[Jetson] Jetson 성능 비교 및 Roadmap

---

이번 시간에는 Jetson 시리즈의 모듈의 구성, 성능을 비교해보고,

앞으로의 로드맵을 알아보겠습니다.

 

 

 

 

 

1. Nvidia Jetson

---

Nvidia Jetson은 Nvidia 회사의 임베디드 보드의 시리즈입니다.

특히 이 보드는 인공지능, 특히 Deep Learning의 결과를 Inference할 수 있습니다.

 

Jetson family of modules

 

 

 

 

2. Benckmarks

---

Jetson 보드에 대한 모델의 성능은 아래와 같고,

GitHub자료를 참고하여 실험해보실 수 있습니다.

 

Jetson 보드별 성능

* Latency more than 15ms.

 

On Jetson Xavier NX and Jetson AGX Xavier, both NVIDIA Deep Learning Accelerator (NVDLA) engines and the GPU were run simultaneously with INT8 precision, while on Jetson Nano and Jetson TX2 the GPU was run with FP16 precision.

 

Notes:

• Each Jetson module was run with maximum performance
  • MAX-N mode for Jetson AGX Xavier
  • 15W for Jetson Xavier NX and Jetson TX2
  • 10W for Jetson Nano
• Minimum latency results
  • The minimum latency throughput results were obtained with the maximum batch size that would not exceed 15ms latency (50ms for BERT) — otherwise, a batch size of one was used.
• Maximum performance results
  • The maximum throughput results were obtained without latency limitation and illustrate the maximum performance that can be achieved.

This methodology provides a balance between deterministic low-latency requirements for real-time applications and maximum performance for multi-stream use-case scenarios. All results are obtained with JetPack 4.4.1.

 

https://developer.nvidia.com/embedded/jetson-benchmarks

Jetson Benchmarks

 

관련 GitHub 자료

https://github.com/NVIDIA-AI-IOT/jetson_benchmarks

NVIDIA-AI-IOT/jetson_benchmarks

 

 

 

 

 

3. Technical Specifications

---

사진을 클릭하면, 확대한 이미지를 볼 수 있습니다.

 

 

https://developer.nvidia.com/embedded/jetson-modules#jetson_nano

Jetson Modules

 

 

 

 

4. Jetson Roadmap

---

Jetson의 향후 Hardware와 Software 로드맵은 다음과 같습니다.

 

 

1) Hardware Roadmap

 

 

2) Software Roadmap

 

 

 

https://developer.nvidia.com/embedded/develop/roadmap

Jetson Roadmap