Saved in this browser, used on every model page.
Yes
Yes. The Tesla P40 runs tiny-voca fully in VRAM at FP16 / BF16 (about 0.5 GB of 24 GB).
- best quant
- FP16 / BF16
- needs
- 0.5 GB
- usable
- 24 GB
Estimates from the model's size.Try other contexts on the tiny-voca page →
Every quantisation of tiny-voca on the Tesla P40
| Quant | Memory | Fit |
|---|---|---|
| Q4_K_M | 0.5 GB | fits |
| Q8_0 | 0.5 GB | fits |
| FP16 / BF16 pick | 0.5 GB | fits |
Speed is a bandwidth-based estimate for generating text (347 GB/s); prompt processing and long contexts are slower.
TTS models that fit the Tesla P40 comfortably
All that fit →| Model | Params | Best quant | Memory |
|---|---|---|---|
| 36M | FP16 / BF16 | ≈0.6 GBFits in memory | |
| 82M | FP16 / BF16 | ≈0.7 GBFits in memory | |
| 66M | FP16 / BF16 | ≈0.7 GBFits in memory | |
| 66M | FP16 / BF16 | ≈0.7 GBFits in memory | |
| 82M | FP16 / BF16 | ≈0.7 GBFits in memory | |
| 100M | FP16 / BF16 | ≈0.7 GBFits in memory |
Other GPUs that run tiny-voca
Frequently asked questions
- Can the Tesla P40 run tiny-voca?
- Yes. The Tesla P40 runs tiny-voca fully in VRAM at FP16 / BF16 (about 0.5 GB of 24 GB).
- Which quantisation of tiny-voca should I use on the Tesla P40?
- FP16 / BF16: the largest one that fits. Expected quality: lossless compared with the full-precision weights.
Memory and speed are estimates from the model's size and the card's memory bandwidth; real numbers depend on the runtime, drivers and context length. Community figures are medians of user reports.