Overview
The ADSP-21160NCB-100 is a high-performance 32-bit Digital Signal Processor (DSP) from Analog Devices Inc., part of the SHARC DSP family. It is the second iteration of the ADSP-21160, fabricated in a 0.18 micron CMOS process, which offers higher performance and lower power consumption compared to its predecessor, the ADSP-21160M. This processor is optimized for high-performance DSP applications and is application source code compatible with first-generation ADSP-2106x SHARC DSPs in Single Instruction, Single Data (SISD) mode, although some code changes are necessary to leverage its Single Instruction, Multiple Data (SIMD) capability.
Key Specifications
| Specification | Value |
|---|---|
| Core Instruction Rate | 100 MHz (10.5 ns) |
| Peak Performance | 570 MFLOPS |
| Sustained Performance (FIR) | 380 MFLOPS |
| Instruction Execution | Single-cycle, including SIMD operations |
| Computational Units | Two 32-bit IEEE floating-point computation units |
| Data Address Generators (DAGs) | Dual DAGs with modulo and bit-reverse addressing |
| On-Chip Memory | 4M-bit dual-ported SRAM |
| Package | 400-ball 27 × 27 mm Metric PBGA |
| Test Access Port | IEEE 1149.1 JTAG standard |
| Looping and Loop Setup | Zero-overhead looping and single-cycle loop setup |
Key Features
- SIMD Architecture: The ADSP-21160NCB-100 features a Single Instruction, Multiple Data (SIMD) architecture with two computational processing elements, enabling concurrent execution and code compatibility at the assembly level with the ADSP-2106x SHARC DSP family.
- Integrated Peripherals: Includes an integrated I/O processor, 4M bits of on-chip dual-ported SRAM, glueless multiprocessing features, and various ports such as serial, link, external bus, and JTAG.
- Efficient Program Sequencing: Offers zero-overhead looping and single-cycle loop setup, providing efficient program sequencing.
- Multi-Bus Architecture: Features multiple internal buses to eliminate I/O bottlenecks.
- Power and Performance: Fabricated in a 0.18 micron CMOS process, it offers higher performance and lower power consumption compared to its predecessor.
Applications
The ADSP-21160NCB-100 is optimized for various high-performance DSP applications, including:
- Medical Applications: Suitable for medical imaging and diagnostic equipment.
- Military Applications: Used in military communications and imaging systems.
- Imaging Applications: Ideal for high-resolution imaging and processing.
- Communications: Applied in advanced communication systems requiring high DSP performance.
- Multiprocessing Systems: Designed for multiprocessing environments, including industrial and speech recognition systems.
Q & A
- What is the core instruction rate of the ADSP-21160NCB-100?
The core instruction rate is 100 MHz (10.5 ns).
- What is the peak performance of the ADSP-21160NCB-100?
The peak performance is 570 MFLOPS.
- Does the ADSP-21160NCB-100 support SIMD operations?
Yes, it supports Single Instruction, Multiple Data (SIMD) operations with two computational units.
- What type of on-chip memory does the ADSP-21160NCB-100 have?
The ADSP-21160NCB-100 has 4M bits of on-chip dual-ported SRAM.
- What is the package type of the ADSP-21160NCB-100?
The package type is a 400-ball 27 × 27 mm Metric PBGA.
- Is the ADSP-21160NCB-100 compatible with previous SHARC DSPs?
Yes, it is application source code compatible with first-generation ADSP-2106x SHARC DSPs in SISD mode, but some code changes are needed to leverage SIMD capability.
- What are the key peripherals included in the ADSP-21160NCB-100?
The key peripherals include an integrated I/O processor, serial ports, link ports, external parallel bus, and JTAG.
- What is the power consumption characteristic of the ADSP-21160NCB-100 compared to its predecessor?
The ADSP-21160NCB-100 offers lower power consumption compared to the ADSP-21160M.
- What are some typical applications of the ADSP-21160NCB-100?
Typical applications include medical imaging, military communications, imaging systems, and advanced communication systems.
- Does the ADSP-21160NCB-100 support glueless multiprocessing?
Yes, it supports glueless multiprocessing features.
