New Approach Could Sink Floating Point Computation

25 bookmarks. First posted by fileformat 9 days ago.

We caught up with Gustafson at ISC19. For that particular crowd, the supercomputing set, one of the primary advantages of the posits format is that you can get more precision and dynamic range using less bits than IEEE 754 numbers. And not just a few less. Gustafson told us that a 32-bit posit can replace a 64-bit float in almost all cases, which would have profound implications for scientific computing. Cutting the number of bits in half not only reduces the amount of cache, memory and storage to hold these values, but also substantially reduces the bandwidth needed to shuffle these values to and from the processor. It’s the main reason why he thinks posit-based arithmetic would deliver a two-fold to four-fold speedup compared to IEEE floats.

floatingpoint
posits
unum
6 days ago by euler

RT @alpinegizmo: New Approach Could Sink Floating Point Computation

from twitter
7 days ago by clonezone

In 1985, the Institute of Electrical and Electronics Engineers (IEEE) established IEEE 754, a standard for floating point formats and arithmetic that

7 days ago by synergyfactor

A new approach to floating point maths - major nerd stuff ahoy https://t.co/YhQAkKwqyP

— Mike Lynch (@bombinans) July 10, 2019

twitter
— Mike Lynch (@bombinans) July 10, 2019

7 days ago by mikelynch

Posits might beat out floating point?

IEEE
IEEE754
floating
point
math
Posit
computer
science
mathematics
algorithm
research
technology
7 days ago by asteroza

RT @newsycombinator: John Gustafson’s crusade to replace floating point with something better

from twitter
8 days ago by hyperfekt

8 days ago by unconsidered

There is ACTUALLY a credible alternative to IEEE 754 for floating point.

floatingpoint
programming
computerscience
via:HackerNews
8 days ago by mcherm

New Approach Could Sink Floating Point Computation

floatingPoint
posits
math
cpus
computerScience
8 days ago by gojomo

tags

algorithm architecture bestpractices computer computers computerscience cpus development everynumber floating floatingpoint floats ieee ieee754 ifttt interesting machine_learning math mathematics performance pocket point posit posits programming research science technology twitter unum via:hackernews