Graphics Reference
In-Depth Information
Fig. 5.2
Basic architecture of the target computing platform. The overall design contains the
imaging sensors (e.g., at WVGA resolution), a low-cost FPGA (e.g., a Spartan 6 Model 45 or
better), and an external high-speed communications controller (e.g., USB 2.0 or 3.0, GigaEthernet).
The overall processing pipeline, including the FIFO aimed at handling transfers to/from the high-
speed communications controller and a softcore that supervises the whole system, is synthesized
into the reconfigurable logic of the FPGA
ment, compactness, and reduced bill of materials. Concerning power requirements,
the overall design has a power consumption of about 2.0W supplied by a standard
USB 2.0 data connector. The overall size of the computing platform depicted in
Fig.
5.2
, excluding the imaging sensors modules, has an area smaller than a credit
card. Finally, the bill of materials can be summarized by considering a small amount
of inexpensive hardware devices. Namely, it results in the FPGA, the imaging sen-
sors and the external USB or GigaEthernet controller plus some standard electronic
components such as clocks and passive devices. According to this overview, it is clear
that the considered computing platform is not equipped with any external memory
device such as a SRAMor a DDR. This choice simplifies the overall design (enabling
power, area, and costs reduction) but, on the other hand, enforces strong constraints
to the processing capabilities of such a processing platform that will be thoroughly
discussed in the remainder of this chapter.
Observing Fig.
5.2
, we can also notice that our design contains a
softcore
synthe-
sized into the FPGA logic; this is a small RISC processor mainly aimed at handling
communications with the host computer in order to change camera parameters, such
as the frame rate or other features of the imaging sensors, by means of standard
serial communication protocols (e.g., I2C). The softcore, by means of software com-
mands issued by the host, also allows us to select, among those available within the
processing pipeline, the video streams actually required by the user. For video stream
configuration, the softcore is tightly coupled with the Data/Video manager unit as
