USR is not Flash at all. This is the built-in RAM where you can sample and edit audio.
Data recorded via the sampleR is written here, in a type of memory called SDRAM. There is 128MB of USR SDRAM memory. The audio data here is in volatile memory... Means it disappears when you power down. This memory is automatically backed up when you Save an ALL (.X3A), an ALL VOICE (.X3V). If your sample is used in a Voice or if you sample to the Sequencer, the audio goes to this same SDRAM. Sampled audio that is recorded to the sequencer is automatically backed up when you save an ALL (.X3A), ALL SONG (.X3S) or ALL PATTERN (.X3P) file.

FL1 and FL2 are your Flash Boards. This type of memory is a special type of storage media with lightning fast access times. It has to retrieve audio data and deliver 128 channels of audio with unnoticeable delay (measured in nanoseconds). Flash is all about robust audio retrieval.

Data is kept semi-permanently (this means data written to Flash will remain there until you either DELETE it or FORMAT the Board). Because you are placing data in semi-permanent locations, all of YOUR files will contain a roadmap of where YOU have installed YOUR custom data. This roadmap is called the "Waveform List".

When you wish to install audio data to your Flash Board there are two methods: you can COPY it from USR SDRAM or you can LOAD it from a FILE.
When you read about Waveforms and Samples... At first you think they are the same thing. They are not. Appreciating this difference is the first step toward understanding.

A digital recording to the XF is called a SAMPLE. it is the audio data. A .Wav file can be imported to the XF. It is equal to a sample.
In order for the XF to use that audio it must apply a set of parameters that allow it to "playback" that audio. The very basic parameters include Key Range and Velocity Range. These additional parameters are what organize samples into something you can access are the Waveform. A Waveform can contain as few as one and as many as 256 samples.

The .wav is no longer just a .wav, the XF has added a set of parameters that elevate it to a musical level. It is mapped to a equal temperament keyboard, and responds to a 1-127 range of velocities... Something it did not do as a .wav - it is now a Motif XF Waveform. The single audio sample itself could be 64MB in size. The set of parameters organizing it could be as small as 2KB.

Of significance is the relative size of the data. While a set of audio Samples can be huge (the Chick Corea Mark V is 400MB), the Waveform data is only a few kilobytes in size.

When you LOAD data from a File, you can select the TYPE of data you are going to LOAD or SAVE.
The SAVE Options appear on the lower left of the FILE screen.
The LOAD Options appear on the lower right of the FILE screen.

SAVE
You can choose to SAVE a File "with Sample" or "without Sample". Why would you choose one or the other?
You only need to SAVE "with sample" when making a backup archive file because you've added new (or deleted) data to your Flash Board library.
You normally would SAVE "without sample" on a day to day basis. Say you've just recorded a new Pattern or edited a Voice. You don't need to make a 40 minute backup of all your Flash data just to Save the new Pattern or a small tweak. It is the rare occasion that you SAVE "with sample".

The small Waveform data (the parameters that organize the samples) is automatically written to every File. You choose whether or not the big Sample data is also Saved.

LOAD
When you LOAD a new Library, the extended File Name extension will tell you if a File contains data for your Flash Boards or not, and for which Flash Board it is intended. You decide where you would like to direct this data.
.n3.X3A is an ALL data File with no samples for either FL1 or FL2
.n2.X3A is an ALL data File with no samples for FL2
.n1.X3A is an ALL data File with no samples for FL1
.X3A is an ALL data File with Samples for both FL1 and FL2

If you are loading Voices and the Waveforms are already on your instruments Waveform List, there is no need to have the XF load data that is already in memory. The XF is fairly smart, but it will rely on you to know what you want to do.

The XF can detect Waveforms with the same name. If two Waveforms have exactly the same name and come from the same File, the XF will "know" and it will not duplicate the Waveform.

It cannot detect the difference between Waveforms with the same name that originate in different Files. It will depend on you (the human) to direct the data and make the decision to Load specific data. In other words, when loading from one of your own files, the XF will recognize the names of the Waveforms and if the data is already installed to your Board, it will simply repoint the Voice data to the location on the List. It will not duplicate the Waveform. So why all these load options...?

All of the options are there for different situations. Touring musicians want to take just their Flash Board and thumb drives on the road, and they will rent a backline Motif XF in the city they go to... Motif XFs are everywhere. They can pop in their Boards and load their data. In less than 1 minute they are up and running!

You may never need some of the Load Options. Particularly if you are just working with one XF. But as you see you can Load with, or without samples, you redirect data to the other Board ...as you may require... Or you can set the Flash Board to NONE which allows no changes to your installed data. NONE is a very important option. One you will use most often. Again, in the day to day use of your XF, you are not installing data to your Flash Boards... When you do not want to change anything concerning your Flash Board set this To NONE.

Hope that helps shed some light on it.