Secrets from the GPU
This provides a low-cost, high-throughput solution for autonomous trusted architectures in cryptographic applications, though it is incremental as it builds on existing GPU and RSA methods.
The paper tackled the challenge of accelerating cryptographic applications on GPUs by proposing a layered trusted architecture for random bits generation and parallelized RSA computations, achieving a throughput of 32-40 GB/s for random integers and supporting encryptions with up to 16,128-bit long exponents on mid-range GPUs.
Acceleration of cryptographic applications on massively parallel computing platforms, such as Graphics Processing Units (GPUs), becomes a real challenge as their decreasing cost and mass production makes practical implementations attractive. We propose a layered trusted architecture integrating random bits generation and parallelized RSA cryptographic computations on such platforms. The GPU-resident, three-tier, MR architecture consists of a RBG, using the GPU as a deep entropy pool; a bignum modular arithmetic library using the Residue Number System; and GPU APIs for RSA key generation, encryption and decryption. Evaluation results of an experimental OpenCL implementation show a 32-40 GB/s throughput of random integers, and encryptions with up to 16,128-bit long exponents on a commercial mid-range GPUs. This suggests an ubiquitous solution for autonomous trusted architectures combining low cost and high throughput.