Key Points

  1. SiMa.ai has launched Palette 3.0, an on-chip software operating system that enhances AI workloads on its Modalix edge AI processors.
  2. The new platform improves application sharing, introduces techniques for accelerating text generation, and offers a model browser for developers to optimize model performance.
  3. Palette 3.0 is most relevant to teams whose perception and reasoning models already work on their own but compete for compute and memory when they run together.

The New Palette 3.0: What It Means for Edge AI Developers

SiMa.ai has launched Palette 3.0, an on-chip software operating system that enhances AI workloads on its Modalix edge AI processors. The new platform improves application sharing, introduces techniques for accelerating text generation, and offers a model browser for developers to optimize model performance.

Palette 3.0 is most relevant to teams whose perception and reasoning models already work on their own but compete for compute and memory when they run together. For a robot, this can come down to whether object detection responds within its deadline while the language model generates an answer.

At a Glance

Main CompanySiMa.ai
Developer of the Palette 3.0 edge AI platform
Main ExecutiveManuel López Roldán
GenAI product manager at SiMa.ai
PlatformPalette 3.0
On-chip software operating system for edge AI processors
Key TechnologySpeculative decoding
Technique used to accelerate text generation
Key Figure3.5X
Speedup in text generation performance

Behind the Scenes: How SiMa.ai Optimized Palette 3.0 for Edge AI

Behind the Scenes: How SiMa.ai Optimized Palette 3.0 for Edge AI

SiMa.ai's GenAI product manager, Manuel López Roldán, confirmed that the changes run on existing silicon. The company's associate vice president of machine learning accelerator (MLA) software, Ashok Sudarsanam, said his team builds the compiler that translates models into instructions for the accelerator and the software that schedules them to run.

SiMa.ai's decoding figures put a number on generation speed. The mixed-workload demos leave that deadline question for developers to test on their own applications.

What Developers Need to Know About Palette 3.0 and Its Implications

What Developers Need to Know About Palette 3.0 and Its Implications

  • Palette 3.0 is most relevant to teams whose perception and reasoning models already work on their own but compete for compute and memory when they run together.
  • Developers can assign priorities to individual requests, giving higher priority to tasks that require continuous processing.
  • SiMa.ai's decoding figures put a number on generation speed, but the mixed-workload demos leave the deadline question for developers to test on their own applications.