Can StarCoder 15B run on AMD Instinct MI250 128GB?

YES — Runs Great

A74Great

Estimated from fit model

StarCoder 15B needs ~39.4 GB VRAM. AMD Instinct MI250 128GB has 128.0 GB. With Q5_K_M quantization, expect ~206 tok/s.

Runtime: OllamaCapacity: RoomyBandwidth: HighStack: BasicBottleneck: Balanced

Operating mode

Choose the run profile you care about

Interactive favors responsiveness, while light API and scale-out lean harder on serving readiness. The fit stays the same, but the recommendation lens changes.

Current mode

Balanced

Balanced for general local use. Keeps the ranking neutral across personal and serving workflows.

Capabilities:

Select quantization to explore

Q5_K_M (High quality) — 39.4 GB, 205.5 tok/s, Runs well

39.4 GB required128.0 GB available

31% VRAM used

Fit status

Runs well

Decode

205.5 tok/s

TTFT

942 ms

Safe context

Memory

39.4 GB / 128.0 GB

Memory breakdown

Weights10.8 GB

KV Cache14.6 GB

Runtime1.2 GB

Headroom12.8 GB

See how fast it feels

See how fast it feelsStarCoder 15B on AMD Instinct MI250 128GB
1st promptCold start — includes initialization
>Write a Python function to sort a list
Here's a Python function that sorts a list using quicksort:

def quicksort(arr):
    if len(arr) <= 1:
        return arr
    pivot = arr[len(arr) // 2]
    left   = [x for x in arr if x < pivot]
    middle = [x for x in arr if x == pivot]
    right  = [x for x in arr if x > pivot]
    return quicksort(left) + middle + quicksort(right)

# Example usage
numbers = [3, 6, 8, 10, 1, 2, 1]
print(quicksort(numbers))  # [1, 1, 2, 3, 6, 8, 10]

The algorithm picks a pivot, partitions the list into smaller, equal, and larger elements, then recursively sorts each partition and concatenates the results.
2nd promptWarm — model ready, no init overhead
>Now add type hints and a docstring
def quicksort(arr: list[int]) -> list[int]:
    """Sort a list using the quicksort algorithm.

    Args:
        arr: List of integers to sort.

    Returns:
        A new sorted list.
    """
    if len(arr) <= 1:
        return arr
    pivot = arr[len(arr) // 2]
    left   = [x for x in arr if x < pivot]
    middle = [x for x in arr if x == pivot]
    right  = [x for x in arr if x > pivot]
    return quicksort(left) + middle + quicksort(right)

Added type hints for the input and return type, plus a Google-style docstring.
Estimated: 205.5 tok/s decode · 942ms TTFT (warm) · 514 tok/s prefill

What limits this setup

This setup is broadly balanced for this model.

No major red flags

This recommendation has enough memory headroom and acceptable estimated speed for the selected workload.

Best improvement path

Performance by workload

Workload	Grade	Fit	Decode	TTFT	Context
Chat	A	Runs well	205.5 tok/s	514 ms	8K
Coding	A	Runs well	205.5 tok/s	942 ms	8K
Agentic Coding	A	Runs well	205.5 tok/s	1370 ms	8K
Reasoning	A	Runs well	205.5 tok/s	1113 ms	8K
RAG	A	Runs well	205.5 tok/s	1713 ms	8K

Inference speed

StarCoder 15B inference speed — tokens per second by GPU & Mac

Estimated decode speed (tokens/sec) for StarCoder 15B at Q5_K_M across popular GPUs and Apple Silicon, using the fastest local runtime per device. Fastest is RTX 5090 32GB at ~113 tok/s. Speed is memory-bandwidth bound, so cards that fit the whole model in VRAM run far faster than ones that offload to system RAM.

GPU / Mac	Memory	Quant	Speed (tok/s)	Fits?
RTX 5090 32GB	32 GB	Q5_K_M	113.4	Tight
Mac Studio M3 Ultra 256GB	256 GB	Q5_K_M	52.6	Fits
Mac Studio M2 Ultra 128GB	128 GB	Q5_K_M	43.8	Fits
Mac Studio M1 Ultra 128GB	128 GB	Q5_K_M	41.6	Fits
RTX 4090 24GB	24 GB	Q5_K_M	36.3	Too big
RX 7900 XTX 24GB	24 GB	Q5_K_M	32.8	Too big
RTX 3090 24GB	24 GB	Q5_K_M	31.1	Too big
MacBook Pro M4 Max 128GB	128 GB	Q5_K_M	30.0	Fits
MacBook Pro M4 Max 64GB	64 GB	Q5_K_M	30.0	Fits
MacBook Pro M3 Max 64GB	64 GB	Q5_K_M	22.7	Fits
MacBook Pro M1 Max 64GB	64 GB	Q5_K_M	20.8	Fits
MacBook Pro M4 Pro 48GB	48 GB	Q5_K_M	18.3	Tight
RTX 4080 Super 16GB	16 GB	Q5_K_M	13.1	Too big
RTX 4070 12GB	12 GB	Q5_K_M	5.4	Too big
RTX 3060 12GB	12 GB	Q5_K_M	3.4	Too big
RTX 4060 8GB	8 GB	Q5_K_M	2.8	Too big

Estimates for single-stream decoding at Q5_K_M; real tokens/sec varies with prompt length, context, batch size, and runtime build. Prompt processing (prefill) is faster than the decode figures shown here.

Quantization options

How StarCoder 15B (15B params) fits at each quantization level on AMD Instinct MI250 128GB (128.0 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	5.9 GB	Low	B63
Q3_K_S	3	7.4 GB	Low	B63
NVFP4	4	8.4 GB	Medium	B63
Q4_K_M	4	9.2 GB	Medium	B63
Q5_K_M	5	10.8 GB	High	B64
Q6_K	6	12.3 GB	High	B64
Q8_0	8	16.1 GB	Very High	B64
F16Best for your GPU	16	30.7 GB	Maximum	B66

Get started

Copy-paste commands to run StarCoder 15B on your machine.

Run

lms load starcoder && lms server start

Your hardware

More models your AMD Instinct MI250 128GB can run

Model	Params	Grade	Decode
Devstral 2 123B Instruct	123B	S	31.5 tok/s
Qwen3-Coder 30B A3B Instruct	30.5B	S	329 tok/s
Qwen 3.5 27B	27B	S	142.7 tok/s
Qwen 3.6 27B	27B	S	143.1 tok/s
Qwen 3.5 122B A10B	122B	S	87.5 tok/s

Frequently asked questions

Can AMD Instinct MI250 128GB run StarCoder 15B?

Yes, AMD Instinct MI250 128GB can run StarCoder 15B with a A grade (Runs well). Expected decode speed: 205.5 tok/s.

How much VRAM does StarCoder 15B need?

StarCoder 15B (15B parameters) requires approximately 39.4 GB of memory with Q5_K_M quantization.

What is the best quantization for StarCoder 15B?

The recommended quantization for StarCoder 15B is Q5_K_M, which balances quality and memory efficiency.

What speed will StarCoder 15B run at on AMD Instinct MI250 128GB?

On AMD Instinct MI250 128GB, StarCoder 15B achieves approximately 205.5 tokens per second decode speed with a time-to-first-token of 942ms using Q5_K_M quantization.

Can AMD Instinct MI250 128GB run StarCoder 15B for coding?

For coding workloads, StarCoder 15B on AMD Instinct MI250 128GB receives a A grade with 205.5 tok/s and 8K context.

What context window can StarCoder 15B use on AMD Instinct MI250 128GB?

On AMD Instinct MI250 128GB, StarCoder 15B can safely use up to 8K tokens of context. The model's official context limit is 8K, but available memory constrains the safe maximum.

See all results for AMD Instinct MI250 128GB See all hardware for StarCoder 15B

Embed this result▼

Paste this snippet into any page to show a live fit card.

<iframe src="https://willitrunai.com/embed/starcoder-15b-on-instinct-mi250-128gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>

Preview:

Workload

Grade

Fit

Decode

TTFT

Context

Chat

Runs well

205.5 tok/s

514 ms

Coding

Runs well

205.5 tok/s

942 ms

Agentic Coding

Runs well

205.5 tok/s

1370 ms

Reasoning

Runs well

205.5 tok/s

1113 ms

RAG

Runs well

205.5 tok/s

1713 ms

StarCoder 15B inference speed — tokens per second by GPU & Mac

GPU / Mac

Memory

Quant

Speed (tok/s)

Fits?

RTX 5090 32GB

32 GB

Q5_K_M

113.4

Tight

Mac Studio M3 Ultra 256GB

256 GB

Q5_K_M

52.6

Fits

Mac Studio M2 Ultra 128GB

128 GB

Q5_K_M

43.8

Fits

Mac Studio M1 Ultra 128GB

128 GB

Q5_K_M

41.6

Fits

RTX 4090 24GB

24 GB

Q5_K_M

36.3

Too big

RX 7900 XTX 24GB

24 GB

Q5_K_M

32.8

Too big

RTX 3090 24GB

24 GB

Q5_K_M

31.1

Too big

MacBook Pro M4 Max 128GB

128 GB

Q5_K_M

30.0

Fits

MacBook Pro M4 Max 64GB

64 GB

Q5_K_M

30.0

Fits

MacBook Pro M3 Max 64GB

64 GB

Q5_K_M

22.7

Fits

MacBook Pro M1 Max 64GB

64 GB

Q5_K_M

20.8

Fits

MacBook Pro M4 Pro 48GB

48 GB

Q5_K_M

18.3

Tight

RTX 4080 Super 16GB

16 GB

Q5_K_M

13.1

Too big

RTX 4070 12GB

12 GB

Q5_K_M

5.4

Too big

RTX 3060 12GB

12 GB

Q5_K_M

3.4

Too big

RTX 4060 8GB

8 GB

Q5_K_M

2.8

Too big

Quantization options

How StarCoder 15B (15B params) fits at each quantization level on AMD Instinct MI250 128GB (128.0 GB usable).

Quant	Bits	VRAM	Quality	Fit
Q2_K	2	5.9 GB	Low	B63
Q3_K_S	3	7.4 GB	Low	B63
NVFP4	4	8.4 GB	Medium	B63
Q4_K_M	4	9.2 GB	Medium	B63
Q5_K_M	5	10.8 GB	High	B64
Q6_K	6	12.3 GB	High	B64
Q8_0	8	16.1 GB	Very High	B64
F16Best for your GPU	16	30.7 GB	Maximum	B66

Model

Params

Grade

Decode

Capabilities

Devstral 2 123B Instruct

123B

31.5 tok/s

Qwen3-Coder 30B A3B Instruct

30.5B

329 tok/s

Qwen 3.5 27B

27B

142.7 tok/s

Qwen 3.6 27B

27B

143.1 tok/s

Qwen 3.5 122B A10B

122B

87.5 tok/s

Frequently asked questions

Can AMD Instinct MI250 128GB run StarCoder 15B?

Yes, AMD Instinct MI250 128GB can run StarCoder 15B with a A grade (Runs well). Expected decode speed: 205.5 tok/s.

How much VRAM does StarCoder 15B need?

StarCoder 15B (15B parameters) requires approximately 39.4 GB of memory with Q5_K_M quantization.

What is the best quantization for StarCoder 15B?

The recommended quantization for StarCoder 15B is Q5_K_M, which balances quality and memory efficiency.

What speed will StarCoder 15B run at on AMD Instinct MI250 128GB?

On AMD Instinct MI250 128GB, StarCoder 15B achieves approximately 205.5 tokens per second decode speed with a time-to-first-token of 942ms using Q5_K_M quantization.

Can AMD Instinct MI250 128GB run StarCoder 15B for coding?

For coding workloads, StarCoder 15B on AMD Instinct MI250 128GB receives a A grade with 205.5 tok/s and 8K context.

What context window can StarCoder 15B use on AMD Instinct MI250 128GB?

On AMD Instinct MI250 128GB, StarCoder 15B can safely use up to 8K tokens of context. The model's official context limit is 8K, but available memory constrains the safe maximum.