I am running into roadblocks with my newly purcchased DGX Spark. PyTorch 2.8 and NGC 25.09 missing support for GB10 GPUs with sm_121 for FLUX training. Anyone here who has run into the same problem?
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Effective PyTorch and CUDA | 23 | 10425 | January 12, 2026 | |
| DGX Dashboard Playbook: pytorch in sample code not supporting cuda 12.1? | 2 | 523 | November 10, 2025 | |
| Anyone got nanochat training working on the DGX spark? | 10 | 1606 | November 21, 2025 | |
| Having trouble with my dgx spark-digits | 3 | 140 | December 5, 2025 | |
| GB10 and Docling | 3 | 410 | February 17, 2026 | |
| Can I downgrade DGX SPARK to cuda 12? | 6 | 1081 | November 2, 2025 | |
| PyTorch 2.4 and below: 10x slowdown on NVIDIA GB10 (sm_121) due to missing kernel support | 2 | 416 | October 25, 2025 | |
| Dearest CUTLASS TEAM, When the hell are you going to properly fix tcgen05 FP4 support for DGX Spark / GB10 (SM121)? | 37 | 2085 | April 25, 2026 | |
| DGX Spark (SM121) Software Support is Severely Lacking - Official Roadmap Needed | 41 | 4711 | February 15, 2026 | |
| To NVIDIA Staff: Stop leeching off community developers, Get your act together and start shopping fixes the broken VLLM & TensorLLM Packes | 4 | 267 | March 31, 2026 |