A100 PCIe 40GB Specifications
Original A100. 40 GB is tight for modern model weights — check fit before buying used.
Buy it for
Adding Ampere capability to standard servers, and for MIG — it partitions into up to seven isolated instances, which suits shared research clusters and multi-tenant inference.
Look elsewhere if
You are serving models above roughly 20B parameters in FP16. 40 GB is the binding constraint, and the 80 GB version usually costs little more used.
Will it fit your system?
Passively cooled — needs server airflow
This card has no fans. It relies on a chassis forcing air through its heatsink. In a normal desktop it will throttle within minutes and can overheat outright. It needs a rack server with high-static-pressure fans, or an aftermarket shroud and blower.
- Form factor
- PCIe dual-slot
- Cooling
- passive (no fans)
- Power connectors
- 1 × 8-pin EPS (CPU-style — not a PCIe 8-pin)
- Board power
- 250 W
- Suggested PSU
- 650 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 A100 PCIe 40GB 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
600 GB/s GPU-to-GPU
Compute-only die — no ray tracing hardware
Corrects single-bit errors — expected for long training runs
What fits in memory
Largest model that fits on one A100 PCIe 40GB, assuming ~20% of the 40 GB is reserved for KV cache and activations.
| Model | FP16 / BF16 | INT4 / FP4 |
|---|---|---|
| Llama 8B | Fits | Fits |
| Llama 70B | 5× GPUs | 2× GPUs |
| Mixtral 8x22B | 9× GPUs | 3× GPUs |
| Llama 405B | 26× GPUs | 7× 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 (3,066 kWh/year).
At the US average rate, three years of continuous operation costs $1,104 — roughly 11% of the card's $10,000 launch price, before any cooling or hosting fees.
Availability & market
The weakest value in the A100 family now. Enough cheap 80 GB cards exist that the 40 GB rarely justifies its discount.
Discontinued. Expect secondary-market stock only, with no manufacturer warranty and limited driver support ahead.
Memory
- Capacity
- 40 GB
- Type
- HBM2
- Bandwidth
- 1,555 GB/s
- Bus width
- 5,120-bit
- GB per TFLOPMemory relative to compute — higher favours large models
- 0.128
Compute
- FP64
- 9.7 TFLOPS
- FP32
- 19.5 TFLOPS
- TF32 dense
- 156 TFLOPS
- FP16 dense
- 312 TFLOPS
- FP16 sparseRequires structured sparsity; rarely reached in practice
- 624 TFLOPS
- FP8 dense
- —
- INT8 dense
- 624 TOPS
Silicon
- Codename
- GA100
- Process node
- TSMC 7N
- Transistors
- 54.2B
- Die size
- 826 mm²
- CUDA cores
- 6,912
- Tensor cores
- 432
- SM count
- 108
- Boost clock
- 1,410 MHz
Platform & power
- TDP
- 250 W
- TFLOPS per wattDense FP16 ÷ TDP — drives your power bill at scale
- 1.25
- Form factor
- PCIe dual-slot
- Cooling
- passive
- Interconnect
- NVLink bridge
- Interconnect speed
- 600 GB/s
- PCIe generation
- Gen 4
- MIG
- Yes — up to 7 instances
- Launch
- 2020-05
- MSRP at launch
- $10,000
A100 PCIe 40GB — 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 A100 PCIe 40GB have?
- The A100 PCIe 40GB has 40 GB of HBM2 memory with 1,555 GB/s of bandwidth. In practice plan on roughly 32 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 A100 PCIe 40GB run?
- On a single A100 PCIe 40GB, about 16B parameters at FP16, or roughly 64B 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 A100 PCIe 40GB in a desktop PC?
- Only with added cooling. The A100 PCIe 40GB is a passively cooled card — it has no fans at all and depends on a rack chassis forcing air through its heatsink. Dropped into an ordinary desktop it will throttle within minutes and can overheat outright. It needs either a rack server with high-static-pressure fans or an aftermarket shroud and blower.
- How much power does the A100 PCIe 40GB use?
- The A100 PCIe 40GB is rated at 250 W board power. It is fed through 1 × 8-pin EPS (CPU-style — not a PCIe 8-pin). For a single-card system, a 650 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 A100 PCIe 40GB?
- At the US commercial average of $0.12/kWh, running one A100 PCIe 40GB continuously costs about $368 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 11% of what the card cost at launch.
- Does the A100 PCIe 40GB support MIG?
- Yes. The A100 PCIe 40GB can be partitioned into up to 7 Multi-Instance GPU slices, each with isolated memory and compute. This is what makes it practical to share one card across several users or tenants without them affecting each other's latency.
- Does the A100 PCIe 40GB support FP8?
- No. The A100 PCIe 40GB 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 A100 PCIe 40GB support FP4?
- No. The A100 PCIe 40GB 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 A100 PCIe 40GB support INT8?
- Yes. The A100 PCIe 40GB supports INT8 acceleration at 624 dense TOPS. INT8 remains the most widely supported quantised inference format across frameworks.
- Does the A100 PCIe 40GB support BF16?
- Yes. The A100 PCIe 40GB 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 A100 PCIe 40GB have NVLink?
- Yes. The A100 PCIe 40GB supports NVLink (NVLink bridge) at 600 GB/s, so GPU-to-GPU traffic bypasses PCIe. This matters for tensor-parallel serving, where every token crosses the link.
- Does the A100 PCIe 40GB have RT cores?
- No. The A100 PCIe 40GB is a compute-only part with no ray-tracing hardware. For rendering or virtual workstations, look at the L40S or an RTX workstation card instead.
- Does the A100 PCIe 40GB have ECC memory?
- Yes. The A100 PCIe 40GB has ECC, so single-bit memory errors are corrected rather than silently corrupting a long run.
- Is the A100 PCIe 40GB worth buying?
- Buy it for: Adding Ampere capability to standard servers, and for MIG — it partitions into up to seven isolated instances, which suits shared research clusters and multi-tenant inference. Look elsewhere if: You are serving models above roughly 20B parameters in FP16. 40 GB is the binding constraint, and the 80 GB version usually costs little more used.