BPU Stream Engine

High-speed realtime streaming engine demo using ESP32 devices.

This project demonstrates a reliable serial data pipeline:

ESP32-WROOM → ESP32-S3 → PC
with COBS framing, CRC16 validation, and live visualization.

Designed for stress-testing embedded streaming systems.

---

Demo

Realtime draw streaming over UART (921600 baud) with live CRC/sequence monitoring.

---

Project Goal

To build and validate a high-reliability, high-throughput embedded streaming pipeline for real-time visualization and stress testing.

---

Features

• COBS framing (0x00 delimited packets)
• CRC16-CCITT integrity check
• Sequence number validation
• High-rate draw stress generator
• Live visualization on PC (Python + matplotlib)
• Throughput and error statistics

---

System Architecture

ESP32-WROOM (Generator) | UART @921600 | ESP32-S3 (Bridge) | USB CDC | PC Viewer

Repository Structure

bpu-stream-engine/
 ├─ s3_uart_bridge/
 │   └─ s3_uart_bridge.ino
 │
 ├─ bpu_r4_safe.ino
 ├─ bpu_r4.cpp
 ├─ bpu_r4.h
 │
 ├─ bpu_decode.py
 ├─ bpu_draw_viewer.py
 ├─ bpu_mat_viewer.py
 │
 ├─ demo.gif
 └─ README.md

Firmware

Main Files

File	Description
bpu_r4_safe.ino	Main firmware entry
bpu_r4.cpp	Core stream engine
bpu_r4.h	Engine definitions

Runs on ESP32-WROOM.

---

Build & Flash

ESP32-WROOM

1. Open bpu_r4_safe.ino
2. Select board: ESP32 Dev Module
3. Select correct COM port
4. Upload firmware

ESP32-S3 Bridge

Upload s3_uart_bridge/s3_uart_bridge.ino
Board: ESP32-S3 Dev Module

---

Wiring

WROOM TX → S3 RX
WROOM RX → S3 TX (optional)
GND → GND

Baudrate: 921600

---

PC Viewer

Install

pip install pyserial matplotlib numpy

---

List Ports

python -m serial.tools.list_ports

---

Run

python bpu_mat_viewer.py COM11

(Change COM port if needed)

---

Viewer Output

• Realtime draw points
• PPS / DPS
• CRC errors
• Sequence gaps

Stable example:

crc=0 seqGap=0 drop=0

---

Packet Format

[MAGIC | TYPE | SEQ | LEN | PAYLOAD | CRC16]

MAGIC = 0xB2
CRC = CCITT

COBS framed, 0x00 delimited.

---

Performance

UART: 921600
1000+ pkt/s
10k+ pts/s
Zero CRC errors

---

Project Status

Streaming: Stable
Validation: Verified
Viewer: Working
Stress: Enabled

---

Future Work

• Bidirectional control
• Compression
• Web viewer
• Sync
• FPGA offload

Quick Start

1. Flash firmware to ESP32-WROOM
2. Flash bridge firmware to ESP32-S3
3. Connect UART (TX/RX/GND)
4. Connect S3 to PC via USB

Run viewer:

pip install pyserial matplotlib python bpu_mat_viewer.py COM11

---

License

MIT License

---

Author

Himchan Choi
Embedded Systems Engineer
High-Speed Data Streaming & Visualization

GitHub: https://github.com/choihimchan