X100 / Strix Halo
AMD's flagship targeting AI and robotics edge workloads. Strix Halo represents an x86-based alternative to the Jetson ecosystem, offering strong CPU performance with unified memory for LLM/VLM workloads.
Platforms
What each platform is on paper, what it was actually configured to for these runs, and how it performed. Specifications are the vendors' own; everything measured is from the runs in this repository.
Vendor specifications for the three platforms evaluated.
| Feature | X100 / Strix Halo | Jetson Thor | Jetson AGX Orin |
|---|---|---|---|
| CPU | 16x Zen 5 (32T) @ 5.1 GHz | 14x Arm Neoverse V3AE @ 2.6 GHz | 12x Arm Cortex-A78AE @ 2.2 GHz |
| GPU architecture | RDNA 3.5 | Blackwell | Ampere |
| GPU cores | 2560 Shaders (40 CUs) | 2560 CUDA + 96 Tensor | 2048 CUDA + 64 Tensor |
| NPU / DLA | XDNA 2 (50 TOPS) | None | 2x NVDLA 2.0 |
| RAM | Up to 128 GB LPDDR5X | 128 GB LPDDR5X | 64 GB LPDDR5 |
| Memory bandwidth | 256 GB/s | 273 GB/s | 204.8 GB/s |
| Power (TDP) | 45-120 W | 40-130 W | 15-60 W |
| Vendor | AMD | NVIDIA | NVIDIA |
AMD's flagship targeting AI and robotics edge workloads. Strix Halo represents an x86-based alternative to the Jetson ecosystem, offering strong CPU performance with unified memory for LLM/VLM workloads.
NVIDIA's next-generation embedded AI module built on the Blackwell GPU architecture. Thor is designed as the next platform for physical AI and advanced robotics applications and enables edge AI such as LLM/VLM and foundation models.
NVIDIA's established embedded AI platform widely adopted in robotics which require edge AI such as detection, segmentation, and reinforcement learning. The Orin is the current workhorse of many production robotics deployments.
The tested TDP and other key configurations.
| Configuration | X100 / Strix Halo | Jetson Thor | Jetson AGX Orin |
|---|---|---|---|
| TDP, Max Power (W) | 120 | 130 | 65 |
| TDP, Max Power (AI Workload Optimized) (W) | 120 | 130 | n/a |
| TDP, Balanced Power (W) | 85 | 90 / 70 | 65 |
| Software stack | Ubuntu, ROCm | JetPack, CUDA | JetPack 6.2.2, CUDA 12.6 |
| ROS distribution | ROS 2 Jazzy | ROS 2 Jazzy | ROS 2 Jazzy |
| DDS | CycloneDDS | CycloneDDS | CycloneDDS |
| Run length | 900 s at 1 Hz sampling | 900 s at 1 Hz sampling | 900 s at 1 Hz sampling |
| Nav2 configuration | nav2_params.yaml | nav2_params.yaml | nav2_params.yaml |