The New Chapter: Lessons From the Best Engineer I’ve Ever Known
It’s Only Common Sense: Time for Some Mind-blowing, Earth-shaking, Putting-a-dent-in-the-world Changes
The Chemical Connection: Can the Limits of Subtractive Etching Be Extended?Cross-Industry Hardware Specification to Accelerate AI Software Development
September 16, 2022 | IntelEstimated reading time: 1 minute
Arm, Intel and Nvidia have jointly authored a paper describing an 8-bit floating point (FP8) specification and its two variants E5M2 and E4M3 to provide a common interchangeable format that works for both artificial intelligence (AI) training and inference. This cross-industry specification alignment will allow AI models to operate and perform consistently across hardware platforms, accelerating AI software development.
Computational requirements for AI have been growing at an exponential rate. New innovation is required across hardware and software to deliver computational throughput needed to advance AI.
One of the promising areas of research to address this growing compute gap is to reduce the numeric precision requirements for deep learning to improve memory and computational efficiencies. Reduced-precision methods exploit the inherent noise-resilient properties of deep neural networks to improve compute efficiency.
Intel plans to support this format specification across its AI product roadmap for CPUs, GPUs and other AI accelerators, including Habana® Gaudi® deep learning accelerators.
FP8 minimizes deviations from existing IEEE 754 floating point formats with a good balance between hardware and software to leverage existing implementations, accelerate adoption and improve developer productivity.
The guiding principle of this format proposal from Arm, Intel and Nvidia is to leverage conventions, concepts and algorithms built on IEEE standardization. This enables the greatest latitude for future AI innovation while still adhering to current industry conventions.
Share on:
Suggested Items
Siemens Collaborates with Samsung Foundry to Expand 3D-IC Enablement Tools
06/14/2024 | SiemensSiemens Digital Industries Software today announced that, in collaboration with Samsung Foundry, they have developed compelling new capabilities for the manufacture of multi-die packaged designs at advanced nodes and achieved a host of new product certifications for many of Siemens’ industry-leading IC design and verification technologies.
A/NZ Software Spending Slowed in 2023, Artificial Intelligence Grew 62% YoY
06/10/2024 | IDCAccording to the International Data Corporation (IDC) Worldwide Semiannual Software Tracker (2H23), annual software spending across A/NZ grew 17.3% year-over-year (YoY) in 2023, reaching US$20.9 billion in constant currency.
Airbus, Helsing to Collaborate on AI for the Teaming of Manned and Unmanned Military Aircraft
06/05/2024 | AirbusAirbus Defence and Space and Helsing, Europe’s leading defence AI and software company, signed a framework cooperation agreement at the ILA aerospace trade show in Berlin.
Foxconn Shareholders Approve Record Cash Dividend for 2023
05/31/2024 | FoxconnHon Hai Technology Group, at its annual shareholders’ meeting, approved its distributable earnings plan totaling NT$74.86 billion in cash. The cash dividend of NT$5.40 per share for 2023 sets a new high since Foxconn listed in 1991, with the cash dividend payout ratio exceeding 50% for the fifth consecutive year.
Intel AI Platforms Accelerate Microsoft Phi-3 GenAI Models
05/27/2024 | Intel CorporationIntel has validated and optimized its AI product portfolio across client, edge and data center for several of Microsoft’s Phi-3 family of open models.