CommTech FASTCOM SuperFASTCOM, Hardware Reference Manual

Page: 23 / 32

16
MEMORY MANAGEMENT
High-speed communications requires that large amounts of transmitted/received data be
buffered so as to prevent data loss and maintain data throughput. The SuperFASTCOM
is
designed to utilize system memory directly, bypassing the system processor. The
SuperFASTCOM
accesses system memory through high-speed PCI Bus Mastering, using its
on-board bus master controller and supplied software. This procedure for direct memory
access (DMA) is by far the fastest, most efficient method of handling the large amounts of data
that high-speed communications generates. Data is transferred to and from the
SuperFASTCOM
adapter at data bus speed. Buffering a high-speed data stream is critical in
order to maintain data integrity and reliability, to avoid data loss, and to prevent data overrun.
The SuperFASTCOM
does all this with minimal system processor involvement, therefore there
is no reduction in processor performance due to data communication overhead.
How it works - The user designates in his program how much system memory will be allocated
to communications: the more memory in your system, the more you can allocate to the
communication process. As data is received, the on-board communication processor fills a
small local buffer. When this is full, the bus master controller on the SuperFASTCOM
transfers
this data directly to system memory. The local buffer is reset and the process begins again.
Transmitting data works in reverse; data from system memory is directly transferred to the
local buffer on the SuperFASTCOM. When the local buffer is empty, new data from memory is
directly transferred in. All of this occurs without slowing or interfering with the system
processor.
The high speed PCI Bus Mastering procedure for direct memory access provides the fastest,
most efficient method of data transfer. It also improves efficiency and reduces the cost of the
SuperFASTCOM
board by eliminating the need for on-board memory or a processor.