Calculate the TPS of the P102-100 on local AI models
Every model in our catalogue assessed against this card at the context length and minimum quality you choose. Speed is an estimate for a single request, calculated from this card's memory bandwidth and the size of each model once compressed.
Calculated for this card
679 models in our catalogue altogether
Largest model it holds
Sailor-7B-Chat
7.7B · Q3_K_M · 55.4 tok/s
Fastest model
Gemma 3 QAT 1B
159 tok/s · 1B
Which AI models can run on a P102-100?
Set the inputs, read the answer
More context means more memory for the conversation cache. Speed is for a fresh conversation and does not change with this setting.
Hides models that would only fit by being compressed below this point.
203 models match
Calculating| Quantisation | Fit | ||||||
|---|---|---|---|---|---|---|---|
|
159
tok/s
55–317 · low confidence |
Gemma 3 1B | 1B | Mar 2025 | 1.8 GB | 33k tokens | Q8_0 | Comfortable |
|
159
tok/s
55–317 · low confidence |
Gemma 3 QAT 1B | 1B | Apr 2025 | 1.8 GB | 33k tokens | Q8_0 | Comfortable |
|
159
tok/s
55–317 · low confidence |
HGRN 1B (WT 103) ≈ | 1B | Nov 2023 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
159
tok/s
55–317 · low confidence |
LLama 3..2 Typhoon 2 1B ≈ | 1B | Dec 2024 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
159
tok/s
55–317 · low confidence |
OLMo-1B ≈ | 1B | Feb 2024 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
159
tok/s
55–317 · low confidence |
Pythia-1b ≈ | 1B | Apr 2023 | 1.8 GB | 131k tokens ? | Q8_0 | Comfortable |
|
147
tok/s
51–294 · low confidence |
OpenELM-1.1B ≈ | 1.1B | May 2024 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
144
tok/s
50–288 · low confidence |
DeciCoder-1B ≈ | 1.1B | Aug 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
144
tok/s
50–288 · low confidence |
SantaCoder ≈ | 1.1B | Jan 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
144
tok/s
50–288 · low confidence |
TinyLlama-1.1B (1T token checkpoint) ≈ | 1.1B | Oct 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
144
tok/s
50–288 · low confidence |
TinyLlama-1.1B (3T token checkpoint) ≈ | 1.1B | Oct 2023 | 1.9 GB | 131k tokens ? | Q8_0 | Comfortable |
|
132
tok/s
46–264 · low confidence |
EXAONE 4.0 (1.2B) ≈ | 1.2B | Jul 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
132
tok/s
46–264 · low confidence |
MinerU2.5 ≈ | 1.2B | Sep 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
132
tok/s
46–264 · low confidence |
Pleias 1.0 1.2B ≈ | 1.2B | Dec 2024 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
132
tok/s
46–264 · low confidence |
Pleias-RAG-1B ≈ | 1.2B | Apr 2025 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
129
tok/s
45–258 · low confidence |
Llama 3.2 1B | 1.2B | Sep 2024 | 2.2 GB | 84k tokens | Q8_0 | Comfortable |
|
127
tok/s
44–254 · low confidence |
MiniCPM-1.2B ≈ | 1.2B | Jun 2024 | 2.0 GB | 131k tokens ? | Q8_0 | Comfortable |
|
122
tok/s
43–244 · low confidence |
DeepSeek Coder 1.3B ≈ | 1.3B | Jan 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
122
tok/s
43–244 · low confidence |
DeepSeek-VL-1.3B ≈ | 1.3B | Mar 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
122
tok/s
43–244 · low confidence |
DigiRL ≈ | 1.3B | Jun 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
122
tok/s
43–244 · low confidence |
GLA Transformer 1.3B ≈ | 1.3B | Aug 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
122
tok/s
43–244 · low confidence |
Janus 1.3B ≈ | 1.3B | Oct 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
122
tok/s
43–244 · low confidence |
Kosmos-2.5 ≈ | 1.3B | Aug 2024 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
122
tok/s
43–244 · low confidence |
Otter ≈ | 1.3B | May 2023 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
|
122
tok/s
43–244 · low confidence |
Phi-1 ≈ | 1.3B | Oct 2023 | 2.1 GB | 131k tokens ? | Q8_0 | Comfortable |
Speeds are estimates for a single request — one conversation at a time — calculated from memory bandwidth, model size and quantisation. Real throughput varies with the inference runtime and its version. Figures published by hardware vendors measure many simultaneous requests and are much higher.
On record
P102-100 full specification
Everything on record for this board, ordered by how much it bears on running a language model rather than by how a spec sheet would list it. Memory comes first because it decides the outcome; the rest is context.
Memory
The two specifications that decide what this card can run and how quickly. Capacity sets which models fit; bandwidth sets how many tokens per second they produce once they do.
- Memory size
- 5 GB
- Memory bandwidth
- 440 GB/s
- Memory type
- GDDR5X
- Memory bus width
- 320 bit
- Memory clock
- 1.38 GHz
The chip
Which processor is on the board and how it was manufactured. A smaller process size generally means more performance for the same power.
- Graphics processor
- GP102
- Architecture
- Pascal
- Generation
- Mining GPUs
- Foundry
- TSMC
- Process size
- 16 nm
- Transistors
- 11.8 billion
- Transistor density
- 25,100 K/mm²
- Die size
- 471 mm²
- Package
- BGA-2397
- Released
- 12 February 2018
Clock speeds
How fast the processor runs. Worth far less here than on a gaming benchmark: generating text is limited by memory bandwidth, so a higher clock barely moves the result.
- Base clock
- 1.58 GHz
- Boost clock
- 1.68 GHz
Processing units
What the chip contains. These drive graphics performance and matter mainly for processing a long prompt rather than for producing the answer.
- Shading units
- 3,200
- Texture mapping units
- 200
- Render output units
- 80
- Streaming multiprocessors
- 25
- L1 cache
- 48 KB
- L2 cache
- 2.5 MB
Theoretical performance
Peak arithmetic rates published for the board. These are ceilings that no real workload reaches, and generating text reaches a small fraction of them because it is limited by memory rather than arithmetic.
- Half precision (FP16)
- 168.3 GFLOPS
- Single precision (FP32)
- 10.8 TFLOPS
- Double precision (FP64)
- 336.6 GFLOPS
- Pixel rate
- 135 GPixel/s
- Texture rate
- 337 GTexel/s
The board
What it takes to physically install and power the card — the practical constraints that decide whether it fits the machine you already own.
- Power draw (TDP)
- 250 W
- Suggested power supply
- 600 W
- Power connectors
- 2x 8-pin
- Bus interface
- PCIe 1.0 x4
- Slot width
- Dual-slot
- Dimensions
- 267 mm
Software support
Which graphics and compute interfaces the card supports. CUDA compute capability is the one that bears on inference: below 7.0 there are no tensor cores, and modern inference software falls back to slower code paths.
- CUDA compute capability
- 6.1
- DirectX
- 12.1
- OpenGL
- 4.6
- Vulkan
- 1.4
- OpenCL
- 3.0
- Shader model
- 6.8
Listings
Where to buy a P102-100
No vendor is currently listing this card. Listings come from vendors who publish them here directly — browse the vendor directory to see who is selling what.
What the numbers mean
Memory: the specification that decides everything
Memory
5 GB
Bandwidth
440 GB/s
Largest model
Sailor-7B-Chat
At 5 GB of GDDR5X the P102-100 is limited to the smaller end of the catalogue. About 4.5 GB is actually available to a runtime, and a model has to fit entirely inside it before generating anything at all.
The memory bus moves 440 GB/s across a 320-bit bus. That is the number that governs generation speed — arithmetic per byte read is small enough that the bus, not the cores, is what everything waits on.
The figure is the memory clock — 1.38 GHz here — multiplied by the bus width. It is why core counts predict generation speed so poorly.
Put together, the largest model that fits is Sailor-7B-Chat at 7.7B, running Q3_K_M and producing around 55.4 tokens per second.
The chip and how it was built
The P102-100 is built on the GP102 graphics processor, using NVIDIA's Pascal architecture, as part of the Mining GPUs generation.
The chip is manufactured by TSMC, on a 16 nm process, with a die measuring 471 mm², holding 11.8 billion transistors. A smaller process generally means more performance for the same power, though for language models it matters far less than the memory subsystem.
It was released in February 2018, roughly 8 years ago. Inference software support tends to follow hardware by a year or two, so a card of this age generally has mature, well-optimised code paths available to it.
Compute throughput, and why it matters less than it looks
FP16
168.3 GFLOPS
FP64
336.6 GFLOPS
On paper the P102-100 reaches 168.3 GFLOPS at half precision and 10.8 TFLOPS at single precision. These are peak figures no real workload sustains, and generating text reaches only a small fraction of them — decoding is limited by memory rather than arithmetic, which is why a card can look enormously powerful here and still produce tokens at an ordinary rate.
Double-precision throughput is 336.6 GFLOPS. It has no bearing on running a language model — no inference runtime uses it — but it separates datacentre parts from consumer ones, since the latter deliberately restrict it.
Clocks run from 1.58 GHz at base to 1.68 GHz boosted. Worth far less here than on a gaming benchmark: raising the clock speeds up the arithmetic, and the arithmetic is not what generation is waiting on.
Cache and processing units
The P102-100 has 48 KB of L1 cache, backed by 2.5 MB of L2. Cache absorbs a share of the memory traffic that would otherwise hit the main bus, which is the one place on this page where a number other than bandwidth quietly affects generation speed — a large L2 lets more of the working set stay close to the cores.
There are 3,200 shading units, 200 texture mapping units, and 80 render output units. These drive graphics workloads and contribute to prompt processing, but they sit idle for much of the time a model spends generating a reply.
Power, size and installation
Power draw
250 W
The P102-100 is rated at 250 W, with a 600 W power supply suggested for the whole system. Running a language model keeps a card busy in bursts rather than continuously — it draws hard while generating and idles between requests — so sustained draw over a working day is usually well below the rated figure.
The board occupies a dual-slot, measuring 267 mm long, and needs 2x 8-pin. Worth checking against the case and power supply already in the machine, since the largest cards need considerably more of both than a typical desktop provides.
It connects over PCIe 1.0 x4. The interface governs how quickly a model is loaded from disk into the card, not how fast it runs once there, so a narrower link costs a few seconds at startup and nothing thereafter.
The extremes
The largest AI models that run on a P102-100
The biggest open-weight models that fit on this card, newest first. Each is shown at the best compression the card can hold.
The fastest AI models on a P102-100
Where this card produces tokens quickest. Smaller models dominate here, because generating each token means reading the whole model out of memory once.
Step by step
How to work out the tokens per second of a P102-100
You do not have to calculate anything by hand — the gputps.com calculator on this page has already worked it out for every model this card can hold. Reading off the answer takes six steps.
-
01
Find the model in the table
The table lists 203 models this P102-100 can run. Search by name, or by size — typing 27b matches on the parameter count even when the name never states it.
-
02
Decide how long your conversations run
Longer conversations cost memory on top of the weights. With 5 GB to work in, that is frequently the difference between a model fitting and not.
-
03
Choose how far you will compress
By default the table picks the least-compressed copy that fits. Setting a floor removes models that only qualify through heavy compression.
-
04
Read the speed and the range
The figures are calculated, not measured. 159 tok/s on Gemma 3 QAT 1B is the fastest result on this card, and like every row it carries a range that reflects how much the runtime matters.
-
05
Read the fit verdict last
A tight fit runs but leaves no room to raise the context later; comfortable has headroom. The memory column shows what each model needs against the 5 GB available.
-
06
Check the same model from the other side
Every model name in the table links to its own page, which runs the same calculation across every card we hold. That is where you see whether the P102-100 is the right buy for it or merely a card that fits.
Answers
P102-100 — common questions
Does the P102-100 support CUDA?
Yes. The P102-100 reports CUDA compute capability 6.1, which predates tensor cores. Capability 7.0 and above has tensor cores, which modern inference software uses; below that it falls back to slower code paths for quantised models.
What bus interface does the P102-100 use?
It uses PCIe 1.0 x4. This governs how fast a model is loaded onto the card rather than how fast it runs once loaded, so it costs a few seconds at startup and nothing during generation.
Is the P102-100 good for running local AI models?
Its memory limits it to smaller models and its bandwidth gives usable, if unspectacular, generation speeds. In total it runs 203 of the models we track. Whether that is enough depends entirely on which model you want — the table above answers that directly.
Can a P102-100 run a model that does not fit in its memory?
Only partly. Layers beyond the 5 GB sit in system memory and run at a fraction of the speed, so a mostly-offloaded model is rarely worth using. Every figure here assumes it is fully resident on the card.
Would two P102-100 cards be twice as fast?
Pairing P102-100 cards buys headroom rather than pace: 10 GB of combined memory, at roughly the same generation speed as one.
What AI models can a P102-100 run?
203 of the 679 open-weight language models we track fit on a P102-100 and can be run locally on it. The table on this page lists every one, with the memory it needs, the quantisation it runs at and an estimated generation speed.
What is the largest AI model a P102-100 can run?
The largest model in our catalogue that fits on a P102-100 is Sailor-7B-Chat at 7.7B parameters, compressed to Q3_K_M. It generates roughly 55.4 tokens per second and needs about 4.5 GB of the card's memory.
How many tokens per second does a P102-100 produce?
It depends on the model. On a P102-100 the fastest model we track is Gemma 3 QAT 1B at about 159 tokens per second, while larger models run proportionally slower because each token requires reading the whole model out of memory once. Speeds are estimates for a single conversation at a time.
Can a P102-100 run a 7B model?
Yes. For example a P102-100 runs Olmo 3 7B Think at Q3_K_M, using about 4.1 GB of memory and generating around 61.1 tokens per second.
How much memory does a P102-100 have?
A P102-100 has 5 GB of GDDR5X memory. Around a tenth of that is reserved by the inference runtime and the driver, leaving roughly 4.5 GB available for a model and its conversation.
What is the memory bandwidth of a P102-100?
The P102-100 has 440 GB/s of memory bandwidth, across a 320-bit memory bus. This is the single best predictor of how fast it generates text, because producing each token means reading the entire model out of memory once.
What type of memory does a P102-100 use?
It uses GDDR5X clocked at 1.38 GHz. HBM types are found on datacentre accelerators and carry far more bandwidth than the GDDR used on desktop cards, which is why they generate tokens considerably faster at the same capacity.
Who makes the P102-100?
The P102-100 is a NVIDIA product, with the chip manufactured by TSMC, on a 16 nm process.
When was the P102-100 released?
The P102-100 was released in February 2018.
How much power does a P102-100 use?
The P102-100 has a rated board power of 250 W, and a 600 W system power supply is suggested. Generating text draws hard in bursts and idles between requests, so average consumption over a working session is normally well below the rated figure.
How much cache does a P102-100 have?
The P102-100 has 48 KB of L1 cache, and 2.5 MB of L2 cache. Cache absorbs part of the memory traffic that would otherwise reach the main bus, so a larger L2 gives a modest lift to generation speed beyond what bandwidth alone predicts.
What are the TFLOPS of a P102-100?
The P102-100 is rated at 168.3 GFLOPS at half precision and 10.8 TFLOPS at single precision. These are peak arithmetic ceilings rather than achievable rates, and text generation reaches only a small fraction of them because it is limited by memory bandwidth instead.
The other direction
Looking at it from the other side?
This page starts from the hardware. If you already know which model you want and need to know what it takes to run it, start from the model instead.