Yes. The Jetson TX2 NX is the only TX2-generation module that fits a 260-pin SO-DIMM slot. It uses the same 260-pin SO-DIMM (69.6x45mm) form factor as the Xavier NX, and it's described as pin-compatible with Nano and Xavier NX. developer... +1developer.nvidia.com ↗“Reduced-form-factor Pascal SoM (Nano/Xavier NX pin-compatible)”developer.nvidia.com ↗“mechanical.form_factor=260-pin SO-DIMM (69.6x45mm)”
Jetson TX2 NX key specs developer...developer.nvidia.com ↗“256 CUDA cores, 6-core CPU, 4GB LPDDR4, 7.5-15W”
| Spec | Jetson TX2 NX |
|---|---|
| Form factor | 260-pin SO-DIMM (69.6x45mm) |
| GPU | Pascal, 256 CUDA cores |
| CPU | 6 cores |
| Memory | 4 GB LPDDR4, 51.2 GB/s |
| Power envelope | 7.5–15 W |
| FP32 | 0.666 TFLOPS |
The other TX2 modules won't fit. The Jetson TX2, TX2 4GB and TX2i all use a 400-pin B2B (50x87mm) connector, not SO-DIMM. developer... +2developer.nvidia.com ↗“identity.jetson_family=Jetson TX2, identity.lifecycle_status=EOL, mechanical.form_factor=400-pin B2B (50x87mm)”developer.nvidia.com ↗“Pascal SoM; 256 CUDA cores, 6-core CPU, 4GB LPDDR4, 7.5-15W; cost-optimized TX2.”developer.nvidia.com ↗“Industrial/rugged Pascal SoM”
Things to check before you switch:
- Smaller module overall. Compared with the Xavier NX it replaces, the TX2 NX has no Tensor cores and no NVDLA engines. Its memory is 4 GB vs. 8 or 16 GB. developer... +1developer.nvidia.com ↗“compute.tensor_cores=0, compute.nvdla_engines=0”developer.nvidia.com ↗“compute.tensor_cores=48, compute.nvdla_engines=2”
- Fitting the slot isn't the same as full support. Matching pins don't guarantee that every carrier feature or software package works the same way. Check with your carrier's vendor that they support the TX2 NX.
- Lifecycle status: the TX2 NX's lifecycle status isn't listed in the catalog. The 400-pin TX2 and TX2 4GB are both marked EOL. developer...developer.nvidia.com ↗“identity.lifecycle_status=EOL”