Tesla P100 SXM2 Specifications
Two firsts that still define the category: HBM stacked memory and NVLink. Also the first Tesla with packed FP16, which is what made mixed-precision training practical.
Buy it for
Nothing you can buy standalone. Historically important as the first HBM2 Tesla and the first with NVLink, which is where multi-GPU scaling stopped being bottlenecked by PCIe.
Look elsewhere if
You are shopping for a card. SXM2 is a bare module that mounts to an OEM baseboard — there is no way to install it in a normal system.
Will it fit your system?
This is not a card you can install
SXM2 is a socketed module that mounts to an OEM baseboard, not a PCIe slot. You buy it inside a complete system — it cannot go into an existing desktop or a standard server under any circumstances.
- Form factor
- SXM2
- Cooling
- passive
- Power connectors
- None — SXM2 module, powered by the baseboard
- Board power
- 300 W
- Interface
- PCIe Gen 3
Supported formats & features
What the Tesla P100 SXM2 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
160 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 Tesla P100 SXM2, assuming ~20% of the 16 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.
Workload character
Plenty of bandwidth per unit of compute. Memory-bound work — single-stream token decode above all — runs close to its ceiling, and compute is usually what runs out first.
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
Modules appear cheaply from decommissioned DGX-1 systems. Without the matching baseboard they are inert.
Discontinued around 2021. Expect secondary-market stock only, with no manufacturer warranty and limited driver support ahead.
Memory
- Capacity
- 16 GB
- Type
- HBM2
- Bandwidth
- 732 GB/s
- Bus width
- 4,096-bit
- GB per TFLOPMemory relative to compute — higher favours large models
- 0.755
Compute
- FP64
- 5.3 TFLOPS
- FP32
- 10.6 TFLOPS
- TF32 dense
- —
- FP16 dense
- 21.2 TFLOPS
- FP16 sparseRequires structured sparsity; rarely reached in practice
- —
- FP8 dense
- —
- INT8 dense
- —
Silicon
- Codename
- GP100
- Process node
- TSMC 16nm FinFET
- Transistors
- 15.3B
- Die size
- 610 mm²
- CUDA cores
- 3,584
- Tensor cores
- —
- SM count
- 56
- Boost clock
- 1,480 MHz
Platform & power
- TDP
- 300 W
- TFLOPS per wattDense FP16 ÷ TDP — drives your power bill at scale
- 0.07
- Form factor
- SXM2
- Cooling
- passive
- Interconnect
- NVLink 1.0
- Interconnect speed
- 160 GB/s
- PCIe generation
- Gen 3
- MIG
- No
- Launch
- 2016-04
- MSRP at launch
- —
Tesla P100 SXM2 — 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 P100 SXM2 have?
- The Tesla P100 SXM2 has 16 GB of HBM2 memory with 732 GB/s of bandwidth. In practice plan on roughly 12 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 P100 SXM2 run?
- On a single Tesla P100 SXM2, about 6B parameters at FP16. 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 Tesla P100 SXM2 in a desktop PC?
- No. The Tesla P100 SXM2 is a SXM2 module, not a PCIe card. It bolts to an OEM baseboard and is only sold inside a complete system, so there is no way to install one in a desktop or a standard server.
- How much power does the Tesla P100 SXM2 use?
- The Tesla P100 SXM2 is rated at 300 W board power. It draws that from the baseboard rather than from PSU cables, so there is nothing to wire up yourself.
- How much does it cost to run the Tesla P100 SXM2?
- At the US commercial average of $0.12/kWh, running one Tesla P100 SXM2 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 P100 SXM2 support FP8?
- No. The Tesla P100 SXM2 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 P100 SXM2 support FP4?
- No. The Tesla P100 SXM2 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 P100 SXM2 support INT8?
- No. The Tesla P100 SXM2 has no INT8 tensor path, so quantised inference gives you a smaller memory footprint but no arithmetic speedup.
- Does the Tesla P100 SXM2 support BF16?
- No. The Tesla P100 SXM2 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 P100 SXM2 have NVLink?
- Yes. The Tesla P100 SXM2 supports NVLink (NVLink 1.0) at 160 GB/s, so GPU-to-GPU traffic bypasses PCIe. This matters for tensor-parallel serving, where every token crosses the link.
- Does the Tesla P100 SXM2 have RT cores?
- No. The Tesla P100 SXM2 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 P100 SXM2 have ECC memory?
- Yes. The Tesla P100 SXM2 has ECC, so single-bit memory errors are corrected rather than silently corrupting a long run.
- What replaced the Tesla P100 SXM2?
- The Tesla V100 PCIe 32GB (2018) is its successor, with 32 GB of HBM2 at 900 GB/s. It is about 5.3x faster in dense FP16. On the used market that gap is what sets a fair price for the Tesla P100 SXM2.
- Is the Tesla P100 SXM2 worth buying?
- Buy it for: Nothing you can buy standalone. Historically important as the first HBM2 Tesla and the first with NVLink, which is where multi-GPU scaling stopped being bottlenecked by PCIe. Look elsewhere if: You are shopping for a card. SXM2 is a bare module that mounts to an OEM baseboard — there is no way to install it in a normal system.