Qualcomm Hexagon NPU
Vendor: Qualcomm
Category: NPU Neural Processor
Architecture: Hexagon (HTA + HVX)
Introduction
Qualcomm Hexagon NPU integrated in Snapdragon SoCs, providing efficient AI inference acceleration. Accessed through Qualcomm AI Engine Direct and SNPE frameworks, supporting INT8/INT16 quantized inference.
Specifications
| Model | Compute | Memory | Interface | TDP | Process |
|---|---|---|---|---|---|
| Snapdragon 8 Elite | 45 TOPS (INT8) | LPDDR5X (shared) | Integrated SoC | 10W | 3nm |
| Snapdragon X Elite | 45 TOPS (INT8) | LPDDR5X (shared) | Integrated SoC | 23W | 4nm |
Official Website
Driver Downloads
Linux
Related Documentation
OS Support
| Windows | Linux | macOS | Android |
|---|---|---|---|
| ✅ (Snapdragon) | ✅ (Android) | ❌ | ❌ |
Version History
| Version | Release Date | Description |
|---|---|---|
| QNN 2.x | 2024 | Snapdragon 8 Elite support |
Performance Benchmarks
| Model | Task | Performance Metric |
|---|---|---|
| Snapdragon 8 Elite | NPU TOPS | 45 TOPS (Hexagon) |
| Snapdragon X Elite | NPU TOPS | 45 TOPS |
| Snapdragon 8 Elite | Stable Diffusion Mobile | ~10s/img |
Pricing
| Model | Reference Price | Notes |
|---|---|---|
| Snapdragon 8 Elite | Provided with SoC | Flagship phone SoC |
| Snapdragon X Elite | Provided with SoC | Windows on ARM |
Quick Installation
Android / Windows on Snapdragon
Hexagon NPU is invoked through Qualcomm AI Engine Direct.
# Install QNN SDK (download from Qualcomm Developer Network)
# Supports TFLite/ONNX model compilation and deployment
Code Examples
Python (QNN Runtime)
# Run compiled model using QNN SDK
from qnn import QnnRuntime
runtime = QnnRuntime()
model = runtime.load("model.qnn")
output = model.run(input_data)
Architecture Highlights
- Hexagon DSP + HVX: Qualcomm scalar/vector DSP architecture, supporting AI inference and signal processing
- HTA (Hexagon Tensor Accelerator): Dedicated tensor accelerator optimized for Transformer models
- AI Engine: Unified AI software stack covering Hexagon + Adreno GPU + Kryo CPU
Model Compatibility
| Model/Framework | Support | Notes |
|---|---|---|
| QNN SDK | ✅ Native | Best support |
| TFLite | ✅ | NNAPI/Hexagon backend |
| ONNX | ✅ | QNN compile |
| Stable Diffusion | ✅ | Mobile-optimized version |
| Whisper | ✅ | On-device voice |
Related Products
If you're evaluating alternatives, the following products may also fit your scenario:
- AMD Ryzen AI NPU — AMD (NPU Neural Processor)
- MediaTek NeuroPilot — MediaTek (NPU Neural Processor)
- Qualcomm Adreno GPU — Qualcomm (GPU Graphics Processor)
- Intel NPU (Neural Processing Unit) — Intel (NPU Neural Processor)
- Samsung Gauss AI — Samsung (NPU Neural Processor)
- Huawei Ascend — Huawei (NPU Neural Processor)
- NVIDIA GPU / CUDA — NVIDIA (GPU Graphics Processor)