gpu-compute

A pseudo operating system for local and cloud compute (Opensourcing on 5000th star)

CUDA kernels in any language supported by LLVM

Revives the last functional CUDA deep-learning stack on macOS High Sierra. NVIDIA CUDA 10.2 + cuDNN 7.6.5 + PyTorch 1.7.0 fully working on GTX Pascal GPUs — a discontinued ecosystem resurrected.

Computable GPU Index (CGI): an open, verifiable price index for GPU compute, in USD per GPU-hour. Reproduce any published value from the public record.

Real-Time AI Frame Interpolation for AMD GPUs

Deterministic, integer-only SLM inference runtime — no floating point on the inference path, so the same model and prompt produce bit-identical output across CPUs, GPUs, compilers, and vendors.

Daily-verified free LLM APIs, AI token prices, and GPU compute deals with no-card filters, official evidence, and a local-first dashboard.

A ray-tracer/path-tracer implemented in JavaScript using the WebGL 2 API

Experimental open-source compatibility research for running DLSS Neural Rendering workloads on AMD RDNA2/gfx1030. Includes ISA translation, host emulation, proof-oriented validation, and bounded hardware testing. Early development: no end-to-end game frame yet.

.NET binding for wgpu-native

Browser-based Finite Element Analysis solver showcasing WebGPU compute and Rust → WebAssembly numerical pipelines.

Experimental host-side Metal compute API prototype with a Vulkan executor and offline parity research.

Fastest Minecraft coordinates calculator using rotations for Windows.

Experimental tensor compiler with a Python frontend, typed SSA IR, semantics preservation tests, and Metal lowering.

Autonomous ML research agent skill (Agent Skills format). Designs experiments, deploys to Modal GPUs, tracks budget, iterates autonomously.

Silicon Fingerprinting & Hardware OSINT Engine leveraging GPU-accelerated telemetry analysis and microscopic clock-skew variance. Built with Rust, C, and HIP/ROCm.

Programmable, verifiable execution on Xbox Series X Developer Mode, with agent-native workflows, XVM, CPU/GPU runtimes and evidence-bound execution.

The .NET wrapper of the Cocaine C library. Cocaine is a multi-platform C library that can be used to accelerate large workloads/big data/anything really with the power of a GPU with ease.

🌊✨There's something about jellyfish that makes you stop and watch. They don't swim like other creatures. They pulse. They drift. They become the water itself.

A cross-platform C++ GPU compute microbenchmark supporting Vulkan, DX12, DX11, OpenGL, and Metal. Features hardware timestamp profiling and deep cross-architecture analysis.