L40 Specifications
Graphics-oriented sibling of the L40S, with lower clocks. Aimed at Omniverse and virtual workstations rather than pure AI throughput.
Buy it for
Omniverse, rendering, and virtual workstations — the Ada datacentre card with RT cores and 48 GB, aimed at graphics rather than training.
Look elsewhere if
Your workload is inference or training. The L40S is the same card tuned for compute and is substantially faster at it for the same slot and power.
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 × 16-pin (12VHPWR)
- 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 L40 does and does not support in hardware.
Blackwell-generation feature — B200 has it
E4M3 and E5M2 — the standard format for modern inference
Wider exponent than FP16 — fewer overflow issues when training
Drop-in FP32 speedup with no code changes
Quantised inference supported
Not yet confirmed against vendor documentation.
Precision must be managed manually
Multi-GPU traffic goes over PCIe — a real limit for distributed training
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 L40, assuming ~20% of the 48 GB is reserved for KV cache and activations.
| Model | FP16 / BF16 | FP8 |
|---|---|---|
| Llama 8B | Fits | Fits |
| Llama 70B | 4× GPUs | 2× GPUs |
| Mixtral 8x22B | 8× GPUs | 4× GPUs |
| Llama 405B | 22× GPUs | 11× 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 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 18% of the card's $7,500 launch price, before any cooling or hosting fees.
Availability & market
Largely displaced by the L40S. Only worth it where the graphics and vGPU certifications are the point.
Memory
- Capacity
- 48 GB
- Type
- GDDR6 ECC
- Bandwidth
- 864 GB/s
- Bus width
- 384-bit
- GB per TFLOPMemory relative to compute — higher favours large models
- 0.265
Compute
- FP64
- —
- FP32
- 90.5 TFLOPS
- TF32 dense
- 90.5 TFLOPS
- FP16 dense
- 181 TFLOPS
- FP16 sparseRequires structured sparsity; rarely reached in practice
- 362 TFLOPS
- FP8 dense
- 362 TFLOPS
- INT8 dense
- 362 TOPS
Silicon
- Codename
- AD102
- Process node
- TSMC 4N
- Transistors
- 76.3B
- Die size
- 609 mm²
- CUDA cores
- 18,176
- Tensor cores
- 568
- SM count
- 142
- Boost clock
- 2,490 MHz
Platform & power
- TDP
- 300 W
- TFLOPS per wattDense FP16 ÷ TDP — drives your power bill at scale
- 0.60
- Form factor
- PCIe dual-slot
- Cooling
- passive
- Interconnect
- None
- Interconnect speed
- —
- PCIe generation
- Gen 4
- MIG
- No
- Launch
- 2022-10
- MSRP at launch
- $7,500
L40 — 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 L40 have?
- The L40 has 48 GB of GDDR6 ECC memory with 864 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 L40 run?
- On a single L40, about 19B parameters at FP16, or roughly 38B quantised to FP8. Anything larger has to be split across several GPUs, which is slow on this card because it has no NVLink. These are weight-only figures — a long context window can consume tens of gigabytes on its own.
- Can you put the L40 in a desktop PC?
- Only with added cooling. The L40 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 L40 use?
- The L40 is rated at 300 W board power. It is fed through 1 × 16-pin (12VHPWR). 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 L40?
- At the US commercial average of $0.12/kWh, running one L40 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 18% of what the card cost at launch.
- Does the L40 support FP8?
- Yes. The L40 has hardware FP8 tensor support, so FP8 weights and activations run at full tensor-core rate — 362 dense TFLOPS. This is the format most current inference servers default to for large models.
- Does the L40 support FP4?
- No. The L40 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 L40 support INT8?
- Yes. The L40 supports INT8 acceleration at 362 dense TOPS. INT8 remains the most widely supported quantised inference format across frameworks.
- Does the L40 support BF16?
- Yes. The L40 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 L40 have NVLink?
- No. The L40 has no NVLink. Multi-GPU work goes over PCIe Gen 4, which is an order of magnitude slower and makes tensor parallelism across cards expensive.
- Does the L40 have RT cores?
- Yes. The L40 includes dedicated ray-tracing cores, so it is usable for rendering and visualisation as well as compute.
- Does the L40 have ECC memory?
- Yes. The L40 has ECC, so single-bit memory errors are corrected rather than silently corrupting a long run.
- Is the L40 worth buying?
- Buy it for: Omniverse, rendering, and virtual workstations — the Ada datacentre card with RT cores and 48 GB, aimed at graphics rather than training. Look elsewhere if: Your workload is inference or training. The L40S is the same card tuned for compute and is substantially faster at it for the same slot and power.