![Let’s take an example:
FIR filters (Finite Impulse Response)
y[n]=b0 x[n] + b1 x[n-1] + b2 x[n-2] + ……. + bN x[n-N]
Structurally, FIR filters consist of just two things:
◦ a sample delay line and
◦ a set of coefficients.
Round
Or
Truncate
(at fixed point)](https://image.slidesharecdn.com/dspaimp-231007041826-e5552065/85/dspa-details-33-320.jpg)
dspa details
- 2. DSP processors ◦ Architecture ◦ Data handling ◦ Program flow ◦ Programming ◦ Applications
- 3. DSP processors ◦ Architecture ◦ Data handling ◦ Program flow ◦ Programming ◦ Applications
- 4. A digital signal processor is a specialized microprocessor with an architecture optimized for the fast operational needs of digital signal processing. ◦ DSP is the application of mathematical operations to digitally represent signals. The source of these signals can be ◦ Audio ◦ Image
- 5. Digital signal processing enjoys several advantages over analog signal processing: ◦ DSP systems are able to accomplish tasks inexpensively that would be difficult or even impossible using analog electronics. (Examples of such applications include speech synthesis and speech recognition). ◦ Insensitivity to environment. ◦ Insensitivity to component tolerances. ◦ Repeatable behavior. ◦ Re-programmability. ◦ Size.
- 6. ◦ Arithmetic and Multiplication (add, subtract, increment, decrement, negate, round, absolute value) and multiplication. With the exception of the Texas Instruments TMS320Clx processor provide multiply-accumulate instructions as well. ◦ Logic Operations and, or, exclusive-or, and not. ◦ Shifting Arithmetic (left and right). Logical (left and right).
- 7. ◦ Rotation Left. Right. ◦ Comparison Most processors provide a set of status bits (ex: zero-Bit, minus Bit and overflow Bit) that provide information about the results of arithmetic operations. used in conditional branches or conditional execution instructions. ◦ Looping ◦ Subroutine Calls may be called jump-to-subroutine instructions.
- 8. ◦ Branching jump or goto instructions on some processors. Conditional Un-conditional
- 10. DSP processors ◦ Architecture ◦ Data handling ◦ Program flow ◦ Programming ◦ Applications
- 11. Instruction sets ◦ A basic DSP processor supports RISC (Reduce Instruction Set Computers) and CISC (Complex Instruction Set Computers) instructions. ◦ Single instruction, multiple data (SIMD) ◦ Instruction-level parallelism (ILP)
- 12. Single instruction, multiple data (SIMD) Single instruction, multiple data describes computers with multiple processing elements that perform the same operation on multiple data simultaneously.
- 13. Instruction-level parallelism (ILP) ◦ Instruction-level parallelism (ILP) is a measure of how many of the operations in a computer program can be performed simultaneously. ◦ Ex: 1. e = a + b 2. f = c + d independent 3. g = e * f ◦ ILP allows the compiler and the processor to overlap the execution of multiple instructions or even to change the order in instructions
- 14. ◦ Transferring information to and from memory includes data, such as samples from the input signal and the filter coefficients, as well as program instructions, the binary codes that go into the program sequencer. ◦ Ex. a b×a
- 15. There are mainly three types of architectures employed for the processors: 1. Von Neumann architecture 2. Harvard architecture 3. Super Harvard Architecture
- 16. contains a single memory and a single bus for transferring data into and out of the central processing unit (CPU). For example, Memory (instruction and data) CPU a b×a
- 17. • Advantages: • This type of architecture is cheap, and • Simple to use because the programmer can place instructions or data anywhere throughout the available memory. • Disadvantages: • Von Neumann computers spend a lot of time moving data to and from the memory, and hence slows the computer.
- 18. Separate memories for data and program instructions, with separate buses for each. • For example, Program Memory (instruction only) CPU Data Memory (data only) a b×a
- 19. • Advantages: • Since the buses operate independently, program instructions and data can be fetched at the same time, improving the speed over the single bus design. • Disadvantages: • data memory bus is busier than the program memory bus.
- 20. Improves upon the Harvard design by adding an instruction cache and dedicated I/O controller. • For example, a b×a Program Memory (instruction and secondary data) CPU Data Memory (data only) I/O Controller Instruction Cache Data
- 21. • Advantages: • the instruction cache improves the performance of the Harvard architecture. • I/O controller is connected to data memory.This dedicated hardware allows the data streams to be transferred directly into memory without having to pass through the CPU's registers. • Disadvantages: • If we were executing random instructions, this situation would be no better at all.
- 25. Now let's look inside the CPU
- 26. At the top of the diagram are two blocks labeled Data Address Generator (DAG), one for each of the two memories. ◦ These control the addresses sent to the program and data memories, specifying where the information is to be read from or written to.
- 27. The data register section : contains16 general purpose registers of 40 bits each. ◦ These can hold intermediate calculations, ◦ prepare data for the math processor, ◦ serve as a buffer for data transfer, ◦ hold flags for program control.
- 28. The math processing is broken into three sections, ◦ a multiplier (MAC), ◦ an arithmetic logic unit (ALU), and ◦ a shifter.
- 29. DSP processors ◦ Architecture ◦ Data handling ◦ Program flow ◦ Programming ◦ Applications
- 30. DSP processors fall into two major categories based on the way they represent numerical values and implement numerical operations internally. Fixed Point Floating Point
- 31. Floating point ◦ Floating point processors primarily represent numbers in floating point format. ◦ Advantages: Easier to develop code.. The large dynamic range available means that dynamic range limitations can be practically ignored in a design. ◦ Disadvantages: More expensive because they implement more functionality (complexity )in silicon and have wider buses (32 bit).
- 32. Fixed point ◦ Fixed point processors represent and manipulate numbers as integers. ◦ Advantages: lower cost and higher speed ◦ Disadvantages: Added design effort for algorithm implementation analysis, and data and Coefficient scaling to avoid accumulator overflow (16-20-24 bit).
- 33. Let’s take an example: FIR filters (Finite Impulse Response) y[n]=b0 x[n] + b1 x[n-1] + b2 x[n-2] + ……. + bN x[n-N] Structurally, FIR filters consist of just two things: ◦ a sample delay line and ◦ a set of coefficients. Round Or Truncate (at fixed point)
- 34. DSP processors ◦ Architecture ◦ Data handling ◦ Program flow ◦ Programming ◦ Applications
- 35. Pipelined Hardware-controlled looping
- 36. Pipelined Hardware-controlled looping
- 37. A pipeline is a set of data processing elements connected in series, so that the output of one element is the input of the next one. Instruction pipelines, used in processors to allow overlapping execution of multiple instructions. Fetch Decode Execute Fetch ‘A’ •2nd CLK Cycle Fetch ‘B’ Decode ‘A’ •3rd CLK Cycle Fetch ‘C’ Execute ‘A’ Decode ‘B’ •1st CLK Cycle
- 38. 1st Approach ◦ Each clock cycle = 20ns ◦ One instruction = 80 ns ◦ each stage of instruction execution is idle 75 % of the time.
- 39. 2nd Approach ◦ One instruction is now completed every clock cycle (every 20 ns)
- 40. Pipelined Hardware-controlled looping
- 41. DSP algorithms frequently involve the repetitive execution of a small number of instructions (ex: FIR and IIR filters, FFTs and matrix multiplication) DSP processors have evolved to include features to efficiently handle this sort of repeated execution. MOV #16,B LOOP : MAC (R0)+,(R4+),A DEC B JNE LOOP RPT #16 MAC (R0)+,(R4+),A
- 42. DSP processors ◦ Architecture ◦ Data handling ◦ Program flow ◦ Programming ◦ Applications
- 43. Most DSPs are programmed in special versions of C. DSP vendors will almost always provide support for C++ programming, but it is not very popular in the DSP software industry. Some DSP software programmers will resort to assembly programming for DSPs.
- 44. DSP processors ◦ Architecture ◦ Data handling ◦ Program flow ◦ Programming ◦ Applications
- 45. Digital cameras. Digital radios. High-resolution printers. Satellites.