Calculate the TPS of the A30X on local AI models

NVIDIA 24 GB HBM2e 1,220 GB/s April 2021

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

502 of 679 models it can run

Largest model it holds

Mixtral 8x7B

46.7B · Q3_K_M · 108 tok/s

Fastest model

Gemma 3 QAT 1B

517 tok/s · 1B

What AI models can a A30X run?

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.

502 models match

Calculating
Quantisation Fit
517 tok/s

439–620

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

439–620

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

310–827 · low confidence

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

310–827 · low confidence

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

310–827 · low confidence

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

310–827 · low confidence

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

287–765 · low confidence

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

282–752 · low confidence

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

282–752 · low confidence

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

282–752 · low confidence

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

282–752 · low confidence

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

258–689 · low confidence

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

258–689 · low confidence

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

258–689 · low confidence

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

258–689 · low confidence

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

357–504

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

249–663 · low confidence

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

238–636 · low confidence

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

238–636 · low confidence

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

238–636 · low confidence

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

238–636 · low confidence

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

238–636 · low confidence

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

238–636 · low confidence

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

238–636 · low confidence

Otter 1.3B May 2023 2.1 GB 131k tokens ? Q8_0 Comfortable
397 tok/s

238–636 · low confidence

Phi-1 1.3B Oct 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

A30X 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
24 GB
Memory bandwidth
1,220 GB/s
Memory type
HBM2e
Memory bus width
3,072 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
12 April 2021

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
1.04 GHz
Boost clock
1.44 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
3,584
Texture mapping units
224
Render output units
96
Streaming multiprocessors
56
Tensor cores
224
L1 cache
192 KB
L2 cache
24 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)
10.3 TFLOPS
Single precision (FP32)
10.3 TFLOPS
Double precision (FP64)
5.2 TFLOPS
Pixel rate
138 GPixel/s
Texture rate
323 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)
230 W
Suggested power supply
550 W
Power connectors
1x 16-pin
Bus interface
PCIe 4.0 x8
Slot width
Dual-slot
Dimensions
267 mm

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 A30X

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

Why memory is the number that matters here

Memory

24 GB

Bandwidth

1,220 GB/s

Largest model

Mixtral 8x7B

The A30X carries 24 GB of HBM2e, which covers the mid-sized models most people actually run — about 21.6 GB of it after the runtime and driver reserve their working space.

Bandwidth is 1,220 GB/s across a 3,072-bit bus. Generating a token means reading every weight once, so that figure sets the pace more than any other number here, and at this level text arrives faster than most people read.

That comes from a 1.59 GHz memory clock across the bus width above. 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.

Put together, the largest model that fits is Mixtral 8x7B at 46.7B, running Q3_K_M and producing around 108 tokens per second.

The chip and how it was built

The A30X is built on the GA100 graphics processor, using NVIDIA's Ampere architecture, as part of the Server Ampere(Axx) generation.

The chip is manufactured by TSMC, on a 7 nm process, 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 April 2021, roughly 5 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

10.3 TFLOPS

FP64

5.2 TFLOPS

Tensor cores

224

On paper the A30X reaches 10.3 TFLOPS at half precision and 10.3 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 is 5.2 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 224 tensor cores across 56 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 1.04 GHz at base to 1.44 GHz boosted. 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

The A30X has 192 KB of L1 cache, backed by 24 MB of L2. 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 3,584 shading units, 224 texture mapping units, and 96 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

230 W

The A30X is rated at 230 W, with a 550 W power supply suggested for the whole system. 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 a dual-slot, measuring 267 mm long, and needs 1x 16-pin. 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 x8. 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 a A30X can run

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 Qwen3-Omni-30B-A3B 35.3B · Q4_K_M · Sep 2025 188 tok/s
  2. 02 InternVL2_5-38B 38.4B · Q3_K_M · Dec 2024 36.3 tok/s
  3. 03 TeleChat2-35B 35B · IQ4_XS · Oct 2024 36.3 tok/s
  4. 04 InternVL2-40B 40.1B · Q3_K_M · Jul 2024 34.8 tok/s
  5. 05 JIUTIAN-139MoE 38.8B · Q3_K_M · Jun 2024 35.9 tok/s
  6. 06 VILA1.5-40B 40B · Q3_K_M · May 2024 34.9 tok/s
  7. 07 LLaVA-NeXT-34B (LLaVA-1.6) 34.8B · IQ4_XS · Jan 2024 36.5 tok/s
  8. 08 Mixtral 8x7B 46.7B · Q3_K_M · Dec 2023 108 tok/s
  9. 09 Falcon-40B 40B · Q3_K_M · Mar 2023 34.9 tok/s
  10. 10 gpt-sw3-40b 40B · Q3_K_M · Mar 2023 34.9 tok/s

The fastest AI models on a A30X

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 517 tok/s
  2. 02 Gemma 3 1B 1B · Q8_0 · 1.8 GB 517 tok/s
  3. 03 LLama 3..2 Typhoon 2 1B 1B · Q8_0 · 1.8 GB 517 tok/s
  4. 04 OLMo-1B 1B · Q8_0 · 1.8 GB 517 tok/s
  5. 05 HGRN 1B (WT 103) 1B · Q8_0 · 1.8 GB 517 tok/s
  6. 06 Pythia-1b 1B · Q8_0 · 1.8 GB 517 tok/s
  7. 07 OpenELM-1.1B 1.1B · Q8_0 · 1.9 GB 478 tok/s
  8. 08 TinyLlama-1.1B (1T token checkpoint) 1.1B · Q8_0 · 1.9 GB 470 tok/s
  9. 09 TinyLlama-1.1B (3T token checkpoint) 1.1B · Q8_0 · 1.9 GB 470 tok/s
  10. 10 DeciCoder-1B 1.1B · Q8_0 · 1.9 GB 470 tok/s

Step by step

How to work out the tokens per second of a A30X

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

    Find the model in the table

    The table lists 502 models this A30X can run. Search by name, or by size — typing 27b matches on the parameter count even when the name never states it.

  2. 02

    Decide how long your conversations run

    Drag the slider to the conversation length you plan to work at. The cache grows with the conversation, and on 24 GB it is often what pushes a large model over the edge.

  3. 03

    Set a minimum quality if you need one

    By default the table picks the least-compressed copy that fits. Setting a floor removes models that only qualify through heavy compression.

  4. 04

    Take the range as the answer

    Each speed is an estimate for a single conversation, with a range beneath it — 517 tok/s on Gemma 3 QAT 1B at the top end here. The same card and model vary by thirty to fifty per cent between inference runtimes.

  5. 05

    Check the memory column before committing

    The fit column separates models that just fit from those with room to spare — worth checking against the card's 24 GB before settling on one.

  6. 06

    Open the model to compare cards

    Each model page repeats this calculation for the whole catalogue. Worth a look before deciding: it shows what else runs the same model, and how the A30X compares.

Answers

A30X — common questions

01

How much power does a A30X use?

The A30X has a rated board power of 230 W, and a 550 W system power supply is suggested. Generating text draws hard in bursts and idles between requests, so average consumption over a working session is normally well below the rated figure.

02

How much cache does a A30X have?

The A30X has 192 KB of L1 cache, and 24 MB of L2 cache. 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.

03

What are the TFLOPS of a A30X?

The A30X is rated at 10.3 TFLOPS at half precision and 10.3 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.

04

How many tensor cores does a A30X have?

The A30X has 224 tensor cores across 56 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.

05

Does the A30X support CUDA?

Yes. The A30X 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.

06

What bus interface does the A30X use?

It uses PCIe 4.0 x8. 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.

07

Is the A30X good for running local AI models?

Its memory comfortably covers the mid-sized models most people run locally and its bandwidth is high enough to generate text faster than most people read. In total it runs 502 of the models we track. Whether that is enough depends entirely on which model you want — the table above answers that directly.

08

Can a A30X run a model that does not fit in its memory?

Only partly. Layers beyond the 24 GB sit in system memory and run at a fraction of the speed, so a mostly-offloaded model is rarely worth using. Every figure here assumes it is fully resident on the card.

09

Would two A30X cards be twice as fast?

Pairing A30X cards buys headroom rather than pace: 48 GB of combined memory, at roughly the same generation speed as one.

10

What AI models can a A30X run?

502 of the 679 open-weight language models we track fit on a A30X and can be run locally on it. The table on this page lists every one, with the memory it needs, the quantisation it runs at and an estimated generation speed.

11

What is the largest AI model a A30X can run?

The largest model in our catalogue that fits on a A30X is Mixtral 8x7B at 46.7B parameters, compressed to Q3_K_M. It generates roughly 108 tokens per second and needs about 21.0 GB of the card's memory.

12

How many tokens per second does a A30X produce?

It depends on the model. On a A30X the fastest model we track is Gemma 3 QAT 1B at about 517 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.

13

Can a A30X run a 7B model?

Yes. For example a A30X runs Multi-Token Prediction 7B at Q8_0, using about 7.9 GB of memory and generating around 77.1 tokens per second.

14

Can a A30X run a 13B model?

Yes. For example a A30X runs DeepSeekMoE-16B at Q8_0, using about 17.5 GB of memory and generating around 179 tokens per second.

15

Can a A30X run a 30B model?

Yes. For example a A30X runs Nemotron 3-Nano-30B-A3B at Q4_K_M, using about 18.1 GB of memory and generating around 210 tokens per second.

16

How much memory does a A30X have?

A A30X has 24 GB of HBM2e memory. Around a tenth of that is reserved by the inference runtime and the driver, leaving roughly 21.6 GB available for a model and its conversation.

17

What is the memory bandwidth of a A30X?

The A30X has 1,220 GB/s of memory bandwidth, across a 3,072-bit memory bus. 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.

18

What type of memory does a A30X 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.

19

Who makes the A30X?

The A30X is a NVIDIA product, with the chip manufactured by TSMC, on a 7 nm process.

20

When was the A30X released?

The A30X was released in April 2021.

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