shown, the Compressed memory is always less than the resident memory, even if it is by just a tiny bit. If resident and compressed were not related this way, then there would be examples of compressed being larger than resident memory. Again, just a guess based on what I'm seeing.

I also assume that if that program scheduled for execution, any pages the program wants to access which are compressed will first have to be uncompressed before the program can use that page of memory.

Student_Learner wrote:

Actually, I thought my question was stated simply enough.

What difference does it make. Are you experiencing some sort of a problem?

The introduction of compressed memory goes back some 5 years:

OS X 10.9 Mavericks: The Ars Technica Review | Ars Technica

I think you are thinking of something else. Memory compression is for the memory used by any given application.

I do not understand your question.

The screenshot you show, says your memory is overcommitted 119MB which has been moved to a swap file. Since your Memory Pressure is not yellow, you are most likely not compressing memory in RAM at the moment.

As to compressing files, the worst offenders for filling up a disk are often music, picture or video files which are already stored in a highly compressed form, so it is not really possible to compress them further.

Can you be more specific about what you are asking?