simt

A lightweight Python-based GPU architecture simulator that demonstrates how parallel threads, registers, memory, and instructions work on a GPU.

Implementation of a custom GPU ISA microarchitecture called GBox16 based around NVIDIA and AMD microarchitectures

Synthesizable 16-lane SIMT GPU compute unit in SystemVerilog with a bit-exact Python golden model, assembler, constrained-random verification (200/200 seeds) and an RTL performance study: latency hiding, bank conflicts, branch divergence.

A SpinalHDL-based GPGPU for flexible architecture exploration.

High-performance Discrete Fourier Transform (DFT) implementation using NVIDIA CUDA with CPU vs GPU benchmarking.

Four single-cycle Verilog processor cores from scratch — 16-bit MIPS-lite, 32-bit CISC (x86-style), 32-bit RISC (ARM-style), and a 32-bit SIMT GPU — with full sim setup, FPGA constraints, and microarchitecture deep-dive docs

Tiny tensor compiler: one tensor program compiled to a GPU simulator (tiny-gpu) and a systolic-array TPU (tiny-tpu), checked bit-exact against numpy.

Twelve CUDA-style kernel puzzles on a simulated GPU. Write a kernel, pick the launch, and see warps, divergence, memory coalescing, bank conflicts, atomics and data races, graded against a par time. Built with Claude Code.

Four-warp, eight-lane programmable SIMT processor with scoreboarding, divergence, banked memory, differential verification, and an ASIC RTL-to-GDS flow.

BradVector — the Brad Devices GPU/accelerator ISA: spec, reference RTL, and toolchain

Browser game for learning CUDA: a simulated GPU runs your kernels warp by warp

Repository of the lab8 assignment for the Parallel Programming course.

Review of the paper A Formal Analysis of the NVIDIA PTX Memory Consistency Model

GPU-accelerated approximate nearest neighbor search engine with multi-arch CUDA (sm_75-sm_90) warp-cooperative HNSW, stride-257 IVF-PQ ADC, zero-copy PyTorch bindings, Rust orchestration, and a real-time 3D visualizer.

Vortex, the open-source RISC-V GPGPU, taken as a black box

SIMT GPU warp execution simulator tracking active thread divergence masks, reconvergence points, and warp vote intrinsics (__ballot_sync, __all_sync, __any_sync).

EPFL CS-550 project report

WaveSim: GPU wavefront (SIMT) execution simulator in C++17: exec masks, divergence stack, LDS bank-conflict and coalescing models. 11 kernels bit-exact vs scalar oracles at wave32/wave64, 58 tests.