Can RTX 5090 run Nemotron 3.5 Lightning 30B-A3B?
Comfortable: GGUF Q4_0 needs 19.8 GB at 4K, batch 1; headroom 12.2 GB.
Text input, FP16 KV cache, one card. Unified-memory capacity uses the GPU guide’s usable limit; negative headroom may require host offload. Speed source settings may differ from this 4K estimate.
Measured runs
No measured run for this pair. Sizing is an estimate, not a runtime test.
Quant memory at 4K
| Quant | Total GB | Verdict | Weight sizing |
|---|---|---|---|
| GGUF Q4_0 | 19.8 | Comfortable | Verified file |
| GGUF Q8_0 | 35.1 | Does not fit | Verified file |
Context capacity · GGUF Q4_0
| Context | Total GB | Verdict |
|---|---|---|
| 4K | 19.8 | Comfortable |
| 16K | 19.9 | Comfortable |
| 32K | 20.0 | Comfortable |
| 128K | 20.6 | Comfortable |
Generation speed
GGUF Q4_0: 1050 tok/s · Bandwidth ceiling (not measured)