Title: Low Power IP Design Methodology for Rapid Development of DSP Intensive SOC Platforms
1Low Power IP Design Methodology for Rapid
Development of DSP Intensive SOC Platforms
T. ArslanA.T. ErdoganS. MasupeC. Chun-FuD.
Thompson
2Contents
- Introduction to power consumption
- Introduction to Main Concepts
- Low Power Design Methodology
- IP implementations
- Results and conclusions
3Power Consumption in CMOS-Based DSP Systems
4Common Approaches to Low Power Design
- Supply Voltage Reduction
- Clock Gating
- Disadvantage
- Added design effort
5Systematic Low Power Design Approach
- Exploit Algorithmic Correlations and
Redundancies within an algorithm, then Map to
hardware.
6Systematic Design Implementation Framework
7Rapid Design and IP-Based Integration Platforms
IPx
WL
IP
N
. . .
. . .
Multiplier
Algorithm
IPy
8Developed IPs
9Parameterisation Options
10Design Flow for Filter IPs
11FIR Filter Implementation
12Typical Single Multiplier DSP Processor
Architecture
13Transpose Direct Form (TDF) FIR Structure
14Modified DSP Processor Architecture for TDF FIR
Filter Implementation
15An Example SFG for IP2
16Coefficient Memory Configuration with Coefficient
Ordering
- Order coefficients such that adjacent
coefficients are highly correlated.
17- Coefficient Word
- SF Shift Flag
- SF 1 shift
- SF 0 no shift
- PCVMA Pre-Calculated Value Memory Address
18Coefficient Word Decomposition (Verilog Code)
19An Example SFG for IP3
20Memory Operations (Verilog Code)
21Software Implementation Example for IP3
22Power Evaluation
23Filter Specifications
24Power Reductions Achieved (wordlength 16 bit)
25An example of a 6-tap FIR filter with block size
of 3
26Power Reductions for IP4 (wordlength 16 bit)
27Reductions in Number of Memory Accesses ()
28Coefficient Segmentation Algorithm
29Example Segmentations
30Example Segmentations
31Coefficient Segmentation Algorithm for Twos
Complement Coding
32Coefficient Segmentation Algorithm for
Sign-Magnitude Coding
33Total switching activity of H and M coefficient
sets with Twos Complement Coding
34Total switching activity of H and M coefficient
sets with Sign-Magnitude Coding
35Simplified Filter Architecture for Mixed-Mode
Multiplication
36Simplified Filter Architecture for
Sign-Magnitude Multiplication
37Example Switching Activity Distribution with
Twos Complement Coding (N89, W16)
38Example Switching Activity Distribution with
Sign-Magnitude Coding (N89, W16)
39Power Reductions Achieved with Coefficient
Segmentation
40Power Reduction in Multiplier Circuit (wordlength
16 bit)
35
47
44
53
62
41Power Reduction (wordlength 16 bit)
42Power Reduction at Coefficient Bus (wordlength
16 bit)
37
37
49
54
54
43DCT Implementation Scheme
442-D DCT Implementation Approach
45Simplified Architecture of the DCT Processor
46Conventional Programmable FIR Filter Architecture
47TDF with Coefficient Ordering Programmable FIR
Filter Architecture
48Power Reduction ()
49Top View of IP1
50Block Report for IP1
51Top View of IP2
52Block Report for IP2
53Top View of IP3
54Block Report for IP3
55Area Comparison
56Top View of IP4
57Top View of IP5
58Top View of IP6
59Case Study a 34-tap bandpass filter
60Area and Power Characteristics for the Example
Filter
61Conclusions
- A methodology for Low Power Implementation of DSP
functions has been presented. - The methodology has been used to develop a number
of IPs. - Significant reductions in Power is reported.
- Power reduction is achieved in the multiplier and
system buses. - Methodology can be used for prototyping other DSP
functions.