
Altera Corporation
7–45
September 2004
Stratix Device Handbook, Volume 2
Implementing High Performance DSP Functions in Stratix & Stratix GX Devices
Butterworth Filter Implementation Results
Table 7–16 shows the results of implementing a 4th order Butterworth Butterworth Filter Design Example
Download the 4th Order Butterworth Filter (butterworth.zip) design
example from the Design Examples section of the Altera web site at
www.altera.com.
Matrix
Manipulation
DSP relies heavily on matrix manipulation. The key idea is to transform
the digital signals into a format that can then be manipulated
mathematically.
This section describes an example of matrix manipulation used in 2-D
convolution filter, and its implementation in a Stratix device.
Background on Matrix Manipulation
A matrix can represent all digital signals. Apart from the convenience of
compact notation, matrix representation also exploits the benefits of
linear algebra. As with one-dimensional, discrete sequences, this
advantage becomes more apparent when processing multi-dimensional
signals.
In image processing, matrix manipulation is important because it
requires analysis in the spatial domain. Smoothing, trend reduction, and
sharpening are examples of common image processing operations, which
are performed by convolution. This can also be viewed as a digital filter
operation with the matrix of filter coefficients forming a convolutional
kernel, or mask.
Table 7–16. 4th Order Butterworth Filter Implementation Results
Part
EP1S10F780C6
Utilization
Lcell: 251/10570(2%)
DSP Block 9-bit elements: 16/48 (33%)
Memory bits: 0/920448 (0%)
Performance
80 MHz
Latency
4 clock cycles