A Difference World: High-performance, NVM-invariant, Software-only Intermittent...
summary
Camel is a compiler+runtime that makes software-only intermittent computation practical on Flash devices by splitting scarce SRAM into two mixed-volatility buffers, using a checksum buffer swap and differential updates to lock forward progress. Implemented on MSP430 hardware (G2955 Flash and FR6989 FRAM) and evaluated on eight benchmarks in hardware and simulation, Camel cuts completion time versus SRAM checksum checkpointing up to 5x improvement and improves task-based FRAM systems ~2x, removes the need for voltage monitors, and works end-to-end.
AI-written summary · about
chapters (17)
- 0:00NVM-invariant intermittent computing
- 1:06Batteryless IoT challenges
- 2:50Flash limits in intermittent systems
- 3:52High-performance persistent memories
- 4:57SRAM checkpointing with Total Recall
- 6:18Task models for peripherals
- 7:22SRAM checkpoints in task models
- 8:04Camel’s mixed-volatility SRAM
- 9:28Persistent memory task models vs Camel
- 10:10Camel’s in-place task execution
- 11:35Non-idempotent tasks in Camel
- 13:19Evaluation on MSP430 platforms
- 14:01Performance vs Total Recall
- 14:44Correctness with peripheral interactions
- 15:26Camel’s buffer advantage
- 16:30Conclusion and future
- 16:52Q&A
related talks
15mA Managed Memory System for micro controllers with NOR flash memory
57mCortex-M(etal): Hyper-optimized Zephyr-Friendly Context Switching for a Mature Architecture
14mFreedom from memory interference- 57mCortex-M(etal): Hyper-optimized Zephyr-Friendly Context Switching for a Mature Architecture
01h 00mIntermittent Computer Systems on Earth and in Space
34mRISC-V-Based Sandboxing for Secure and Efficient Software Execution
16mPatchlings: Enabling A/B-Equivalent OTA Updates for Deeply Embedded...
24mBM-310 Small and Efficient MCU Core