Calculate the TPS of the Xeon Phi 5120D on local AI models

Intel 8 GB GDDR5 320 GB/s June 2013

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

337 models it can run

679 models in our catalogue altogether

Largest model it holds

Baichuan 1-13B

13.3B · Q3_K_M · 17.9 tok/s

Fastest model

Gemma 3 QAT 1B

88.1 tok/s · 1B

Which AI models can run on a Xeon Phi 5120D?

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.

337 models match

Calculating
Quantisation Fit
88.1 tok/s

53–141 · low confidence

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

53–141 · low confidence

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

53–141 · low confidence

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

53–141 · low confidence

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

53–141 · low confidence

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

53–141 · low confidence

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

49–131 · low confidence

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

48–128 · low confidence

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

48–128 · low confidence

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

48–128 · low confidence

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

48–128 · low confidence

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

44–117 · low confidence

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

44–117 · low confidence

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

44–117 · low confidence

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

44–117 · low confidence

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

43–115 · low confidence

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

42–113 · low confidence

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

41–108 · low confidence

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

41–108 · low confidence

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

41–108 · low confidence

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

41–108 · low confidence

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

41–108 · low confidence

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

41–108 · low confidence

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

41–108 · low confidence

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

41–108 · 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

Xeon Phi 5120D 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
8 GB
Memory bandwidth
320 GB/s
Memory type
GDDR5
Memory bus width
512 bit
Memory clock
1.25 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
Knights Corner
Architecture
Knights
Generation
Knights Corner(x100)
Foundry
Intel
Process size
22 nm
Transistors
5 billion
Transistor density
6,900 K/mm²
Die size
720 mm²
Released
17 June 2013

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
1.05 GHz
Boost clock
1.05 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
960
Texture mapping units
32

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.

Single precision (FP32)
2 TFLOPS
Texture rate
34 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)
245 W
Suggested power supply
550 W
Bus interface
PCIe 2.0 x16
Slot width
Dual-slot
Dimensions
248 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.

OpenCL
1.2
Shader model
5.0

Listings

Where to buy a Xeon Phi 5120D

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

8 GB

Bandwidth

320 GB/s

Largest model

Baichuan 1-13B

Xeon Phi 5120D carries only 8 GB of GDDR5. That limits it to the smaller end of the catalogue, and a model has to fit entirely inside before it generates anything at all. A runtime actually gets about 7.2 GB.

Memory bandwidth reaches 320 GB/s across a bus of 512 bits. That is the number governing generation speed — arithmetic per byte read is small enough that the bus, not the cores, is what everything waits on.

Bandwidth is clock times bus width, and this card clocks its memory at 1.25 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.

The practical ceiling is Baichuan 1-13B, 13.3B, compressed to Q3_K_M and generating around 17.9 tokens per second.

The chip and how it was built

Xeon Phi 5120D is built on the graphics processor Knights Corner, using the architecture Knights from Intel, as part of the generation Knights Corner(x100).

The chip is manufactured by Intel, on a process of 22 nm, with a die measuring 720 mm², holding 5 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 June 2013, roughly 13.122911060137 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

Clocks run from a base of 1.05 GHz to a boost of 1.05 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

There are 960 shading units, 32 texture mapping 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

245 W

Xeon Phi 5120D is rated at 245 W, and the suggested system power supply is 550 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 dual-slot, measuring 248 mm long. 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 2.0 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 Xeon Phi 5120D

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 OLMo 2 Furious 13B 13B · Q3_K_M · Dec 2024 18.3 tok/s
  2. 02 Cambrian-1-13B 13B · Q3_K_M · Jun 2024 18.3 tok/s
  3. 03 Fugaku-LLM 13B · Q3_K_M · May 2024 18.3 tok/s
  4. 04 OpenThaiGPT v1.0.0 (13B) 13.1B · Q3_K_M · Apr 2024 18.2 tok/s
  5. 05 Aya 13B · Q3_K_M · Feb 2024 18.3 tok/s
  6. 06 Elyza 13B · Q3_K_M · Dec 2023 18.3 tok/s
  7. 07 NexusRaven-V2 13B · Q3_K_M · Dec 2023 18.3 tok/s
  8. 08 Baize-v2-13B (白泽) 13B · Q3_K_M · Dec 2023 18.3 tok/s
  9. 09 Stockmark-13B 13.2B · Q3_K_M · Oct 2023 18.0 tok/s
  10. 10 Baichuan 1-13B 13.3B · Q3_K_M · Jul 2023 17.9 tok/s

The fastest AI models on a Xeon Phi 5120D

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 88.1 tok/s
  2. 02 Gemma 3 1B 1B · Q8_0 · 1.8 GB 88.1 tok/s
  3. 03 LLama 3..2 Typhoon 2 1B 1B · Q8_0 · 1.8 GB 88.1 tok/s
  4. 04 OLMo-1B 1B · Q8_0 · 1.8 GB 88.1 tok/s
  5. 05 HGRN 1B (WT 103) 1B · Q8_0 · 1.8 GB 88.1 tok/s
  6. 06 Pythia-1b 1B · Q8_0 · 1.8 GB 88.1 tok/s
  7. 07 OpenELM-1.1B 1.1B · Q8_0 · 1.9 GB 81.6 tok/s
  8. 08 TinyLlama-1.1B (1T token checkpoint) 1.1B · Q8_0 · 1.9 GB 80.1 tok/s
  9. 09 TinyLlama-1.1B (3T token checkpoint) 1.1B · Q8_0 · 1.9 GB 80.1 tok/s
  10. 10 DeciCoder-1B 1.1B · Q8_0 · 1.9 GB 80.1 tok/s

Step by step

How to work out the tokens per second of a Xeon Phi 5120D

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 337 models the card handles. The search box takes a name or a size such as 27b, which matches on parameter count.

  2. 02

    Set the context length you will actually use

    Set the context to your real working length. Short questions cost almost nothing, but a long document can consume a large share of 8 GB so the setting is worth getting right.

  3. 03

    Pin the comparison to one quality level

    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

    Look at the range, not just the number

    Each speed is an estimate for a single conversation, with a range beneath it. The top end here is 88.1 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

    The fit column separates models that just fit from those with room to spare — worth checking before settling on one, against an available 8 GB.

  6. 06

    Check the same model from the other side

    Following a model through to its own page lists all the hardware that can run it, so you can see how it compares against Xeon Phi 5120D.

Answers

Xeon Phi 5120D — common questions

01

Xeon Phi 5120D— can it run 7B models?

Yes. For example it runs MetaMath 7B (LLaMa finetune) at Q4_K_M, using about 6.5 GB of memory and generating around 29.1 tokens per second.

02

Xeon Phi 5120D— can it run 13B models?

Yes. For example it runs Gemma 4 12B at Q3_K_M, using about 6.5 GB of memory and generating around 19.9 tokens per second.

03

Xeon Phi 5120D— how much memory does it have?

This card has 8 GB of GDDR5. Around a tenth is reserved by the inference runtime and the driver, leaving roughly 7.2 GB available for a model and its conversation.

04

Xeon Phi 5120D— what is its memory bandwidth?

Memory bandwidth reaches 320 GB/s across a bus of 512 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.

05

Xeon Phi 5120D— what type of memory does it use?

It uses GDDR5 clocked at 1.25 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.

06

Xeon Phi 5120D— who makes it?

This is a product of Intel, with the chip manufactured by Intel, on a process of 22 nm.

07

Xeon Phi 5120D— when was it released?

It was released in June 2013.

08

Xeon Phi 5120D— how much power does it use?

Rated board power is 245 W, and the suggested system power supply is 550 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.

09

Xeon Phi 5120D— does it support CUDA?

No. CUDA is NVIDIA-only, and this is a card from Intel. It runs language models through ROCm, Vulkan or Metal depending on the software, which are less mature than the CUDA path — our estimates apply a penalty for that.

10

Xeon Phi 5120D— what bus interface does it use?

It uses PCIe 2.0 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.

11

Xeon Phi 5120D— is it good for running local AI models?

Its memory limits it to smaller models though its bandwidth means generation will feel slow on larger models. In total it runs 337 of the models we track. Whether that is enough depends entirely on which model you want — the table above answers that directly.

12

Xeon Phi 5120D— can it run a model that does not fit in its memory?

Only partly. Layers beyond the card's 8 GB is possible and usually a false economy: the system-memory portion is slow enough to dominate the result.

13

Would two Xeon Phi 5120D cards be twice as fast?

Capacity adds, throughput does not. Two of them give you 16 GB which lets you hold models neither could hold alone, but generation does not split that way. These figures describe one card.

14

Xeon Phi 5120D— which AI models can it run?

337 of the 679 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.

15

Xeon Phi 5120D— what is the largest AI model it can run?

The largest model in our catalogue that fits is Baichuan 1-13B at 13.3B parameters, compressed to Q3_K_M. It generates roughly 17.9 tokens per second and needs about 7.2 GB of the card's memory.

16

Xeon Phi 5120D— 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 88.1 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.

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