Tesla P40 Specifications
First Tesla built for inference rather than training, with INT8 throughput as the headline. Still cheap on the used market for 24 GB of VRAM.
Buy it for
Budget local LLM inference. 24 GB and cheap, which is why it remains a staple of home inference builds long after its datacentre life ended.
Look elsewhere if
Your stack depends on fast FP16. The P40 deliberately runs FP16 at 1/64 rate, so it must be used in FP32 or INT8 — many llama.cpp and vLLM paths are slower on it than the raw specs suggest.
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 3
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 Tesla P40 does and does not support in hardware.
Blackwell-generation feature — B200 has it
Predates hardware FP8; use INT8 or BF16 instead — H100 has it
FP16 only, so training recipes may need loss scaling
Not available on this architecture
Quantised inference supported
Not supported on this generation
Precision must be managed manually
Multi-GPU traffic goes over PCIe — a real limit for distributed training
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 Tesla P40, assuming ~20% of the 24 GB is reserved for KV cache and activations.
| Model |
|---|
| Llama 8B |
| Llama 70B |
| Mixtral 8x22B |
| Llama 405B |
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.
Cost to run
Electricity for one card at 100% duty, including cooling and PSU overhead (3,066 kWh/year).
Availability & market
A well-known cheap 24 GB option. Passive, so budget for a fan shroud, and check the seller has not stripped the cooler.
Discontinued around 2021. Expect secondary-market stock only, with no manufacturer warranty and limited driver support ahead.
Memory
- Capacity
- 24 GB
- Type
- GDDR5
- Bandwidth
- 346 GB/s
- Bus width
- 384-bit
Compute
- FP64
- —
- FP32
- 11.8 TFLOPS
- TF32 dense
- —
- FP16 dense
- —
- FP16 sparseRequires structured sparsity; rarely reached in practice
- —
- FP8 dense
- —
- INT8 dense
- 47 TOPS
Silicon
- Codename
- GP102
- Process node
- TSMC 16nm FinFET
- Transistors
- 11.8B
- Die size
- 471 mm²
- CUDA cores
- 3,840
- Tensor cores
- —
- SM count
- 30
- Boost clock
- 1,531 MHz
Platform & power
- TDP
- 250 W
- Form factor
- PCIe dual-slot
- Cooling
- passive
- Interconnect
- None
- Interconnect speed
- —
- PCIe generation
- Gen 3
- MIG
- No
- Launch
- 2016-09
- MSRP at launch
- —
Tesla P40 — 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 Tesla P40 have?
- The Tesla P40 has 24 GB of GDDR5 memory with 346 GB/s of bandwidth. In practice plan on roughly 19 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 Tesla P40 run?
- On a single Tesla P40, about 9B parameters at FP16. 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 Tesla P40 in a desktop PC?
- Only with added cooling. The Tesla P40 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 Tesla P40 use?
- The Tesla P40 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 Tesla P40?
- At the US commercial average of $0.12/kWh, running one Tesla P40 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.
- Does the Tesla P40 support FP8?
- No. The Tesla P40 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 Tesla P40 support FP4?
- No. The Tesla P40 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 Tesla P40 support INT8?
- Yes. The Tesla P40 supports INT8 acceleration at 47 dense TOPS. INT8 remains the most widely supported quantised inference format across frameworks.
- Does the Tesla P40 support BF16?
- No. The Tesla P40 predates bfloat16 support and can only use FP16 for half-precision work. FP16 training on it usually needs loss scaling to stay numerically stable.
- Does the Tesla P40 have NVLink?
- No. The Tesla P40 has no NVLink. Multi-GPU work goes over PCIe Gen 3, which is an order of magnitude slower and makes tensor parallelism across cards expensive.
- Does the Tesla P40 have RT cores?
- No. The Tesla P40 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 Tesla P40 have ECC memory?
- Yes. The Tesla P40 has ECC, so single-bit memory errors are corrected rather than silently corrupting a long run.
- Is the Tesla P40 worth buying?
- Buy it for: Budget local LLM inference. 24 GB and cheap, which is why it remains a staple of home inference builds long after its datacentre life ended. Look elsewhere if: Your stack depends on fast FP16. The P40 deliberately runs FP16 at 1/64 rate, so it must be used in FP32 or INT8 — many llama.cpp and vLLM paths are slower on it than the raw specs suggest.