GPUs / NVIDIA
RTX 4070 SUPER
- Memory
- GDDR6X
- Bus
- 192-bit
- Published peak
- 504 GB/s
- Capacity
- 12 GB
Measured
- Read
- 472.2GB/s
- Write
- 453.5GB/s
- Copy
- 427.2GB/s
- Random
- 63.6GB/s
- Latency
- 241ns
- Sustain
- 426.6GB/s
Medians of 4 verified runs from 2 operators. Read reaches 94 % of the published peak.
Local AI speed
| Model | Parameters | 4-bit, tokens/s | 8-bit, tokens/s |
|---|---|---|---|
| Llama 3.1 8B | 8.03B | ≤ 118 | ≤ 59 |
| Qwen 2.5 14B | 14.7B | ≤ 64 | Does not fit |
| gpt-oss-20b | 21B, 3.6B active | ≤ 262 | Does not fit |
| Mistral Small 24B | 24B | ≤ 39 | Does not fit |
| Qwen 2.5 32B | 32.5B | Does not fit | Does not fit |
| Llama 3.3 70B | 70.6B | Does not fit | Does not fit |
| gpt-oss-120b | 117B, 5.1B active | Does not fit | Does not fit |