DMPFold TPS calculator

Open weights University College London (UCL) 3.8M parameters November 2018

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 · 9,701 tok/s

Fastest card

B200

891,641 tok/s · 180 GB

Which GPUs can run DMPFold?

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
891,641 tok/s 534,985–1,426,625 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	0.7 GB	Q8_0	Comfortable
891,641 tok/s 534,985–1,426,625 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	0.7 GB	Q8_0	Comfortable
711,998 tok/s 427,199–1,139,196 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
711,998 tok/s 427,199–1,139,196 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
569,424 tok/s 341,654–911,079 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
545,015 tok/s 327,009–872,025 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
545,015 tok/s 327,009–872,025 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
521,610 tok/s 312,966–834,576 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	0.7 GB	Q8_0	Comfortable
462,929 tok/s 277,757–740,686 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
462,929 tok/s 277,757–740,686 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
462,929 tok/s 277,757–740,686 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
439,133 tok/s 263,480–702,613 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
374,489 tok/s 224,694–599,183 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
374,489 tok/s 224,694–599,183 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	0.7 GB	Q8_0	Comfortable
374,489 tok/s 224,694–599,183 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
374,489 tok/s 224,694–599,183 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
374,489 tok/s 224,694–599,183 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
285,147 tok/s 171,088–456,235 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
285,147 tok/s 171,088–456,235 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
237,622 tok/s 142,573–380,196 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
232,551 tok/s 139,531–372,082 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
227,368 tok/s 136,421–363,789 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	0.7 GB	Q8_0	Comfortable
227,368 tok/s 136,421–363,789 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	0.7 GB	Q8_0	Comfortable
227,368 tok/s 136,421–363,789 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	0.7 GB	Q8_0	Comfortable
227,368 tok/s 136,421–363,789 · 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 College London (UCL)
Organisation type: Academia
Country: United Kingdom of Great Britain and Northern Ireland
Published: 29 November 2018
Authors: Joe G. Greener, Shaun M. Kandathil and David T. Jones

What it does

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

Domain: Biology
Task: Proteins, Protein folding prediction, Protein contact and distance prediction

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: 3.8M
Training data: 1,516,215,024 tokens
Epochs: 75

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)
Training code: Open source

How it is classified

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

Record confidence: Likely
Citations: 174

Sources

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

Reference: Deep learning extends de novo protein modelling coverage of genomes using iteratively predicted structural constraints
Last updated: 28 November 2025

The extremes

The ten fastest GPUs that run DMPFold

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 DMPFold

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

891,641 tok/s

DMPFold is small enough at 3.8M 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 9,701 tokens per second.

The quickest result comes from a B200 at around 891,641 tokens per second — its 8,000 GB/s of bandwidth is what buys that.

Where it came from

DMPFold was published by University College London (UCL), in United Kingdom of Great Britain and Northern Ireland, in November 2018. The organisation is categorised as academia.

It works in Biology, and is recorded as doing proteins, Protein folding prediction, Protein contact and distance prediction.

The weights being open is what puts this page in the calculator rather than only in the catalogue: it is a model you can actually hold.

Understanding the speeds

Half the cards that hold it manage more than 25,037.3 tokens per second, and 818 exceed reading speed outright.

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.

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.

Training and provenance

It was trained on about 1,516,215,024 tokens of text.

Step by step

How to choose a GPU for DMPFold

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

01

Check what it needs before anything else

Look at what DMPFold actually needs — around 0.7 GB at Q8_0. No amount of processing power compensates for a card that cannot hold it.
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 DMPFold.
03

Choose how far you will compress it

Compression is what makes DMPFold fit smaller cards, at some cost in accuracy — Q8_0 on the smallest card that fits. A minimum quality removes the ones that go too far.
04

Rank by throughput rather than spec sheet

The speed ordering for DMPFold is effectively an ordering by memory bandwidth, which is why the B200 tops it at 891,641 tok/s.
05

Look at the headroom, not just the fit

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

Open the card you have settled on

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

Answers

DMPFold — common questions

Can I run DMPFold 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 149,350 tokens per second — a comfortable fit.

Is DMPFold open source?

Its weights are published, so DMPFold 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 DMPFold have?

DMPFold has 3.8M parameters. Based on Fig. 9: Distance predictor network: [Total = 17684] (1) Maxout2D: 0 parameters (2) ResBlock x 18: 2*64*18 = 16384 parameters (a shift and scale parameter for every InstanceNorm2D layer) (3) Conv2D 1x1: 64*20 + 20 = 1300 parameters (4) Softmax: 0 parameters Hydrogen bond predictor network: [Total = 3691073] (1) Maxout2D: 0 parameters (2) ResBlock x 18: (2*64*2 + 5*5*64*64*2)*18 = 3 691 008 parameters (3) Conv2D 1x1: 64*1 + 1 = 65 parameters (4) Sigmoid: 0 parameters Torsion angles and errors network [Total = 115459] (1) Maxout2D: 0 parameters (2) ResBlock x 18: 2*64*18 = 16384 parameters (a shift and scale parameter for every InstanceNorm2D layer) (3) BLSTM: 4×2×(N+M)×M where M is 128 (hidden units in BLSTM layer) and N is 64 (input dimensionality) = 98304 (4) Conv1D 1x1: 256*3 + 3 = 771 parameters Estimate total parameters = 3.8e6 parameters [See Section 'Additional constraint types and iterative predictions': "a bidirectional recurrent LSTM layer with 128 hidden units (BLSTM in Fig. 9c), which embeds each row of the final 2-D 64-channel feature map in a single 256-D vector (concatenation of 128-D final timestep output states of the forward and reverse direction LSTM passes)"]. 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 DMPFold?

DMPFold was published by University College London (UCL), based in United Kingdom of Great Britain and Northern Ireland, categorised as academia.

When was DMPFold released?

DMPFold was published in November 2018. 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 DMPFold used for?

DMPFold works in Biology, and is recorded as handling proteins, Protein folding prediction, Protein contact and distance prediction. 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 DMPFold?

The weights for DMPFold 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 DMPFold if it does not fit in my GPU?

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

Would two GPUs run DMPFold faster?

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

Why does the quantisation differ between cards for DMPFold?

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

How accurate are these DMPFold speed estimates?

They are calculated from specifications rather than measured, and each carries a range — 534,985–1,426,625 tok/s on the B200, for instance. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

What GPU do I need to run DMPFold?

The smallest card in our catalogue that holds DMPFold is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 0.7 GB, and produces roughly 9,701 tokens per second. 818 cards in total can run it.

How fast is DMPFold on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 891,641 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 DMPFold clear that.

How much VRAM does DMPFold 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 DMPFold 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 166,068 tokens per second — a comfortable fit.

Can I run DMPFold 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 101,692 tokens per second — a comfortable fit.

Can I run DMPFold 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 125,944 tokens per second — a comfortable fit.

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