Arao AI
Back to search results

Qwen3.5-9B · FP16

Start a discussion
2x AMD Strix HaloOllama v0.27L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.5-9B (FP16) on 2x AMD Strix Halo with Ollama, contributed by 1 independent source platforms; metrics are averages of published measurements.

58.36 tok/s

Decode

Decode speed

233.8 tok/s

Prefill

Prefill speed

1.32 s

TTFT

Time to first token

37.06 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

2 days ago

Last verified

Performance

  1. Ollama · FP16 (current)Decode 58.36 · Prefill 233.8 ·
  2. Strata · IQ3_XS Decode 234.55 · Prefill 886.87 ·

Core figures

Decode (avg)
58.36 tok/s
Prefill (avg)
233.8 tok/s
TTFT (avg)
1.32 s
VRAM (avg)
37.06 GB
MTP acceptance rate
—
TTFB
0.47 GB
Power draw
98 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.5-9B
Quantization
FP16
Framework
Ollama
Version
v0.27
Context length
8192 tokens
Batch size
512
GPU layers
99
Flash Attention
On

Hardware

Nominal and measured figures are shown side by side; whether it runs is the reader's call

GPU
2x AMD Strix Halo
Measured VRAM (avg)
37.06 GB
OS
Ubuntu 22.04
Driver
565.57
CUDA
N/A
Power draw
98 W

Sources & evidence

1 measured records in total, each traceable to its original source

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-16

    v0.27 · Ubuntu 22.04 · CUDA N/A · 8192 ctx

    58.36 tok/s

    Decode

    233.8 tok/s

    Prefill

    1.32 s

    TTFT

    37.06 GB

    VRAM

    —

    MTP

    98 W

    Power draw

    L2 batch seed 2026-10-16: Ollama on hw_id=88 running model_id=49 at FP16