First, I want to encourage you to go read the tutorial 
Managing Large Data Sets in LabVIEW.  Using the techniques there, I have streamed data to disk at 12MB/sec on a 650MHz PIII (much more on a modern system).  You should be able to do what you want to do, provided you are not trying to do it on a laptop (disk drive probably cannot keep up).  There are some simple improvements you can make that will probably help a lot.
- You take the data using DBLs, then scale it and save it as I16.  This scaling and conversion takes a significant amount of time and is unnecessary.  Fetch your data as unscaled I16 and you will probably see quite a bit of speed increase.  You can save your scaling factors as the first two numbers in your binary file.  Or you can use NI-HWS for your data storage, which natively handles the scaling for you (you still want to fetch and save as I16 to avoid a factor of 4 increase in data size).
 - You happen to be saving to disk at about the right chunk size.  The optimum is about 65,000 bytes - you have 60,0000 byte chunks.  However, this may not be the optimum chunk size to read from the DAQ device.  I don't know what this is.  For NI-SCOPE devices, it is about 300,000 bytes.  You can optimize both the read and write sizes by double buffering your acquisition.  You can roll your own with a producer-consumer architecture, or you can use NI-HWS, which automatically double-buffers and optimizes the write, leaving you to only have to worry about the read.
 
You will find it impossible to create a 1.5GByte RAM buffer in LV.  Due to memory fragmentation issues, the largest single array you can create in LV7.1 is a bit over 1GByte.  You can create an architecture with multiple arrays accessed as a single array and get up to about 1.4GBytes.  System considerations prevent you from getting any more.  You can compress your data into memory and get more (maybe).  It is a lot easier to stream to disk efficiently.
Take home message - streaming to disk, as you are trying to do, is within your PC's and LabVIEW's capabilities.  Let us know if you have further problems.