Andes Extended RISC V Features
summary
Andes extends RISC-V with a priority-based preemption CLINT and vectored CLINT that pass interrupt IDs to targets and store 32-bit vector addresses to reach a 4GB range, plus a stack-protection mode that monitors SP and records peak usage. Instruction set adds larger-immediate load/stores, compare/branch-with-const, compressed 16-bit encodings, and a DSP extension (≈130–150 instructions) with 64-bit add/sub/macc, zero-overhead loops and C intrinsics; benchmarks show ~50% cycle reduction on an MP3 decoder and up to 5× speed with compiler optimizations. A custom-instruction framework generates C/RTL/simulator artifacts and test benches automatically.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 0:18About Andes
- 1:00Andes extended features
- 2:15Priority based preemption
- 6:25Preemption
- 7:32Stax
- 8:48Codons
- 9:40DSP
- 10:47Codec
- 11:52Framework
related talks
22mAndes RISC V Processor Solutions From MCU to Datacenters
10mAn Introduction to RISC-V Vector Programming with C Intrinsics
01h 25mRISC V Bit Manipulation ISA Extension Spec, Hardware, Software
01h 10mWhy RISC-V Through ISA Customization Demonstration
23mCode Density Improvements Beyond The C Standard Extension
20mAnalysis for Code Size Opportunities in RISC V
12mDemo: Boosting AI on Semidynamics RISC-V Cores with Custom Tensor Instructions
18mPreview RISC-V GCC Features and Optimization Improvements for GCC-14
