GNN TPS calculator

Open weights University of Siena 0K parameters December 2008

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 · 1,228,800,000 tok/s

Fastest card

B200

112,941,176,471 tok/s · 180 GB

Which GPUs can run GNN?

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
112,941,176,471 tok/s 67,764,705,882–180,705,882,353 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	0.7 GB	Q8_0	Comfortable
112,941,176,471 tok/s 67,764,705,882–180,705,882,353 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	0.7 GB	Q8_0	Comfortable
90,186,352,941 tok/s 54,111,811,765–144,298,164,706 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
90,186,352,941 tok/s 54,111,811,765–144,298,164,706 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
72,127,058,824 tok/s 43,276,235,294–115,403,294,118 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
69,035,294,118 tok/s 41,421,176,471–110,456,470,588 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
69,035,294,118 tok/s 41,421,176,471–110,456,470,588 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
66,070,588,235 tok/s 39,642,352,941–105,712,941,176 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	0.7 GB	Q8_0	Comfortable
58,637,647,059 tok/s 35,182,588,235–93,820,235,294 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
58,637,647,059 tok/s 35,182,588,235–93,820,235,294 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
58,637,647,059 tok/s 35,182,588,235–93,820,235,294 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
55,623,529,412 tok/s 33,374,117,647–88,997,647,059 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
47,435,294,118 tok/s 28,461,176,471–75,896,470,588 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
47,435,294,118 tok/s 28,461,176,471–75,896,470,588 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	0.7 GB	Q8_0	Comfortable
47,435,294,118 tok/s 28,461,176,471–75,896,470,588 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
47,435,294,118 tok/s 28,461,176,471–75,896,470,588 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
47,435,294,118 tok/s 28,461,176,471–75,896,470,588 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
36,118,588,235 tok/s 21,671,152,941–57,789,741,176 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
36,118,588,235 tok/s 21,671,152,941–57,789,741,176 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
30,098,823,529 tok/s 18,059,294,118–48,158,117,647 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
29,456,470,588 tok/s 17,673,882,353–47,130,352,941 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
28,800,000,000 tok/s 17,280,000,000–46,080,000,000 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	0.7 GB	Q8_0	Comfortable
28,800,000,000 tok/s 17,280,000,000–46,080,000,000 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	0.7 GB	Q8_0	Comfortable
28,800,000,000 tok/s 17,280,000,000–46,080,000,000 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	0.7 GB	Q8_0	Comfortable
28,800,000,000 tok/s 17,280,000,000–46,080,000,000 · 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: University of Siena
Organisation type: Academia
Country: Italy
Published: 9 December 2008
Authors: Franco Scarselli; Marco Gori; Ah Chung Tsoi; Markus Hagenbuchner; Gabriele Monfardini

What it does

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

Domain: Other
Task: Binary classification
Approach: Supervised

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: 0K
Training data: 207 tokens
Epochs: 5,000

Training compute

The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.

Training compute: 1.6 × 10⁹ FLOP
How it was established: Operation counting

The training run

What it physically took to train: which chips, how many, for how long, and what that drew from the wall.

Chips used: 1

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: Historical significance,Highly cited
Record confidence: Confident

Sources

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

Reference: The Graph Neural Network Model
Last updated: 28 November 2025

The extremes

The ten fastest GPUs that run GNN

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 GNN

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

112,941,176,471 tok/s

GNN is small enough at 0K 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 1,228,800,000 tokens per second.

At the other end, a B200 generates roughly 112,941,176,471 tokens per second on it, on the strength of 8,000 GB/s of memory bandwidth.

About this model

GNN was published by University of Siena, in Italy, in December 2008. The organisation is categorised as academia.

It works in Other, and is recorded as doing binary classification.

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

How fast it runs, and why

Half the cards that hold it manage more than 3,171,388,235.3 tokens per second, and 818 exceed reading speed outright.

Being dense, it reads all of itself per token, which is why the ordering by speed below follows the ordering by memory bandwidth so closely.

Memory here is estimated from size rather than computed from the architecture, which is not recorded for this model — the numbers are indicative rather than exact.

What went into building it

Training it took roughly 1.6 × 10⁹ FLOP of computation — a measure of what producing the model cost, not of how fast it answers.

It was trained on about 207 tokens of text.

It is tracked in the underlying dataset for one reason in particular: historical significance,Highly cited.

Step by step

How to choose a GPU for GNN

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

01

Read the memory figure first

Every card here has been checked against GNN — 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

Longer conversations cost memory on top of what the weights need. Move the slider to your real working length before trusting any row for GNN.
03

Choose how far you will compress it

The quantisation column varies by card, because a bigger card holds a more accurate copy of GNN — Q8_0 on the smallest card that fits. Set a floor to hold the comparison at one level.
04

Sort by speed

The speed ordering for GNN is effectively an ordering by memory bandwidth, which is why the B200 tops it at 112,941,176,471 tok/s.
05

Check the fit verdict before buying

The fit column separates cards that just manage GNN from those with room to spare. Buy for the second if the context might grow.
06

Check the card from the other side

Following a card through to its own page shows every other model it can hold, which is the question that follows once GNN is settled.

Answers

GNN — common questions

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

It can be split between the card and system memory, but GNN generates painfully slowly that way. Nothing on this page assumes offloading.

Would two GPUs run GNN faster?

Two cards buy memory rather than speed. That matters for GNN only if one card cannot hold it — 818 can, so a second adds little.

Why does the quantisation differ between cards for GNN?

A larger card holds a more accurate copy. Across the cards that run GNN, 1 compression levels are used; the floor control above pins it to one.

How accurate are these GNN speed estimates?

These are estimates with real error bars. The fastest result here, 67,764,705,882–180,705,882,353 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 GNN?

The smallest card in our catalogue that holds GNN is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 0.7 GB, and produces roughly 1,228,800,000 tokens per second. 818 cards in total can run it.

How fast is GNN on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 112,941,176,471 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 GNN clear that.

How much VRAM does GNN 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 GNN 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 21,035,294,118 tokens per second — a comfortable fit.

Can I run GNN 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 12,880,941,176 tokens per second — a comfortable fit.

Can I run GNN 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 15,952,941,176 tokens per second — a comfortable fit.

Can I run GNN 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 18,917,647,059 tokens per second — a comfortable fit.

Is GNN open source?

Its weights are published, so GNN 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 GNN have?

GNN has 0K parameters. 5*5+5=30 (3 layer network with 5 hidden neurons). 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 GNN?

GNN was published by University of Siena, based in Italy, categorised as academia.

When was GNN released?

GNN was published in December 2008. 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 GNN used for?

GNN works in Other, and is recorded as handling binary classification. A model can carry several of each, so these are the areas it was built for rather than a limit on what it will attempt.

Where can I download GNN?

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

How much compute was used to train GNN?

Around 1.6 × 10⁹ FLOP. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.

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