pascale-joseph
Uploaded by
4 SLIDES
171 VUES
40LIKES

Program Generation with Spiral: Automating Algorithm Selection and Implementation in Computing

DESCRIPTION

This paper discusses advancements in program generation using the Spiral framework, aimed at automating algorithm selection and implementation in computing. Supported by DARPA and NSF, the research tackles the challenge of high-level abstraction to achieve complete automation. The authors explore concepts like common abstraction spaces, rewriting search, and architectural parameters impacting optimization. Key applications include matrix multiplication and Viterbi decoding, demonstrating the framework's utility in diverse computational tasks such as JPEG 2000 compression and synthetic aperture radar.

1 / 4

Télécharger la présentation

Program Generation with Spiral: Automating Algorithm Selection and Implementation in Computing

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Program Generation with Spiral: Beyond Transforms Franz Franchetti, Daniel Mcfarlin, Frédéricde Mesmay, Hao Shen, Tomasz W. Włodarczyk, Srinivas Chellappa, Marek R. Telgarsky, Peter A. Milder, Yevgen Voronenko, Qian Yu, James C. Hoe, JoséM. F. Moura, Markus Püschel Electrical and Computer Engineering Carnegie Mellon University This work was supported by DARPA DESA program, NSF-NGS/ITR, NSF-ACR, Mercury Inc., and Intel

  2. Vision Behind Spiral Current Future Numerical problem Numerical problem human effort algorithm selection C program algorithm selection implementation automated implementation compilation compilation automated Computing platform Computing platform C code a singularity: Compiler hasno access to high level information Challenge: conquer the high abstraction level for complete automation

  3. Main Idea: Program Generation Model: common abstraction = spaces of matching formulas abstraction abstraction νp μ defines rewriting search pick algorithm space architecture space Architectural parameter: Vector length, #processors, … Kernel: problem size, algorithm choice optimization

  4. Expressing Kernels as Operator Formulas Matrix-Matrix Multiplication Viterbi Decoder 11 10 01 01 10 10 11 00 11 10 00 01 10 01 11 00 010001 010001 convolutionalencoder Viterbi decoder £ = JPEG 2000 (Wavelet, EBCOT) Synthetic Aperture Radar (SAR) JPEG 2000 Compression matched filtering preprocessing interpolation 2D iFFT DWT quantization entropy coding (EBCOT + MQ)

More Related