Full-precision Floating-Point Units (FPUs) can be a source of extensive hardware overhead in general-purpose and application-specific processors nowadays. As several modern applications feature an inherent ability to tolerate precision loss, a new computing paradigm has emerged: Transprecision Computing (TC). For such applications, the standard IEEE Floating-Point (FP) formats are over-designed.This thesis proposes contributions that alleviate some State-of-the-Art limitations. First, we propose a minimally-intrusive approach that introduces approximations at the instruction execution/interpretation level, allowing the approximation of virtually all kinds of executable binaries (bare-metal applications, single-/multi-threaded user applicati...
Les unités de calcul à virgule flottante (FP) prennent en charge un sous-ensemble de formats définis p...
Floating-Point (FP) units in processors are generally limited to supporting a subset of formats defin...
Using standard Floating-Point (FP) formats for computation leads to significant hardware overhead si...
Full-precision Floating-Point Units (FPUs) can be a source of extensive hardware overhead in general...
International audienceFull-precision Floating-Point Units (FPUs) can be a source of extensive hardwa...
open3siIn recent years approximate computing has been extensively explored as a paradigm to design h...
The slowdown of Moore's law and the power wall necessitates a shift toward finely tunable precision ...
This paper presents the design and the implementation of a fully combinatorial floating point unit (...
In modern low-power embedded platforms, floating-point (FP) operations emerge as a major contributor...
International audienceFloating-Point (FP) computation using standard IEEE formats has a significant ...
Most of the Floating-Point (FP) hardware units support the formats and the operations specified in t...
The crisis of Moore's law and new dominant Machine Learning workloads require a paradigm shift towar...
Les unités de calcul à virgule flottante (FP) prennent en charge un sous-ensemble de formats définis p...
Floating-Point (FP) units in processors are generally limited to supporting a subset of formats defin...
Using standard Floating-Point (FP) formats for computation leads to significant hardware overhead si...
Full-precision Floating-Point Units (FPUs) can be a source of extensive hardware overhead in general...
International audienceFull-precision Floating-Point Units (FPUs) can be a source of extensive hardwa...
open3siIn recent years approximate computing has been extensively explored as a paradigm to design h...
The slowdown of Moore's law and the power wall necessitates a shift toward finely tunable precision ...
This paper presents the design and the implementation of a fully combinatorial floating point unit (...
In modern low-power embedded platforms, floating-point (FP) operations emerge as a major contributor...
International audienceFloating-Point (FP) computation using standard IEEE formats has a significant ...
Most of the Floating-Point (FP) hardware units support the formats and the operations specified in t...
The crisis of Moore's law and new dominant Machine Learning workloads require a paradigm shift towar...
Les unités de calcul à virgule flottante (FP) prennent en charge un sous-ensemble de formats définis p...
Floating-Point (FP) units in processors are generally limited to supporting a subset of formats defin...
Using standard Floating-Point (FP) formats for computation leads to significant hardware overhead si...