Can RTX 3090 run Clover-35B-A3B-Preview?
Comfortable: GGUF Q4_0 needs 20.9 GB at 4K, batch 1; headroom 3.1 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 Q3_K_M | 18.2 | Comfortable | Verified file |
| GGUF Q4_0 | 20.9 | Comfortable | Verified file |
| GGUF Q4_K_M | 23.4 | Tight | Verified file |
| GGUF Q5_K_M | 28.3 | Does not fit | Verified file |
| GGUF Q8_0 | 38.6 | Does not fit | Verified file |
Context capacity · GGUF Q4_0
| Context | Total GB | Verdict |
|---|---|---|
| 4K | 20.9 | Comfortable |
| 16K | 21.2 | Tight |
| 32K | 21.5 | Tight |
| 128K | 23.6 | Tight |
Generation speed
GGUF Q4_0: 571 tok/s · Bandwidth ceiling (not measured)