triton

LLVM based static binary analysis framework

NVIDIA-accelerated, deep learned model support for image space object detection

NVIDIA-accelerated DNN model inference ROS 2 packages using NVIDIA Triton/TensorRT for both Jetson and x86_64 with CUDA-capable GPU

Binary Ninja plugin that can be used to apply Triton's dead store eliminitation pass on basic blocks or functions.

Standalone static version of Triton's x86/x64 translator

COIN Attacks: on Insecurity of Enclave Untrusted Interfaces in SGX - ASPLOS 2020

Adaptive Callsite-sensitive Control Flow Integrity - EuroS&P'19

Sydr benchmark applications

Triton backend for managing the model state tensors automatically in sequence batcher

Triton backend for https://github.com/OpenNMT/CTranslate2

TensorFlow Lite backend with ArmNN delegate support for Nvidia Triton

pluggable tool to convert an unrolled TritonAST to LLVM-IR, optimize it and get back to TritonAST

MLModelService wrapping Nvidia's Triton Server

Custom CUDA & Triton fused layers for self-stabilizing transformer architectures. Accelerate forward/backward passes and prevent gradient explosions in large-scale LLM training.

FlashAttention from scratch in CUDA: exact attention, tiled through SRAM, N×N score matrix never materialized

Comprehensive CUDA learning curriculum - from basics to Tensor Cores, covering Thrust, CUB, cuBLAS, CUTLASS, and Triton

Threads, concurrent programming, Triton & CUDA Kernels, all in one.