A6000 Specifications
48 GB ECC in a workstation card. Popular for local fine-tuning where a datacenter GPU will not fit or cool.
Buy it for
A 48 GB workstation card you can put under a desk. Actively cooled, ECC, display outputs, and no server chassis required — the usual reason people pick it over an A40.
Look elsewhere if
You are cost-sensitive and have server airflow, in which case the A40 is the same silicon for less, or you need current-generation speed, where the RTX 6000 Ada is roughly twice as fast.
Will it fit your system?
- Form factor
- PCIe dual-slot
- Cooling
- active
- Power connectors
- 1 × 8-pin EPS (CPU-style — not a PCIe 8-pin)
- Board power
- 300 W
- Suggested PSU
- 750 W or better
- Interface
- PCIe Gen 4
PSU figure assumes one card plus a mainstream CPU and drives, with headroom for transient spikes — modern cards can draw well above their rated TDP for milliseconds, which is what trips an undersized supply.
Supported formats & features
What the A6000 does and does not support in hardware.
Blackwell-generation feature — B200 has it
Predates hardware FP8; use INT8 or BF16 instead — H100 has it
Wider exponent than FP16 — fewer overflow issues when training
Drop-in FP32 speedup with no code changes
Quantised inference supported
INT4 tensor operations supported
Precision must be managed manually
112 GB/s GPU-to-GPU
Hardware ray tracing for rendering workloads
Corrects single-bit errors — expected for long training runs
What fits in memory
Largest model that fits on one A6000, assuming ~20% of the 48 GB is reserved for KV cache and activations.
| Model | FP16 / BF16 | INT4 / FP4 |
|---|---|---|
| Llama 8B | Fits | Fits |
| Llama 70B | 4× GPUs | Fits |
| Mixtral 8x22B | 8× GPUs | 2× GPUs |
| Llama 405B | 22× GPUs | 6× GPUs |
Weights only. Long contexts, large batch sizes and optimiser state during training all need substantially more — full fine-tuning typically needs 3–4× the weight memory.
Workload character
Bandwidth and compute are matched closely enough that neither dominates across mixed training and inference.
Cost per unit of compute
Breakeven counts hardware cost against rental only. Power, cooling, hosting and depreciation push the real crossover substantially later — see colocation pricing for what hosting actually costs.
Cost to run
Electricity for one card at 100% duty, including cooling and PSU overhead (3,679 kWh/year).
At the US average rate, three years of continuous operation costs $1,325 — roughly 28% of the card's $4,650 launch price, before any cooling or hosting fees.
Availability & market
Holds value well because desktop-installable 48 GB remains rare. Used prices are reasonable now that the Ada version is established.
Memory
- Capacity
- 48 GB
- Type
- GDDR6 ECC
- Bandwidth
- 768 GB/s
- Bus width
- 384-bit
- GB per TFLOPMemory relative to compute — higher favours large models
- 0.310
Compute
- FP64
- —
- FP32
- 38.7 TFLOPS
- TF32 dense
- —
- FP16 dense
- 154.8 TFLOPS
- FP16 sparseRequires structured sparsity; rarely reached in practice
- 309.7 TFLOPS
- FP8 dense
- —
- INT8 dense
- —
Silicon
- Codename
- GA102
- Process node
- Samsung 8N
- Transistors
- 28.3B
- Die size
- 628 mm²
- CUDA cores
- 10,752
- Tensor cores
- 336
- SM count
- 84
- Boost clock
- 1,800 MHz
Platform & power
- TDP
- 300 W
- TFLOPS per wattDense FP16 ÷ TDP — drives your power bill at scale
- 0.52
- Form factor
- PCIe dual-slot
- Cooling
- active
- Interconnect
- NVLink
- Interconnect speed
- 112 GB/s
- PCIe generation
- Gen 4
- MIG
- No
- Launch
- 2020-12
- MSRP at launch
- $4,650
A6000 — common questions
Answered from the specifications above. Questions this card's data cannot answer with confidence are left out rather than guessed.
- How much VRAM does the A6000 have?
- The A6000 has 48 GB of GDDR6 ECC memory with 768 GB/s of bandwidth. In practice plan on roughly 38 GB for weights: KV cache, activations and fragmentation take the rest, and long contexts or large batches take considerably more.
- What size model can the A6000 run?
- On a single A6000, about 19B parameters at FP16, or roughly 76B quantised to FP4. Anything larger has to be split across several GPUs. These are weight-only figures — a long context window can consume tens of gigabytes on its own.
- Can you put the A6000 in a desktop PC?
- Yes. The A6000 is a self-cooled PCIe dual-slot card and works in a normal desktop, provided the case has clearance and the power supply is rated around 750 W or better.
- How much power does the A6000 use?
- The A6000 is rated at 300 W board power. It is fed through 1 × 8-pin EPS (CPU-style — not a PCIe 8-pin). For a single-card system, a 750 W supply or better leaves headroom for the CPU, drives and the transient spikes that trip undersized units.
- How much does it cost to run the A6000?
- At the US commercial average of $0.12/kWh, running one A6000 continuously costs about $442 a year in electricity. That figure includes cooling and power-supply overhead at a factor of 1.4 over board power, and over three years comes to roughly 28% of what the card cost at launch.
- Does the A6000 support FP8?
- No. The A6000 has no hardware FP8 path. FP8-quantised checkpoints will still load, but the values are upcast before they reach the tensor cores, so you get the memory saving without the speedup.
- Does the A6000 support FP4?
- No. The A6000 predates hardware FP4. You can run 4-bit quantised weights on it and the memory saving is real, but they are dequantised before compute, so there is no arithmetic speedup.
- Does the A6000 support INT8?
- Yes. The A6000 supports INT8 acceleration. INT8 remains the most widely supported quantised inference format across frameworks.
- Does the A6000 support BF16?
- Yes. The A6000 supports bfloat16. Its wider exponent range makes training far less prone to the overflow that plagues FP16, which is why most current training recipes assume it.
- Does the A6000 have NVLink?
- Yes. The A6000 supports NVLink (NVLink) at 112 GB/s, so GPU-to-GPU traffic bypasses PCIe. This matters for tensor-parallel serving, where every token crosses the link.
- Does the A6000 have RT cores?
- Yes. The A6000 includes dedicated ray-tracing cores, so it is usable for rendering and visualisation as well as compute.
- Does the A6000 have ECC memory?
- Yes. The A6000 has ECC, so single-bit memory errors are corrected rather than silently corrupting a long run.
- Is the A6000 worth buying?
- Buy it for: A 48 GB workstation card you can put under a desk. Actively cooled, ECC, display outputs, and no server chassis required — the usual reason people pick it over an A40. Look elsewhere if: You are cost-sensitive and have server airflow, in which case the A40 is the same silicon for less, or you need current-generation speed, where the RTX 6000 Ada is roughly twice as fast.