Re: [AMBER] Posits preparing to pounce on Floating Point

From: Matias Machado <>
Date: Fri, 19 Jul 2019 13:59:38 -0300 (UYT)

"Gustafson told us that a 32-bit posit can replace a 64-bit float in almost all cases"... "he thinks posit-based arithmetic would deliver a two-fold to four-fold speedup compared to IEEE floats."

** WT#! Please, somebody tell Redhat and Canonical not to drop the 32-bit support yet :-P **

"Perhaps the largest opportunity for posits is in machine learning, where 16-bits can be used for training and 8-bits for inference."

** AI with a pocket calculator?? **


Researcher at Biomolecular Simulations Lab.
Institut Pasteur de Montevideo | Uruguay

----- Mensaje original -----
De: "Bill Ross" <>
Para: "AMBER Mailing List" <>
Enviados: Jueves, 18 de Julio 2019 0:29:52
Asunto: [AMBER] Posits preparing to pounce on Floating Point

    Another important advantage to the format is that unlike
    conventional floats, posits produce the same bit-wise results on any
    system, which is something that cannot be guaranteed with the IEEE
    standard (even the same computation on the same system can product
    different results for floats). It also does away with rounding
    errors, overflow and underflow exceptions, subnormal (denormalized)
    numbers, and the plethora of not-a-number (NaN) values.
    Additionally, posits avoids the weirdness of 0 and -0 as two
    distinct values. Instead it uses an integer-like twos complement
    form to encapsulate the sign, which means that simple bit-wise
    comparisons are valid.

AMBER mailing list

AMBER mailing list
Received on Fri Jul 19 2019 - 10:30:03 PDT
Custom Search