Qualcomm Adreno GPU
Vendor: Qualcomm
Category: GPU Graphics Processor
Architecture: Adreno
Introduction
Qualcomm Adreno GPU is built into Snapdragon SoCs, supporting Vulkan, OpenCL, OpenGL ES and other graphics and compute APIs. Suitable for GPU-accelerated computing on mobile and edge devices.
Specifications
| Model | Compute | Memory | Interface | TDP | Process |
|---|---|---|---|---|---|
| Adreno 830 (SD 8 Elite) | 12 TFLOPS (FP16) | LPDDR5X (shared) | Integrated SoC | 10W | 3nm |
| Adreno 750 (X Elite) | 8.5 TFLOPS (FP16) | 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 |
|---|---|---|
| Adreno SDK 3.0 | 2024 | Snapdragon 8 Elite Vulkan 1.3 |
Performance Benchmarks
| Model | Task | Performance Metric |
|---|---|---|
| Adreno 830 (SD 8 Elite) | Stable Diffusion Mobile | ~8s/img |
| Adreno 830 | INT8 Inference | ~45 TOPS |
| Adreno 750 (X Elite) | AI Inference | ~40 TOPS |
Pricing
| Model | Reference Price | Notes |
|---|---|---|
| Adreno 830 | Provided with SoC | Snapdragon 8 Elite |
| Adreno 750 | Provided with SoC | Snapdragon X Elite |
Quick Installation
Android (via Qualcomm AI Engine)
# Use QNN (Qualcomm Neural Network) SDK
# Download from Qualcomm Developer Network
# Compile and deploy TFLite/QNN model
adb push model.dlc /data/local/tmp/
adb shell snpe-qlc-run --container /data/local/tmp/model.dlc
Windows on Snapdragon
Simply install Windows ARM64 drivers to use Adreno GPU.
Code Examples
Python (QNN / TFLite)
# Android: Use TFLite with Adreno GPU acceleration
import tflite_runtime.interpreter as tflite
interpreter = tflite.Interpreter(
model_path="model.tflite",
experimental_delegates=[tflite.load_delegate('libGpuDelegate.so')]
)
interpreter.allocate_tensors()
Architecture Highlights
- Adreno GPU: Qualcomm mobile/PC GPU integrated in Snapdragon SoCs
- Hexagon NPU Co-processing: Adreno GPU works with Hexagon DSP for AI acceleration
- Snapdragon X Elite: PC-class AI processing for Windows on ARM
Model Compatibility
| Model/Framework | Support | Notes |
|---|---|---|
| TFLite | ✅ GPU delegate | Main mobile platform |
| QNN SDK | ✅ Native | Qualcomm native inference |
| ONNX Runtime | ✅ | WinML / QNN backend |
| Stable Diffusion | ⚠️ | Mobile limited |
Related Products
If you're evaluating alternatives, the following products may also fit your scenario:
- Qualcomm Hexagon NPU — Qualcomm (NPU Neural Processor)
- Apple Silicon GPU — Apple (GPU Graphics Processor)
- MediaTek NeuroPilot — MediaTek (NPU Neural Processor)
- NVIDIA GPU / CUDA — NVIDIA (GPU Graphics Processor)
- AMD ROCm / GPU — AMD (GPU Graphics Processor)
- Intel Data Center GPU — Intel (GPU Graphics Processor)