Vendor: Skymizer Category: NPU

Octa-core LPU IP on LISA v3

Octa-core LPU IP on LISA v3 — multi-chip scalable, purpose-built for multimodal, agentic AI at the edge and on-prem.

Contact Provider Request Datasheet

Overview

HyperThought is built on the LISA™ (Language Instruction Set Architecture) foundation and an Octa-Core HTX301 LPU processor. It's a compiler-driven, software-hardware co-design that maps modern language models onto efficient silicon — without demanding a cutting-edge process node.

  • Multimodal and agentic AI at the edge
  • Automotive, industrial, and on-prem inference
  • Multi-chip scalability for extreme performance

Key features

Octa-core, multi-chip

  • Octa-Core HTX301 processor on LISA v3, scalable across chips for extreme performance.

Compression built in

  • Weight compression beats llama.cpp by 9%–17.8%.
  • KV-cache compression with 0.06%–3.52% perplexity loss.

Mature-node ready

  • T28nm with standard LPDDR4/5 — optimized for lower DRAM bandwidth.

Secure & scalable

  • Security-focused instruction architecture with a compact, balanced compute footprint

Block Diagram

HyperThought block diagram

Benefits

  • Octa-Core HTX301: 240 tok/s — Llama2 7B prefill
  • Multi-chip scaling: up to 1200 tok/s — Llama 7B prefill
  • Efficiency point: 30 tok/s at 100 GB/s & 0.5 TOPs
  • Model reach: up to 600B parameters

Specifications

Identity

Part Number
HyperThought
Vendor
Skymizer
Type
Silicon IP

Videos

Skymizer New Product Launch | 2025 新品發表記者會
Skymizer 誠摯邀請您一同線上見證全新一代 AI 加速器 IP —— HyperThought 的首度亮相！ HyperThought: Build Your Own AI Chip with Skymizer’s LPU IP!

Files

Product Sheet Download

Note: some files may require an NDA depending on provider policy.

Provider

Skymizer
HQ: Taiwan
Contact Skymizer

Learn more about NPU IP core

Benchmarking an NPU at Scale

With every SDK release, Quadric automatically recompiles and re-profiles the entire model zoo across a broad sweep of hardware configurations. The results land in DevStudio for any SoC architect to explore — no sales call, no handpicked numbers.

Your NPU Learned to Listen

Whisper runs end-to-end on Chimera. Encoder and decoder compiled as native GPNPU kernels, INT4 weights, FP16 attention, top-1 token match against the float32 reference. Scales to four cores with a flag, no recompile.

Heterogeneous NPU Data Movement Tax: Intel's Own Slides Tell the Story

At Quadric, we have long argued that heterogeneous NPU designs — those that stitch together multiple specialized fixed-function engines — carry an unavoidable hidden cost: data has to move. A lot. And data movement burns power, adds latency, and creates silicon-area overhead that scales with every new generation of AI models. Now, Intel has made that case for us.

The Upcoming NPU Shakeout

The IP industry is no stranger to boom and bust cycles, and it looks to be at the crest of another wave.

Frequently asked questions about NPU IP cores

What is Octa-core LPU IP on LISA v3?

Octa-core LPU IP on LISA v3 is a NPU IP core from Skymizer listed on Semi IP Hub.

How should engineers evaluate this NPU?

Engineers should review the overview, key features, supported foundries and nodes, maturity, deliverables, and provider information before shortlisting this NPU IP.

Can this semiconductor IP be compared with similar products?

Yes. Buyers can compare this product with similar semiconductor IP cores or IP families based on category, provider, process options, and structured technical specifications.

×
Semiconductor IP