Tesla K80 Specifications
Two GK210 dies on one board, presented as two GPUs. The default free-tier GPU on Colab and Kaggle for years, which made it many people's first AI accelerator.
Buy it for
Nothing worth recommending, despite being the cheap GPU that shows up in every "budget CUDA" thread. Its 24 GB is two separate 12 GB GPUs on one board, so no single job ever sees more than 12 GB.
Look elsewhere if
You are buying on the 24 GB figure, or you want to run PyTorch. Kepler support was removed from PyTorch years ago, and building from source for compute capability 3.7 is a project in itself.
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
- 300 W
- Suggested PSU
- 750 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 K80 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
Not 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 K80, 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 (3,679 kWh/year).
Where it sits in the line
A newer part existing does not make this one a bad buy — it makes it cheaper. What matters is price per unit of the throughput you actually need.
Availability & market
Sells for very little and is still overpriced. 300 W passive, needs forced server airflow, and the split memory undermines the one number people buy it for.
Discontinued around 2019. Expect secondary-market stock only, with no manufacturer warranty and limited driver support ahead.
Memory
- Capacity
- 24 GB
- Type
- GDDR5
- Bandwidth
- 480 GB/s
- Bus width
- 768-bit
Compute
- FP64
- 2.9 TFLOPS
- FP32
- 8.7 TFLOPS
- TF32 dense
- —
- FP16 dense
- —
- FP16 sparseRequires structured sparsity; rarely reached in practice
- —
- FP8 dense
- —
- INT8 dense
- —
Silicon
- Codename
- GK210
- Process node
- TSMC 28nm
- Transistors
- 7.1B
- Die size
- 561 mm²
- CUDA cores
- 4,992
- Tensor cores
- —
- SM count
- —
- Boost clock
- 875 MHz
Platform & power
- TDP
- 300 W
- Form factor
- PCIe dual-slot
- Cooling
- passive
- Interconnect
- None
- Interconnect speed
- —
- PCIe generation
- Gen 3
- MIG
- No
- Launch
- 2014-11
- MSRP at launch
- —
Tesla K80 — 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 K80 have?
- The Tesla K80 has 24 GB of GDDR5 memory with 480 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 K80 run?
- On a single Tesla K80, 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 K80 in a desktop PC?
- Only with added cooling. The Tesla K80 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 K80 use?
- The Tesla K80 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 Tesla K80?
- At the US commercial average of $0.12/kWh, running one Tesla K80 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.
- Does the Tesla K80 support FP8?
- No. The Tesla K80 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 K80 support FP4?
- No. The Tesla K80 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 K80 support INT8?
- No. The Tesla K80 has no INT8 tensor path, so quantised inference gives you a smaller memory footprint but no arithmetic speedup.
- Does the Tesla K80 support BF16?
- No. The Tesla K80 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 K80 have NVLink?
- No. The Tesla K80 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 K80 have RT cores?
- No. The Tesla K80 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 K80 have ECC memory?
- Yes. The Tesla K80 has ECC, so single-bit memory errors are corrected rather than silently corrupting a long run.
- What replaced the Tesla K80?
- The Tesla M40 (2015) is its successor, with 24 GB of GDDR5 at 288 GB/s.. On the used market that gap is what sets a fair price for the Tesla K80.
- Is the Tesla K80 worth buying?
- Buy it for: Nothing worth recommending, despite being the cheap GPU that shows up in every "budget CUDA" thread. Its 24 GB is two separate 12 GB GPUs on one board, so no single job ever sees more than 12 GB. Look elsewhere if: You are buying on the 24 GB figure, or you want to run PyTorch. Kepler support was removed from PyTorch years ago, and building from source for compute capability 3.7 is a project in itself.