RTX 4000 Ada vs RTX A4000
Manufacturer specifications of the RTX 4000 Ada and the RTX A4000, side by side, per GPU. A value the manufacturer does not publish for the exact model is marked as such, never estimated.
NVIDIA RTX 4000 Ada 20GB
Ada Lovelace
- GPU memory
- 20 GB
- Memory bandwidth
- Not published
Live price on EverXCloud · No offer right now
Suited workloads
NVIDIA RTX A4000 16GB
Ampere
- GPU memory
- 16 GB
- Memory bandwidth
- 448 GB/s
Live price on EverXCloud · from $0.10 per GPU-hour
Suited workloads
Key differences
- The RTX 4000 Ada has 1.3× the GPU memory of the RTX A4000 (20 GB vs 16 GB).
Specifications
| Specification | RTX 4000 Ada | RTX A4000 |
|---|---|---|
| GPU memory | 20 GB (higher) | 16 GB |
| Memory type | GDDR6 | GDDR6 |
| Memory bandwidth | Not published | 448 GB/s |
| Max power | 130 W | 140 W |
| Architecture | Ada Lovelace | Ampere |
| Interconnect | PCIe 4.0 | PCIe 4.0 |
The higher value of each measured row is shown in bold. Manufacturer specifications, per GPU. Tensor throughput is shown without sparsity. Values the manufacturer does not publish for this exact model are left out.
What fits in memory, on one GPU
| Workload | RTX 4000 Ada | RTX A4000 |
|---|---|---|
| LLM inference, 16-bit | ≈ 8B parameters | ≈ 6B parameters |
| LLM inference, 4-bit | ≈ 25B parameters | ≈ 20B parameters |
| Full training, mixed precision | ≈ 1B parameters | Under 1B parameters |
Cautious estimates counting 80% of memory for model weights; context length, batch size and framework change the real limit. Several GPUs together hold larger models.
RTX 4000 Ada vs RTX A4000: frequently asked questions
Which has more memory, the RTX 4000 Ada or the RTX A4000?
The RTX 4000 Ada has 20 GB of GPU memory, against 16 GB for the RTX A4000.
Can I rent the RTX 4000 Ada and the RTX A4000 by the hour?
Yes, when sellers list them on the EverXCloud spot exchange: you buy GPU hours at the live price from a prepaid balance in euros, US dollars or pounds. Prices are shown excluding tax and EverXCloud is the merchant of record for every rental.
Be ready on launch day
Create your account now: you can top up your balance and rent the GPU that fits your workload as soon as the market opens.
Have GPU servers?
Businesses can join the EverXCloud Network and sell hours of their GPU servers on the spot exchange.