Nodepp is a vertically integrated C++ runtime engineered to eliminate the Hardware Tax in cloud and edge computing. It bridges the gap between high-level developer velocity and low-level mechanical sympathy.
NODEPP UNIFIED ARCHITECTURE: Co-designed components MODEL
=========================================================
[ APPLICATION LAYER ] Logic: High-Level Async
||
+---------||--------------------------------------------+
| || UNIFIED ptr_t DATA CARRIER |
| || (Zero-Copy / Reference Counted) |
| \/ |
| [ PROTOCOL LAYER ] Protocol Layer: HTTP / WS / TLS |
| || Parser: ptr_t Slicing |
| || |
| \/ |
| [ REACTOR LAYER ] Reactor Layer: kernel_t |
| || Engine: Epoll/KQUEUE/IOCP/NPOLL |
+---------||--------------------------------------------+
||
\/ OS Layer: LINUX / WINDOWS / MAC
[ HARDWARE / KERNEL ] Source: Sockets / Registers
Scaling the Talent Bridge for Green Computing: Achieving Silicon-Logic Parity through Deterministic RAII Read the full technical breakdown, including architectural deep-dives into ptr_t, kernel_t and coroutine_t.
In the post-Moore's Law era, hardware is no longer infinitely cheap. Traditional managed runtimes (Node.js, Go, Java) prioritize abstractions that create a 11,000x Virtual Memory Gap. Nodepp reclaims this efficiency, enabling Resource-Dense Computing where a single node does the work of an entire cluster.
Test: 100k requests | 1k Concurrency | Environment: Localhost | Device: Educational-grade Dual-Core Apollo lake Chromebook see benchmark
| Metric | Bun (v1.3.5) | Go (v1.18.1) | Nodepp (V1.4.0) | Impact |
|---|---|---|---|---|
| Requests / Sec | 5,985 | 6,139 | 6,851.33 | +11.6% Performance |
| Memory (RSS) | 69.5 MB | 14.1 MB | 2.9 MB | 95.8% Reduction |
| Max Latency | 1,452 ms | 326 ms | 245 ms | Elimination of GC Spikes |
| p99 Latency | 1,159 ms | 249 ms | 187 ms | High-precision SLA stability |
| Energy Efficiency | Low | Medium | Extreme | Maximum hardware utilization |
Test: 1k Cycles | 100k Allocations | Environment: Educational-grade Dual-Core Apollo lake Chromebook see benchmark
| Runtime | Avg. Cycle Time | VIRT (Address Space) | RES (Physical RAM) | Memory Model |
|---|---|---|---|---|
| Nodepp | 3.0 ms (±0.1 ms) | 6.1 MB | 2.7 MB | Deterministic RAII |
| Bun | 7.2 ms (5-11 ms range) | 69.3 GB | 72.6 MB | Generational GC |
| Go | < 1.0 ms* | 703.1 MB | 2.2 MB | Concurrent GC |
Note: Go's <1ms measurement reflects allocation latency only; reclamation is deferred to concurrent garbage collection cycles.
Test: 100k asynchronous tasks | Environment: Educational-grade Dual-Core Apollo lake Chromebook see benchmark
| Runtime | RSS (Memory) | CPU Load | VIRT Memory | Strategy |
|---|---|---|---|---|
| Nodepp (Balanced) | 59.1 MB | 75.9% | 153 MB | Multi-Worker Pool |
| Nodepp (Single) | 59.0 MB | 59.9% | 62 MB | Single Event Loop |
| Bun | 64.2 MB | 24.2% | 69.3 GB | JavaScriptCore Loop |
| Go | 127.9 MB | 169.4% | 772 MB | Preemptive Goroutines |
Test: 4 Valgrind-based stress tests | Environment: Educational-grade Dual-Core Apollo lake Chromebook see benchmark
| Test Case | Objective | Iterations / Load | Memory Leaks | Result |
|---|---|---|---|---|
| Atomic Longevity | High-concurrency HTTP | 100k requests | 0 bytes | PASSED |
| Rapid Lifecycle | Smart Pointer stress | 1M object cycles | 0 bytes | PASSED |
| Broken Pipe | Resilience to I/O failure | 100k interruptions | 0 bytes | PASSED |
| Multi-Thread Atomicity | race conditions stress | 100k Messages * 2 workers | 0 bytes | PASSED |
-
📌: 1. Deterministic RAII (
ptr_t): Eliminates the unpredictable latency spikes (Stop-the-World) of Garbage Collectors. By utilizing Small Stack Optimization (SSO) and reference counting, memory is reclaimed with microsecond precision. -
📌: 2. Cooperative Multitasking (
coroutine_t): Stackless coroutines eliminate context-switching overhead. This allows for massive connection density on low-power hardware, from 8-bit industrial sensors to cloud-scale reactors. -
📌: 3. Platform-Agnostic Reactor (
kernel_t): A unified abstraction over native kernel I/O (Epoll, Kqueue, IOCP, and Npoll). It provides a consistent non-blocking interface across Linux, Windows, Mac, and Bare-Metal, ensuring that I/O multiplexing is always native to the silicon.
Nodepp abstracts complex socket management into a clean, event-driven API.
#include <nodepp/nodepp.h>
#include <nodepp/regex.h>
#include <nodepp/http.h>
#include <nodepp/date.h>
#include <nodepp/os.h>
using namespace nodepp;
void onMain() {
auto server = http::server([]( http_t cli ){
cli.write_header( 200, header_t({
{ "content-type", "text/html" }
}) );
cli.write( regex::format( R"(
<h1> hello world </h1>
<h2> ${0} </h2>
)", date::fulltime() ));
cli.close();
});
server.listen( "0.0.0.0", 8000, []( socket_t /*unused*/ ){
console::log("Server listening on port 8000");
});
}The Nodepp project is supported by a suite of modular extensions designed to follow the same unified design patterns:
- 📌: Data Parsing: XML
- 📌: Tor: Torify, JWT.
- 📌: Security: Argon2,
- 📌: Web: ExpressPP, ApifyPP.
- 📌: IoT/Embedded: SerialPort, Bluetooth.
- 📌: Databases: Redis, Postgres, MariaDB, Sqlite.
Nodepp is the only framework that lets you share logic between the deepest embedded layers and the highest web layers.
- 📌: Hardware: NodePP for Arduino
- 📌: Desktop: Nodepp for Desktop
- 📌: Browser: Nodepp for WASM
- 📌: IOT: Nodepp for ESP32
Nodepp is an open-source project that values Mechanical Sympathy and Technical Excellence.
- 📌: Sponsorship: Support the project via Ko-fi.
- 📌: Bug Reports: Open an issue via GitHub.
- 📌: License: MIT.
Nodepp is distributed under the MIT License. See the LICENSE file for more details.
