Accurate Floating Point Arithmetic through Hardware Error-Free Transformations
نویسندگان
چکیده
This paper presents a hardware approach to performing accurate floating point addition and multiplication using the idea of errorfree transformations. Specialized iterative algorithms are implemented for computing arbitrarily accurate sums and dot products. The results of a Xilinx Virtex 6 implementation are given, area and performance are compared against standard floating point units and it is shown that the time×area product can be improved by a factor of 4 and 6 over software error-free addition and multiplication. Furthermore, it is illustrated that a number of area and performance trade-offs can be made when calculating vector sums and dot products.
منابع مشابه
Error-free transformations in real and complex floating point arithmetic
Error-free transformation is a concept that makes it possible to compute accurate results within a floating point arithmetic. Up to now, it has only be studied for real floating point arithmetic. In this short note, we recall the known error-free transformations for real arithmetic and we propose some new error-free transformations for complex floating point arithmetic. This will make it possib...
متن کاملOn the Design and Performance of Reliable Geometric Predicates using Error-free Transformations and Exact Sign of Sum Algorithms
We study the relevance of algorithms for exact computation of the sign of a sum of floating-point numbers and error-free transformations of arithmetic expressions on floating-point numbers for the design and implementation of low-dimensional geometric predicates. In a case study, we experimentally compare several implementations of planar orientation test and incircle tests that make use of suc...
متن کاملCAMPARY: Cuda Multiple Precision Arithmetic Library and Applications
Many scientific computing applications demand massive numerical computations on parallel architectures such as Graphics Processing Units (GPUs). Usually, either floating-point single or double precision arithmetic is used. Higher precision is generally not available in hardware, and software extended precision libraries are much slower and rarely supported on GPUs. We develop CAMPARY: a multipl...
متن کاملSHORT-SS4: Error-Free Transformation of Matrix Multiplication by A Posteriori Verification
This paper is concerned with accurate computations for matrix multiplication. An error-free transformation of matrix multiplication is developed by the authors. It transforms a product of two floatingpoint matrices to a sum of several floating-point matrices by using only floating-point arithmetic. This transformation is useful not only for accurate matrix multiplication but also for interval e...
متن کاملError-Free Transformation in Rounding Mode toward Zero
In this paper, we provide new error-free transformations for the sum and the product of two floating-point numbers. These error-free transformations are well suited for the CELL processor. We prove that these transformations are error-free, and we perform numerical experiments on the CELL processor comparing these new error-free transformations with the classic ones.
متن کامل