Can A100 80G run Ripple-65B-Preview?
Comfortable: GGUF Q6_K needs 60.9 GB at 4K, batch 1; headroom 19.1 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 IQ1_M | 20.0 | Comfortable | Verified file |
| GGUF IQ2_XXS | 21.6 | Comfortable | Verified file |
| GGUF IQ2_XS | 23.8 | Comfortable | Verified file |
| GGUF IQ2_S | 27.2 | Comfortable | Verified file |
| GGUF IQ3_XXS | 29.9 | Comfortable | Verified file |
| GGUF IQ3_S | 34.5 | Comfortable | Verified file |
| GGUF IQ4_XS | 41.1 | Comfortable | Verified file |
| GGUF Q4_K_M | 41.2 | Comfortable | Verified file |
| GGUF Q5_K_M | 50.8 | Comfortable | Verified file |
| GGUF Q6_K | 60.9 | Comfortable | Verified file |
| GGUF Q8_0 | 72.6 | Tight | Verified file |
Context capacity · GGUF Q6_K
| Context | Total GB | Verdict |
|---|---|---|
| 4K | 60.9 | Comfortable |
| 16K | 61.8 | Comfortable |
| 32K | 62.9 | Comfortable |
| 128K | 69.5 | Comfortable |
Generation speed
GGUF Q6_K: 35 tok/s · Bandwidth ceiling (not measured)