NPD TPS calculator

Open weights IEEE 313.9K parameters August 2014

Each card below is assessed against this model at the context length and minimum quality you choose. Speed is an estimate for a single request, calculated from the card's memory bandwidth and the size of the model once compressed.

Calculated for this model

818 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Tesla C1080

4 GB · Q8_0 · 117,455 tok/s

Fastest card

B200

10,795,509 tok/s · 180 GB

Which GPUs can run NPD?

Set the inputs, read the answer

Context length tokens

A longer conversation needs more memory, which can push this model off smaller cards.

Minimum quality

Hides cards that would only fit the model by compressing it below this point.

Search cards

Card memory

Also show cards that cannot run it

818 cards match

Calculating

					Needs	Quantisation	Fit
10,795,509 tok/s 6,477,305–17,272,815 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	0.7 GB	Q8_0	Comfortable
10,795,509 tok/s 6,477,305–17,272,815 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	0.7 GB	Q8_0	Comfortable
8,620,484 tok/s 5,172,290–13,792,774 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
8,620,484 tok/s 5,172,290–13,792,774 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
6,894,282 tok/s 4,136,569–11,030,851 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
6,598,755 tok/s 3,959,253–10,558,008 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
6,598,755 tok/s 3,959,253–10,558,008 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
6,315,373 tok/s 3,789,224–10,104,596 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	0.7 GB	Q8_0	Comfortable
5,604,893 tok/s 3,362,936–8,967,829 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
5,604,893 tok/s 3,362,936–8,967,829 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
5,604,893 tok/s 3,362,936–8,967,829 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
5,316,788 tok/s 3,190,073–8,506,861 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
4,534,114 tok/s 2,720,468–7,254,582 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
4,534,114 tok/s 2,720,468–7,254,582 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	0.7 GB	Q8_0	Comfortable
4,534,114 tok/s 2,720,468–7,254,582 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
4,534,114 tok/s 2,720,468–7,254,582 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
4,534,114 tok/s 2,720,468–7,254,582 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
3,452,404 tok/s 2,071,442–5,523,846 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
3,452,404 tok/s 2,071,442–5,523,846 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
2,877,003 tok/s 1,726,202–4,603,205 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
2,815,604 tok/s 1,689,362–4,504,966 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
2,752,855 tok/s 1,651,713–4,404,568 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	0.7 GB	Q8_0	Comfortable
2,752,855 tok/s 1,651,713–4,404,568 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	0.7 GB	Q8_0	Comfortable
2,752,855 tok/s 1,651,713–4,404,568 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	0.7 GB	Q8_0	Comfortable
2,752,855 tok/s 1,651,713–4,404,568 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	0.7 GB	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

Full specification

Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.

Origin

Who built this model, where, and when it was published.

Organisation: IEEE
Organisation type: Industry
Country: Multinational
Published: 6 August 2014
Authors: Shengcai Liao, Anil K. Jain, S. Li

What it does

The problem areas the model was built for. A model can carry several of each.

Domain: Vision
Task: Face detection

Size

How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.

Parameters: 313.9K
Training data: 434,600 tokens

Availability

Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.

Weights: Open — downloadable
Model access: Open weights (unrestricted)

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Why it is tracked: SOTA improvement
Record confidence: Confident

Sources

Where this record came from and when it was last checked.

Reference: A Fast and Accurate Unconstrained Face Detector
Last updated: 28 November 2025

The extremes

The ten fastest GPUs that run NPD

Ranked by estimated tokens per second, newest card first where speeds tie. Because generation is bound by memory bandwidth, this ordering follows bandwidth rather than any gaming benchmark.

The smallest GPUs that still run NPD

The cheapest route in, by memory capacity. A tight fit runs the model but leaves nothing spare for a longer conversation.

What the numbers mean

What you need to run it

Minimum card

Tesla C1080

Memory needed

0.7 GB

Fastest

10,795,509 tok/s

NPD is small enough at 313.9K parameters that hardware is rarely the obstacle — 818 of the cards we track can run it, including cards several years old.

The least hardware that works is a Tesla C1080. Its 4 GB is enough at Q8_0 compression, giving roughly 117,455 tokens per second.

The quickest result comes from a B200 at around 10,795,509 tokens per second — its 8,000 GB/s of bandwidth is what buys that.

What this model is

NPD was published by IEEE, in Multinational, in August 2014. It comes out of industry.

It works in Vision, and is recorded as doing face detection.

Because its weights were released, nothing about running it depends on a provider staying available — it is yours once downloaded.

What decides the speed

The median result is around 303,137.9 tokens per second; 818 cards produce text faster than most people read it.

It is a dense model, so every parameter is read for every token produced. That makes speed track memory bandwidth almost exactly — a card with twice the bandwidth generates roughly twice as fast.

Its internal architecture is not on file, so memory is approximated from the parameter count and marked accordingly. Expect the real figure to differ, more so at long context.

What went into building it

The training set ran to roughly 434,600 tokens.

Its inclusion criterion is sOTA improvement.

Step by step

How to choose a GPU for NPD

The table above has already assessed every card we hold specifications for against this model. Getting to your answer takes six steps.

01

Start from the memory column

Every card here has been checked against NPD — around 0.7 GB at Q8_0. Capacity is the gate — a card either holds it or it does not.
02

Decide how long your conversations run

The conversation occupies memory too, and grows as it goes. Set the slider to the length you expect: at long context NPD can slip off a card that handles short questions easily.
03

Choose how far you will compress it

Each card runs the least-compressed copy it can hold — Q8_0 on the smallest card that fits. Setting a floor drops the cards that only manage NPD by squeezing it further than you would want.
04

Compare tokens per second, not specifications

Sort by speed to see how cards rank for NPD. It will not match a gaming ordering — generation is bound by memory bandwidth, which is why the B200 tops it at 10,795,509 tok/s.
05

Check the fit verdict before buying

Tight means NPD loads and works, with no room to raise the context later. Comfortable means you can. The difference matters more than a few tokens per second.
06

Open the card you have settled on

Every card name links to its own page, which runs the same calculation across the whole model catalogue. Worth a look before buying for NPD alone — a card is usually bought for more than one model.

Answers

NPD — common questions

How accurate are these NPD speed estimates?

These are estimates with real error bars. The fastest result here, 6,477,305–17,272,815 tok/s on the B200, could reasonably land anywhere in its published range depending on which runtime you use.

What GPU do I need to run NPD?

The smallest card in our catalogue that holds NPD is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 0.7 GB, and produces roughly 117,455 tokens per second. 818 cards in total can run it.

How fast is NPD on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 10,795,509 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 818 of the cards that can run NPD clear that.

How much VRAM does NPD need?

About 0.7 GB at Q8_0 compression, which is what the smallest card that runs it uses. Less compression needs more: the figures in the memory column above are recalculated for each card, because each one holds the least-compressed version it can.

Can I run NPD on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q8_0, using about 0.7 GB and generating roughly 2,010,664 tokens per second — a comfortable fit.

Can I run NPD on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q8_0, using about 0.7 GB and generating roughly 1,231,228 tokens per second — a comfortable fit.

Can I run NPD on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 0.7 GB and generating roughly 1,524,866 tokens per second — a comfortable fit.

Can I run NPD on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 0.7 GB and generating roughly 1,808,248 tokens per second — a comfortable fit.

Is NPD open source?

Its weights are published, so NPD can be downloaded and run on your own hardware. Note that open weights is not the same as open source in the full sense — it says nothing about the training data, the training code, or the commercial terms attached.

How many parameters does NPD have?

NPD has 313.9K parameters. "Our final detector contains 1,226 deep quadratic trees" "(depth of eight in this paper)" Parameters of the trained decision trees: 1226*(2^8)=313856. That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.

Who created NPD?

NPD was published by IEEE, based in Multinational, categorised as industry.

When was NPD released?

NPD was published in August 2014. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

What is NPD used for?

NPD works in Vision, and is recorded as handling face detection. These are the areas it was designed around; they describe intent rather than a hard boundary.

Where can I download NPD?

The weights for NPD are published, though we do not hold a repository link for it. This site calculates hardware requirements rather than hosting model files.

Can I run NPD if it does not fit in my GPU?

Partly. Layers that do not fit sit in system memory and run at a fraction of the speed, so a mostly-offloaded NPD is rarely worth using. Every figure here assumes the whole model is on the card.

Would two GPUs run NPD faster?

Capacity adds across cards; throughput does not. Since 818 of the cards we track already hold NPD on their own, a second card is rarely the answer here.

Why does the quantisation differ between cards for NPD?

Each card is shown running the least-compressed copy it can hold, and NPD appears at 1 different compression levels across the cards that fit it. Bigger cards get the more accurate version.

Source

Original publication

Record last updated 28 November 2025

The other direction

Looking at it from the other side?

This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.

All AI models