Arao AI
Back to search results

Qwen3.5-27B · IQ2_XS

Start a discussion
2x AMD Strix HaloStrata v0.25L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.5-27B (IQ2_XS) on 2x AMD Strix Halo with Strata, contributed by 1 independent source platforms; metrics are averages of published measurements.

213.76 tok/s

Decode

Decode speed

822.5 tok/s

Prefill

Prefill speed

1.82 s

TTFT

Time to first token

29.32 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 · IQ2_M Decode 146.37 · Prefill 764.74 ·
  2. Strata · IQ2_XS (current)Decode 213.76 · Prefill 822.5 ·

Core figures

Decode (avg)
213.76 tok/s
Prefill (avg)
822.5 tok/s
TTFT (avg)
1.82 s
VRAM (avg)
29.32 GB
MTP acceptance rate
0.76
TTFB
1.71 GB
Power draw
65.1 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.5-27B
Quantization
IQ2_XS
Framework
Strata
Version
v0.25
Context length
32768 tokens
Batch size
512
GPU layers
5
Flash Attention
Off

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)
29.32 GB
OS
Windows 11 23H2
Driver
amdgpu 6.10
CUDA
N/A
Power draw
65.1 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

    v0.25 · Windows 11 23H2 · CUDA N/A · 32768 ctx

    213.76 tok/s

    Decode

    822.5 tok/s

    Prefill

    1.82 s

    TTFT

    29.32 GB

    VRAM

    0.76

    MTP

    65.1 W

    Power draw

    L2 batch seed 2026-10-16: Strata on hw_id=88 running model_id=50 at IQ2_XS