Here is my HW profile and generated using Jetson -stats tools, but some information is unavailable captured from the toosl
- NVIDIA Jetson AGX Orin
- Jetpack v5.0.2 - [L4T 35.1.0]
- NV Power Mode: MAXN - Type: 0
- jetson_stats.service: inactive
- Board info:
- Type: UNKNOWN
- SOC Family: tegra23x - ID:
- Module: UNKNOWN - Board: P3737-000
- Code Name: concord
- CUDA GPU architecture (ARCH_BIN): NONE
- Serial Number: 1421622124350
- CUDA: NOT_INSTALLED
- cuDNN: 188.8.131.52
- TensorRT: 184.108.40.206
- Visionworks: NOT_INSTALLED
- OpenCV: 4.5.4 compiled CUDA: NO
- VPI: ii libnvvpi2 2.1.6 arm64 NVIDIA Vision Programming Interface library
- Vulkan: 1.3.203
- Version 3.1.4
- Works on Python 3.8.10
Below images I got from Grace, Hopper, NVSwitch Detailed at Hot Chips | NVIDIA Blogs, there are several questions need to clarify because we cannot reproduce the score which published here.
Follow the Spec2017 benchmark steps and configuration, run 1 and 12 copies for intrate benchmarking and I only got the int base rate score is 3.27 (1 copy) and 32.5 (12 copies) even enabled jetson clocks to my Jetson AGX Orin 32GB Dev Kit, which is the 20% delta to the score shared.
Question is here:
- The score published is generated from Jetson AGX Orin 32GB or 64GB?
- Is it Jetson AGX Orin Dev Kit or Modular?
- What is the “start sight” after the score shown on SPEC score, is it estimate score from Nvidia?
- Here is the command line we are running the Spec2017 benchmarking, is it correct? Else, what is the right command line? is there I can refers?
- Spec2006 rate Integer Score “269.5” published is used Jetson AGX Orin 64GB, is the score is from Jetson AGX Orin 32GB or 64GB?