# R12: Piranha: A GPU Platform for Secure Computation

Piranha separates device, protocol and application layers for GPU-accelerated secret-sharing computation. PLLM status: tracked.

[View canonical HTML](https://pllm.run/research/papers/r12-piranha/)

Document ID: `pllm.docs.research.papers.r12-piranha`  
Release: `0.1.0`

**Status:** `tracked`
**Year:** 2022
**Authors:** Jean-Luc Watson, Sameer Wagh, Raluca Ada Popa
**External source:** [Piranha: A GPU Platform for Secure Computation](https://www.usenix.org/conference/usenixsecurity22/presentation/watson)

## Source contribution

Piranha separates device, protocol and application layers for GPU-accelerated secret-sharing computation.

## PLLM perspective

Wrap original GPU kernels through a narrow native interface before selective Rust/CUDA ports. Preserve protocol party counts and numeric domains in comparisons.

Previous CPU limb/packing experiments are not a GPU Piranha reproduction.

This paper remains tracked. Named modules below are implementation targets, not finished PLLM components; no reproduction, integration, assurance, or matched benchmark is claimed.

The source's boundary remains binding: Kernel reuse does not turn a multi-online-worker protocol into the preferred single-evaluator topology.

## PLLM modules and documentation

Target module identifiers: `kernel.modular_gpu`, `runtime.device_buffers`. These identifiers describe scoped work unless this page explicitly links an implementation source. Dependencies: None.

- [Research backlog](/research/backlog/)
- [Reimplementation backlog](/research/backlog/)
- [Clean-room workflow](/research/clean-room/)

## Backlog

This source is assigned to reimplementation priority 7, **Preparation, compression, and acceleration**. Its source-registry order is 12. Follow current gates and unfinished work in the [research backlog](/research/backlog/).