Home Publications edited volumes Awards Research Teaching Miscellaneous Full CV [pdf] BLOG
Events
Past Events

Publications of Torsten Hoefler
Johannes de Fine Licht, Christopher A. Pattison, Alexandros Nikolaos Ziogas, David SimmonsDuffin, Torsten Hoefler:
  Fast Arbitrary Precision Floating Point on FPGA
(In Proceedings of the 30th IEEE International Symposium on FieldProgrammable Custom Computing Machines (FCCM'22), May 2022)
AbstractNumerical codes that require arbitrary precision floating point (APFP) numbers for their core computation are dominated by elementary arithmetic operations due to the superlinear complexity of multiplication in the number of mantissa bits. APFP computations on conventional softwarebased architectures are made exceedingly expensive by the lack of native hardware support, requiring elementary operations to be emulated using instructions operating on machinewordsized blocks. In this work, we show how APFP multiplication on compiletime fixedprecision operands can be implemented as deep FPGA pipelines with a recursively defined Karatsuba decomposition on top of native DSP multiplication. When comparing our design implemented on an Alveo U250 accelerator to a dualsocket 36core Xeon node running the GNU Multiple Precision FloatingPoint Reliable (MPFR) library, we achieve a 9.8x speedup at 4.8 GOp/s for 512bit multiplication, and a 5.3x speedup at 1.2 GOp/s for 1024bit multiplication, corresponding to the throughput of more than 351x and 191x CPU cores, respectively. We apply this architecture to general matrixmatrix multiplication, yielding a 10x speedup at 2.0 GMAC/s over the Xeon node, equivalent to more than 375x CPU cores, effectively allowing a single FPGA to replace a small CPU cluster. Due to the significant dependence of some numerical codes on APFP, such as semidefinite program solvers, we expect these gains to translate into realworld speedups. Our configurable and flexible HLSbased code provides as highlevel software interface for plugandplay acceleration, published as an open source project.
Documentsdownload article: download slides:   BibTeX  @inproceedings{, author={Johannes de Fine Licht and Christopher A. Pattison and Alexandros Nikolaos Ziogas and David SimmonsDuffin and Torsten Hoefler}, title={{Fast Arbitrary Precision Floating Point on FPGA}}, year={2022}, month={May}, booktitle={Proceedings of the 30th IEEE International Symposium on FieldProgrammable Custom Computing Machines (FCCM'22)}, source={http://www.unixer.de/~htor/publications/}, } 

