Saved in this browser, used on every model page.
Yes
Yes. The Arc B580 runs Crowd-v1 fully in VRAM at FP16 / BF16 (about 2.8 GB of 12 GB), at roughly 125 tokens/s.
- best quant
- FP16 / BF16
- needs
- 2.8 GB
- usable
- 12 GB
- est. speed
- ~125 tok/s
Estimates for an 8K context; offload assumes 32 GB of system RAM.Try other contexts on the Crowd-v1 page →
Every quantisation of Crowd-v1 on the Arc B580
| Quant | Memory | Fit | Speed |
|---|---|---|---|
| Q2_K | 1.2 GB | fits | ~763 tok/s |
| Q3_K_M | 1.3 GB | fits | ~513 tok/s |
| Q4_K_M | 1.4 GB | fits | ~414 tok/s |
| Q5_K_M | 1.6 GB | fits | ~353 tok/s |
| Q6_K | 1.7 GB | fits | ~304 tok/s |
| Q8_0 | 1.9 GB | fits | ~236 tok/s |
| FP16 / BF16 pick | 2.8 GB | fits | ~125 tok/s |
Speed is a bandwidth-based estimate for generating text (456 GB/s); prompt processing and long contexts are slower.
Measured speed on the Arc B580
No community reports for this pair yet.
Report your own speed →Chat models that fit the Arc B580 comfortably
All that fit →| Model | Params | Best quant | Memory | Speed |
|---|---|---|---|---|
| 1B | Q8_0 | ≈1.8 GBFits in memory | ~236 tok/s | |
| 752M | Q8_0 | ≈2.0 GBFits in memory | ~314 tok/s | |
| 1.2B | Q8_0 | ≈2.0 GBFits in memory | ~191 tok/s | |
| 1.1B | Q8_0 | ≈1.7 GBFits in memory | ~215 tok/s | |
| 1.7B | Q8_0 | ≈2.7 GBFits in memory | ~137 tok/s | |
| 1.5B | Q8_0 | ≈2.6 GBFits in memory | ~153 tok/s |
Other GPUs that run Crowd-v1
Frequently asked questions
- Can the Arc B580 run Crowd-v1?
- Yes. The Arc B580 runs Crowd-v1 fully in VRAM at FP16 / BF16 (about 2.8 GB of 12 GB), at roughly 125 tokens/s.
- How fast does Crowd-v1 run on the Arc B580?
- An estimate from memory bandwidth gives about 125 tokens/s at FP16 / BF16.
- Which quantisation of Crowd-v1 should I use on the Arc B580?
- 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.