Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

TINA: Acceleration of Non-NN Signal Processing Algorithms Using NN Accelerators

Christiaan Boerkamp, Steven van der Vlugt, Zaid Al-Ars

TL;DR

This paper introduces TINA, a novel framework for implementing non Neural Network (NN) signal processing algorithms on NN accelerators such as GPUs, TPUs or FPGAs, and shows that TINA is highly competitive vs alternative frame-works, specifically for complex functions with iterations.

Abstract

This paper introduces TINA, a novel framework for implementing non Neural Network (NN) signal processing algorithms on NN accelerators such as GPUs, TPUs or FPGAs. The key to this approach is the concept of mapping mathematical and logic functions as a series of convolutional and fully connected layers. By mapping functions into such a small substack of NN layers, it becomes possible to execute non-NN algorithms on NN hardware (HW) accelerators efficiently, as well as to ensure the portability of TINA implementations to any platform that supports such NN accelerators. Results show that TINA is highly competitive compared to alternative frameworks, specifically for complex functions with iterations. For a Polyphase Filter Bank use case TINA shows GPU speedups of up to 80x vs a CPU baseline with NumPy compared to 8x speedup achieved by alternative frameworks. The framework is open source and publicly available at https://github.com/ChristiaanBoe/TINA.

TINA: Acceleration of Non-NN Signal Processing Algorithms Using NN Accelerators

TL;DR

This paper introduces TINA, a novel framework for implementing non Neural Network (NN) signal processing algorithms on NN accelerators such as GPUs, TPUs or FPGAs, and shows that TINA is highly competitive vs alternative frame-works, specifically for complex functions with iterations.

Abstract

This paper introduces TINA, a novel framework for implementing non Neural Network (NN) signal processing algorithms on NN accelerators such as GPUs, TPUs or FPGAs. The key to this approach is the concept of mapping mathematical and logic functions as a series of convolutional and fully connected layers. By mapping functions into such a small substack of NN layers, it becomes possible to execute non-NN algorithms on NN hardware (HW) accelerators efficiently, as well as to ensure the portability of TINA implementations to any platform that supports such NN accelerators. Results show that TINA is highly competitive compared to alternative frameworks, specifically for complex functions with iterations. For a Polyphase Filter Bank use case TINA shows GPU speedups of up to 80x vs a CPU baseline with NumPy compared to 8x speedup achieved by alternative frameworks. The framework is open source and publicly available at https://github.com/ChristiaanBoe/TINA.
Paper Structure (20 sections, 20 equations, 3 figures, 1 table)

This paper contains 20 sections, 20 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Building blocks of TINA
  3. Standard convolution
  4. Depthwise convolution
  5. Pointwise convolution
  6. Fully connected layer
  7. Arithmetic functions mapping
  8. Elementwise multiplication
  9. Matrix-matrix multiplication
  10. Elementwise addition
  11. Summation
  12. Signal processing functions mapping
  13. DFT
  14. IDFT
  15. FIR filter
  16. ...and 5 more sections

Figures (3)

  • Figure 1: Runtime of the arithmetic functions vs input size using TINA, NumPy, CuPy & JAX
  • Figure 2: Runtime of signal processing functions vs input size using TINA, NumPy, CuPy & JAX
  • Figure 3: GPU speedups with respect to NumPy (CPU) of PFB without (left column) and with (right column) Fourier transform in TINA, JAX & CuPy