HR-ResNet101 TPS calculator

Open weights Carnegie Mellon University (CMU) 44.5M parameters December 2016

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 · 828 tok/s

Fastest card

B200

76,140 tok/s · 180 GB

Which GPUs can run HR-ResNet101?

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
76,140 tok/s 45,684–121,824 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	0.7 GB	Q8_0	Comfortable
76,140 tok/s 45,684–121,824 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	0.7 GB	Q8_0	Comfortable
60,800 tok/s 36,480–97,280 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
60,800 tok/s 36,480–97,280 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	0.7 GB	Q8_0	Comfortable
48,625 tok/s 29,175–77,800 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
46,541 tok/s 27,924–74,465 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
46,541 tok/s 27,924–74,465 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.7 GB	Q8_0	Comfortable
44,542 tok/s 26,725–71,267 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	0.7 GB	Q8_0	Comfortable
39,531 tok/s 23,719–63,250 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
39,531 tok/s 23,719–63,250 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
39,531 tok/s 23,719–63,250 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	0.7 GB	Q8_0	Comfortable
37,499 tok/s 22,499–59,998 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
31,979 tok/s 19,187–51,166 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
31,979 tok/s 19,187–51,166 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	0.7 GB	Q8_0	Comfortable
31,979 tok/s 19,187–51,166 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
31,979 tok/s 19,187–51,166 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
31,979 tok/s 19,187–51,166 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	0.7 GB	Q8_0	Comfortable
24,350 tok/s 14,610–38,959 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
24,350 tok/s 14,610–38,959 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	0.7 GB	Q8_0	Comfortable
20,291 tok/s 12,175–32,466 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
19,858 tok/s 11,915–31,773 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	0.7 GB	Q8_0	Comfortable
19,416 tok/s 11,649–31,065 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	0.7 GB	Q8_0	Comfortable
19,416 tok/s 11,649–31,065 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	0.7 GB	Q8_0	Comfortable
19,416 tok/s 11,649–31,065 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	0.7 GB	Q8_0	Comfortable
19,416 tok/s 11,649–31,065 · 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: Carnegie Mellon University (CMU)
Organisation type: Academia
Country: United States of America
Published: 13 December 2016
Authors: Peiyun Hu, Deva Ramanan

What it does

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

Domain: Vision
Task: Face detection
Approach: Supervised
Base model: ResNet-101 (ImageNet)

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: 44.5M
Training data: 8,243,968 tokens
Epochs: 170

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: 7.1 × 10¹⁸ FLOP
How it was established: Operation counting
Fine-tuning compute: 7.3 × 10¹⁶ FLOP

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: Finding Tiny Faces
Last updated: 28 November 2025

The extremes

The ten fastest GPUs that run HR-ResNet101

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 HR-ResNet101

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

Hardware requirements in practice

Minimum card

Tesla C1080

Memory needed

0.7 GB

Fastest

76,140 tok/s

HR-ResNet101 is small enough at 44.5M parameters that hardware is rarely the obstacle — 818 of the cards we track can run it, including cards several years old.

The entry point is the Tesla C1080: 4 GB of memory, Q8_0 compression, roughly 828 tokens per second.

A B200 is the fastest we calculate for it: about 76,140 tokens per second, from 8,000 GB/s of memory bandwidth.

What this model is

HR-ResNet101 was published by Carnegie Mellon University (CMU), in United States of America, in December 2016. It comes out of academia.

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

Its starting point was ResNet-101 (ImageNet) — most models at this scale are adapted from an existing base rather than built from nothing.

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.

What decides the speed

The median result is around 2,138.0 tokens per second; 818 cards produce text faster than most people read it.

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.

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

Training it took roughly 7.1 × 10¹⁸ FLOP of computation — a measure of what producing the model cost, not of how fast it answers.

The training set ran to roughly 8,243,968 tokens.

Its inclusion criterion is sOTA improvement.

Step by step

How to choose a GPU for HR-ResNet101

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

The table lists every card that can hold HR-ResNet101 — around 0.7 GB at Q8_0. That figure, not the card's headline performance, is what decides whether it runs.
02

Set the context length you will work at

Set the context to what you will actually use. The cache grows with the conversation, and it is the usual reason HR-ResNet101 stops fitting a card that seemed fine.
03

Set a quality floor

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 HR-ResNet101 by squeezing it further than you would want.
04

Compare tokens per second, not specifications

Ranking by tokens per second for HR-ResNet101 follows memory bandwidth, not core counts, which is why the B200 tops it at 76,140 tok/s.
05

Look at the headroom, not just the fit

Tight means HR-ResNet101 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

Check the card from the other side

Every card name links to its own page, which runs the same calculation across the whole model catalogue. Worth a look before buying for HR-ResNet101 alone — a card is usually bought for more than one model.

Answers

HR-ResNet101 — common questions

Would two GPUs run HR-ResNet101 faster?

A second card roughly doubles the memory available but not the generation rate. With 818 cards already able to run HR-ResNet101 alone, the case for pairing is weak.

Why does the quantisation differ between cards for HR-ResNet101?

Because capacity varies, so does how hard HR-ResNet101 has to be squeezed — 1 distinct levels appear in the table above. Set a minimum quality to compare at one.

How accurate are these HR-ResNet101 speed estimates?

These are estimates with real error bars. The fastest result here, 45,684–121,824 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 HR-ResNet101?

The smallest card in our catalogue that holds HR-ResNet101 is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 0.7 GB, and produces roughly 828 tokens per second. 818 cards in total can run it.

How fast is HR-ResNet101 on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 76,140 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 HR-ResNet101 clear that.

How much VRAM does HR-ResNet101 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 HR-ResNet101 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 14,181 tokens per second — a comfortable fit.

Can I run HR-ResNet101 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 8,684 tokens per second — a comfortable fit.

Can I run HR-ResNet101 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 10,755 tokens per second — a comfortable fit.

Can I run HR-ResNet101 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 12,753 tokens per second — a comfortable fit.

Is HR-ResNet101 open source?

Its weights are published, so HR-ResNet101 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 HR-ResNet101 have?

HR-ResNet101 has 44.5M parameters. ResNet 101. 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 HR-ResNet101?

HR-ResNet101 was published by Carnegie Mellon University (CMU), based in United States of America, categorised as academia.

When was HR-ResNet101 released?

HR-ResNet101 was published in December 2016. 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 HR-ResNet101 used for?

HR-ResNet101 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 HR-ResNet101?

The weights for HR-ResNet101 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 HR-ResNet101?

Around 7.1 × 10¹⁸ 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.

Can I run HR-ResNet101 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 HR-ResNet101 is rarely worth using. Every figure here assumes the whole model is on the card.

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