FAQ.txt

Why are these examples here?

	These examples are here for educational purposes, to illustrate
	some of the interesting issues around Persistent Memory programming.
	Start by reading the top-level README, then read the API document
	LINUX_PMEM_API.txt.  Then start examining source code.  First,
	trivial/trivial.c, then basic/basic.c, then take a look through
	the libpmemalloc example to see how one might approach the issues
	of Persistent Memory allocation and crash resiliency.

	These are just examples, meant to be simple, self-contained,
	simple C based files.  See the REFERENCES section in the README
	for links to other more complex Persistent Memory work going on.

	For more information, questions or comments, contact:

		andy.rudoff@intel.com

What is Persistent Memory?

	See the PERSISTENT MEMORY section in the top-level README.

Where I can purchase Persistent Memory?

	Start by search for NVDIMM products, many of which meet the
	definition of Persistent Memory described in the README and
	should work fine with PMFS and the examples contained here.
	The press is frequently reporting more emerging Persistent
	Memory technologies as well.

Can Persistent Memory be used transparently to applications?

	Certainly.  One can easily imagine hardware completely managing
	Persistent Memory so that the Operating System and applications
	have no idea it is even installed in the system.  Alternatively,
	one can imagine the OS using Persistent Memory itself but not
	exposing it to applications, so again it is transparent at the
	application level.

	The examples and APIs discussed here are for the non-transparent
	case, where an application wants to fully take advantage of
	Persistent Memory by deciding what data structures to store in it,
	when to access it, when to copy data into volatile memory and back,
	etc.  Transparent use of Persistent Memory is also very interesting,
	but is beyond the scope of the examples in this repository.

How do kernel modules use Persistent Memory?

	File systems like PMFS and other kernel modules that want to
	leverage Persistent Memory access it by calling into the device
	driver for the Persistent Memory.  The PMFS team is working on
	getting their "NVM Volume" interfaces into the Linux kernel.
	But the in-kernel interfaces are beyond the scope of these
	examples -- they are for application-level programming.

How do these examples compare with the Mnemosyne or NV Heaps work?

	If you are interesting in Persistent Memory programming, looking
	at the Mnemosyne and NV Heaps work is highly recommended.  The
	examples here are meant to be very simple, working with existing
	compilers, and self-contained.  So these examples do not build on
	those other works.  It is our hope that these examples help
	stimulate even more examples and more PM-related research.

What exactly do I need to do in order to make stores persistent?

	It depends on the Persistent Memory product, and on the platform
	it is installed in.  Check with the manufacturer of the Persistent
	Memory you are using.  For some examples, see the REFERENCES section
	of the top-level README.  The idea is that the gory details on how
	you flush stores to the point of durability in Persistent Memory
	is contained in libraries like libpmem.  See the Linux API document
	for a description of how your code can depend on libpmem for
	making stores persistent.