Apps on Silicon
-
Upload
dr-ernie-prabhakar -
Category
Devices & Hardware
-
view
94 -
download
0
Transcript of Apps on Silicon
Ernest N. Prabhakar, Ph.D. • The Swan Factory, Inc.
Apps on Silicon How Intel Could Disrupt Software and Take Back Control of its Destiny
Intel’s Conundrum
❖ Hardware Rocks
❖ Fast • Reliable • Exponential
❖ Software Sucks
❖ Slow • Buggy • ~Linear
❖ Software Rules
❖ Cheap • Malleable • Experienced
The ChallengeApps
Frameworks
Operating System
Compiler
Instructions
Circuits
Silicon
Users
Intel
Intermediaries
Signal Abstraction Layer
❖ Signal-based systems are already everywhere
❖ Circuits, HTTP, Node.js, Actors, etc.
❖ Just remove idiosyncratic syntax and semantics
❖ Abstract out common elements
❖ Layer directly on top of silicon
❖ Massively streamline app creation
The Hour of NODE
❖ The first new way to learn programming in 20 years
❖ Based on four years of research into computation
❖ Build by a global team from scratch in 70 days
❖ Next-generation virtual hardware/software stack
❖ Multi-core shared memory with interrupt handling
❖ Programmable by 10-year-olds on an iPhone
The Secret Sauce❖ Get rid of instructions
❖ Invented by von Neumann to simplify math
❖ Sacrificed parallelism, composition, state management, and physicality
❖ Instead, use signal abstraction for everything
❖ Hardware • Firmware • Software • Protocols
❖ Compiler • Operating System • Frameworks
The Reality❖ The Odds
❖ 99% chance this will never work
❖ 1% chance this will change everything
❖ The Bet
❖ I left behind my dream job at Apple to bring this theory into the world
❖ Can Intel leave behind enough of the past for your silicon to rule the world?
Our Plan
❖ Develop a complete system over two years
❖ Estimated cost of $600K
❖ Faster if we work with you
❖ Slower if we don’t get funding
❖ Somebody is eventually going to disrupt software
❖ Why not you?