NVIDIA A30 - 24GB usable
24GB of HBM2 at 933 GB/s, so it generates faster than its FP32 figure suggests. Passive and 165W: a server part, not a drop-in.
Specifications
| Brand | NVIDIA |
|---|---|
| Model | A30 |
| Usable VRAM | 24GB |
| Architecture | Ampere |
| CUDA / Stream Processors | 3,584 |
| Memory Bandwidth | 933 GB/s |
| TDP | 165W |
| FP32 TFLOPS | 10.3 |
Current Offers
Used from $2599New from $3890
Prices last updated:
GPUDojo is reader-supported. When you buy through links on our site, we may earn an affiliate commission.
Price History
- eBay$2599mid-range
- Amazon$3890current
- Newegg$3944at high
For AI / LLM Use
Solid choice for 30B models and comfortable 14B inference. Datacenter card with no display output, may need aftermarket cooling.
What Models Can It Run?
- 30B Q4_K_M, 14B full precision, 70B Q2 (tight)
- 14B Q6_K, 30B Q3_K (tight)
- 14B Q4_K_M, 7B full precision
- 7B Q6_K, 14B Q3_K (tight)
- 7B Q4_K_M only
Estimated Performance
Generation: ~70 tokens/sec
Prefill: ~184 tokens/sec
Recommended Quantisations
- Q4_K_M recommended for 30B models
- Q6_K or Q8 for 14B and below
- Full precision for 7B
Pros & Cons
Pros
- 24GB usable VRAM: handles large models
- High memory bandwidth for fast generation
- Ampere architecture: broad CUDA software support
Cons
- No display output: headless only
- May need aftermarket cooling solution
Community Verdict
No community reviews yet for the A30. Know a good review? Let us know.