Calculate the TPS of the A800 SXM4 80 GB on local AI models

NVIDIA 80 GB HBM2e 2,040 GB/s August 2022

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

642 models it can run

721 models in our catalogue altogether

Largest model it holds

dots.llm1

142B · Q3_K_M · 16.4 tok/s

Fastest model

Gemma 3 QAT 1B

864 tok/s · 1B

Which AI models can run on a A800 SXM4 80 GB?

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.

642 models match

Calculating
Quantisation Fit
864 tok/s

734–1,037

Gemma 3 1B 1B Mar 2025 1.8 GB 33k tokens Q8_0 Comfortable
864 tok/s

734–1,037

Gemma 3 QAT 1B 1B Apr 2025 1.8 GB 33k tokens Q8_0 Comfortable
864 tok/s

518–1,382 · low confidence

HGRN 1B (WT 103) 1B Nov 2023 1.8 GB 131k tokens ? Q8_0 Comfortable
864 tok/s

518–1,382 · low confidence

LLama 3..2 Typhoon 2 1B 1B Dec 2024 1.8 GB 131k tokens ? Q8_0 Comfortable
864 tok/s

518–1,382 · low confidence

OLMo-1B 1B Feb 2024 1.8 GB 131k tokens ? Q8_0 Comfortable
864 tok/s

518–1,382 · low confidence

Pythia-1b 1B Apr 2023 1.8 GB 131k tokens ? Q8_0 Comfortable
800 tok/s

480–1,280 · low confidence

OpenELM-1.1B 1.1B May 2024 1.9 GB 131k tokens ? Q8_0 Comfortable
785 tok/s

471–1,257 · low confidence

DeciCoder-1B 1.1B Aug 2023 1.9 GB 131k tokens ? Q8_0 Comfortable
785 tok/s

471–1,257 · low confidence

SantaCoder 1.1B Jan 2023 1.9 GB 131k tokens ? Q8_0 Comfortable
785 tok/s

471–1,257 · low confidence

TinyLlama-1.1B (1T token checkpoint) 1.1B Oct 2023 1.9 GB 131k tokens ? Q8_0 Comfortable
785 tok/s

471–1,257 · low confidence

TinyLlama-1.1B (3T token checkpoint) 1.1B Oct 2023 1.9 GB 131k tokens ? Q8_0 Comfortable
720 tok/s

432–1,152 · low confidence

EXAONE 4.0 (1.2B) 1.2B Jul 2025 2.0 GB 131k tokens ? Q8_0 Comfortable
720 tok/s

432–1,152 · low confidence

LFM2-1.2B 1.2B Jul 2025 2.0 GB 131k tokens ? Q8_0 Comfortable
720 tok/s

432–1,152 · low confidence

MinerU2.5 1.2B Sep 2025 2.0 GB 131k tokens ? Q8_0 Comfortable
720 tok/s

432–1,152 · low confidence

Pleias 1.0 1.2B 1.2B Dec 2024 2.0 GB 131k tokens ? Q8_0 Comfortable
720 tok/s

432–1,152 · low confidence

Pleias-RAG-1B 1.2B Apr 2025 2.0 GB 131k tokens ? Q8_0 Comfortable
702 tok/s

597–843

Llama 3.2 1B 1.2B Sep 2024 2.2 GB 131k tokens Q8_0 Comfortable
693 tok/s

416–1,108 · low confidence

MiniCPM-1.2B 1.2B Jun 2024 2.0 GB 131k tokens ? Q8_0 Comfortable
665 tok/s

399–1,063 · low confidence

DeepSeek Coder 1.3B 1.3B Jan 2024 2.1 GB 131k tokens ? Q8_0 Comfortable
665 tok/s

399–1,063 · low confidence

DeepSeek-VL-1.3B 1.3B Mar 2024 2.1 GB 131k tokens ? Q8_0 Comfortable
665 tok/s

399–1,063 · low confidence

DigiRL 1.3B Jun 2024 2.1 GB 131k tokens ? Q8_0 Comfortable
665 tok/s

399–1,063 · low confidence

GLA Transformer 1.3B 1.3B Aug 2024 2.1 GB 131k tokens ? Q8_0 Comfortable
665 tok/s

399–1,063 · low confidence

Janus 1.3B 1.3B Oct 2024 2.1 GB 131k tokens ? Q8_0 Comfortable
665 tok/s

399–1,063 · low confidence

Kosmos-2.5 1.3B Aug 2024 2.1 GB 131k tokens ? Q8_0 Comfortable
665 tok/s

399–1,063 · low confidence

Otter 1.3B May 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

A800 SXM4 80 GB 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
80 GB
Memory bandwidth
2,040 GB/s
Memory type
HBM2e
Memory bus width
5,120 bit
Memory clock
1.59 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
GA100
Architecture
Ampere
Generation
Server Ampere(Axx)
Foundry
TSMC
Process size
7 nm
Transistors
54.2 billion
Transistor density
65,600 K/mm²
Die size
826 mm²
Package
BGA-2743
Released
11 August 2022

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
1.16 GHz
Boost clock
1.41 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
6,912
Texture mapping units
432
Render output units
160
Streaming multiprocessors
108
Tensor cores
432
L1 cache
192 KB
L2 cache
40 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)
78 TFLOPS
Single precision (FP32)
19.5 TFLOPS
Double precision (FP64)
9.7 TFLOPS
Pixel rate
226 GPixel/s
Texture rate
609 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)
400 W
Suggested power supply
800 W
Power connectors
None
Bus interface
PCIe 4.0 x16
Slot width
SXM Module

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
8.0
OpenCL
3.0

Listings

Where to buy a A800 SXM4 80 GB

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

Capacity and bandwidth

Memory

80 GB

Bandwidth

2,040 GB/s

Largest model

dots.llm1

A800 SXM4 80 GB holds 80 GB of HBM2e. That puts it in the class of hardware that holds the largest open-weight models without splitting them across machines. An inference runtime can reach roughly 72 GB.

Memory bandwidth reaches 2,040 GB/s across a bus of 5,120 bits. That is at the top of what exists. Since each token means reading the whole model out of memory once, it translates almost directly into generation speed — this card is bandwidth-rich enough that model size stops being the limiting factor long before the bus does.

The figure is the bus width multiplied by a memory clock of 1.59 GHz. Widening the bus and raising the clock are the two levers a manufacturer has, which is why a card with unremarkable cores can still generate quickly.

In practice that combination tops out at dots.llm1, 142B, compressed to Q3_K_M and generating around 16.4 tokens per second.

The chip and how it was built

A800 SXM4 80 GB is built on the graphics processor GA100, using the architecture Ampere from NVIDIA, as part of the generation Server Ampere(Axx).

The chip is manufactured by TSMC, on a process of 7 nm, with a die measuring 826 mm², holding 54.2 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 August 2022, roughly 4.0910071351501 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

78 TFLOPS

FP64

9.7 TFLOPS

Tensor cores

432

On paper A800 SXM4 80 GB reaches 78 TFLOPS at half precision, and 19.5 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 reaches 9.7 TFLOPS. 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.

The card carries 432 tensor cores across 108 streaming multiprocessors. These accelerate the matrix arithmetic at the heart of a transformer, and they are what make prompt processing — reading a long document before answering — dramatically faster than it would otherwise be.

Clocks run from a base of 1.16 GHz to a boost of 1.41 GHz. 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

A800 SXM4 80 GB has an L1 cache of 192 KB, backed by an L2 cache of 40 MB. 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 6,912 shading units, 432 texture mapping units, and 160 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

400 W

A800 SXM4 80 GB is rated at 400 W, and the suggested system power supply is 800 W. 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 sxm module. 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 4.0 x16. 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 A800 SXM4 80 GB

The biggest open-weight models that fit on this card, newest first. Each is shown at the best compression the card can hold.

  1. 01 Mistral Medium 3.5 128B · IQ4_XS · Apr 2026 16.6 tok/s
  2. 02 dots.llm1 142B · Q3_K_M · Jul 2025 16.4 tok/s
  3. 03 Pixtral Large 124B · IQ4_XS · Nov 2024 17.1 tok/s
  4. 04 xLAM-8x22B 141B · Q3_K_M · Sep 2024 16.5 tok/s
  5. 05 SaulLM-large 141B · Q3_K_M · Jul 2024 16.5 tok/s
  6. 06 Mixtral 8x22B 141B · Q3_K_M · Apr 2024 59.8 tok/s
  7. 07 WizardLM-2 8x22B 141B · Q3_K_M · Apr 2024 16.5 tok/s
  8. 08 Zephyr 141B-A39B 141B · Q3_K_M · Apr 2024 59.8 tok/s
  9. 09 APUS-xDAN-4.0(MoE) 136B · Q3_K_M · Apr 2024 17.1 tok/s
  10. 10 DBRX 132B · IQ4_XS · Mar 2024 58.9 tok/s

The fastest AI models on a A800 SXM4 80 GB

Where this card produces tokens quickest. Smaller models dominate here, because generating each token means reading the whole model out of memory once.

  1. 01 Gemma 3 QAT 1B 1B · Q8_0 · 1.8 GB 864 tok/s
  2. 02 Gemma 3 1B 1B · Q8_0 · 1.8 GB 864 tok/s
  3. 03 LLama 3..2 Typhoon 2 1B 1B · Q8_0 · 1.8 GB 864 tok/s
  4. 04 OLMo-1B 1B · Q8_0 · 1.8 GB 864 tok/s
  5. 05 HGRN 1B (WT 103) 1B · Q8_0 · 1.8 GB 864 tok/s
  6. 06 Pythia-1b 1B · Q8_0 · 1.8 GB 864 tok/s
  7. 07 OpenELM-1.1B 1.1B · Q8_0 · 1.9 GB 800 tok/s
  8. 08 TinyLlama-1.1B (1T token checkpoint) 1.1B · Q8_0 · 1.9 GB 785 tok/s
  9. 09 TinyLlama-1.1B (3T token checkpoint) 1.1B · Q8_0 · 1.9 GB 785 tok/s
  10. 10 DeciCoder-1B 1.1B · Q8_0 · 1.9 GB 785 tok/s

Step by step

How to work out the tokens per second of a A800 SXM4 80 GB

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.

  1. 01

    Start with the model, not the specification

    The table lists 642 models this card runs. Search narrows the list by name or by size.

  2. 02

    Match the context to your work

    Drag the slider to the conversation length you plan to work at. The cache grows with the conversation, and against a card holding 80 GB it is often what pushes a large model over the edge.

  3. 03

    Choose how far you will compress

    Each model is shown at the best compression this card can hold. A minimum quality hides the ones that only fit by being squeezed further than you would accept.

  4. 04

    Take the range as the answer

    Each speed is an estimate for a single conversation, with a range beneath it. The top end here is 864 tok/s on Gemma 3 QAT 1B. The same card and model vary by thirty to fifty per cent between inference runtimes.

  5. 05

    Check the memory column before committing

    Tight means it works today; comfortable means it still works when the conversation grows. Compare what each model needs against an available 80 GB.

  6. 06

    Cross-check against other hardware

    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 right buy is A800 SXM4 80 GB.

Answers

A800 SXM4 80 GB — common questions

01

A800 SXM4 80 GB— what is the largest AI model it can run?

The largest model in our catalogue that fits is dots.llm1 at 142B parameters, compressed to Q3_K_M. It generates roughly 16.4 tokens per second and needs about 70.1 GB of the card's memory.

02

A800 SXM4 80 GB— how many tokens per second does it produce?

It depends on the model. The fastest model we track here is Gemma 3 QAT 1B at about 864 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.

03

A800 SXM4 80 GB— can it run 7B models?

Yes. For example it runs Gemma 4 E4B at Q8_0, using about 9.8 GB of memory and generating around 192 tokens per second.

04

A800 SXM4 80 GB— can it run 13B models?

Yes. For example it runs DeepSeekMoE-16B at Q8_0, using about 17.5 GB of memory and generating around 300 tokens per second.

05

A800 SXM4 80 GB— can it run 30B models?

Yes. For example it runs ERNIE-4.5-VL-28B-A3B at Q8_0, using about 29.2 GB of memory and generating around 171 tokens per second.

06

A800 SXM4 80 GB— can it run 70B models?

Yes. For example it runs Qwen3-Coder-Next at Q6_K, using about 62.0 GB of memory and generating around 87.2 tokens per second.

07

A800 SXM4 80 GB— how much memory does it have?

This card has 80 GB of HBM2e. Around a tenth is reserved by the inference runtime and the driver, leaving roughly 72 GB available for a model and its conversation.

08

A800 SXM4 80 GB— what is its memory bandwidth?

Memory bandwidth reaches 2,040 GB/s across a bus of 5,120 bits. 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.

09

A800 SXM4 80 GB— what type of memory does it use?

It uses HBM2e clocked at 1.59 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.

10

A800 SXM4 80 GB— who makes it?

This is a product of NVIDIA, with the chip manufactured by TSMC, on a process of 7 nm.

11

A800 SXM4 80 GB— when was it released?

It was released in August 2022.

12

A800 SXM4 80 GB— how much power does it use?

Rated board power is 400 W, and the suggested system power supply is 800 W. Generating text draws hard in bursts and idles between requests, so average consumption over a working session is normally well below the rated figure.

13

A800 SXM4 80 GB— how much cache does it have?

The L1 cache is 192 KB, and the L2 cache is 40 MB. 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.

14

A800 SXM4 80 GB— what are its TFLOPS?

It is rated at 78 TFLOPS at half precision and 19.5 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.

15

A800 SXM4 80 GB— how many tensor cores does it have?

It has 432 tensor cores across 108 streaming multiprocessors. They accelerate the matrix arithmetic a transformer is built from, which mainly speeds up processing a long prompt rather than producing the reply.

16

A800 SXM4 80 GB— does it support CUDA?

Yes. It reports CUDA compute capability 8.0. 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.

17

A800 SXM4 80 GB— what bus interface does it use?

It uses PCIe 4.0 x16. 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.

18

A800 SXM4 80 GB— is it good for running local AI models?

Its memory is large enough for models most desktop hardware cannot touch and its bandwidth puts it among the fastest hardware available for generation. In total it runs 642 of the models we track. Whether that is enough depends entirely on which model you want — the table above answers that directly.

19

A800 SXM4 80 GB— can it run a model that does not fit in its memory?

It can be split, with the overflow held in system memory beyond the card's 80 GB is possible and usually a false economy: the system-memory portion is slow enough to dominate the result.

20

Would two A800 SXM4 80 GB cards be twice as fast?

No. A second card doubles the memory to 160 GB to work with rather than twice the tokens per second — every figure here is for a single card.

21

A800 SXM4 80 GB— which AI models can it run?

642 of the 721 open-weight language models we track fit on this card and can be run locally. The table on this page lists every one, with the memory it needs, the quantisation it runs at and an estimated generation speed.

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.

All GPUs