Arao AI
Back to search results

Qwen3.8-9B · IQ2_M

Start a discussion
2x AMD Strix HaloStrata v0.53L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.8-9B (IQ2_M) on 2x AMD Strix Halo with Strata, contributed by 1 independent source platforms; metrics are averages of published measurements.

243.64 tok/s

Decode

Decode speed

1,152.92 tok/s

Prefill

Prefill speed

0.18 s

TTFT

Time to first token

11.93 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

2 days ago

Last verified

Performance

  1. llama.cpp · IQ1_S Decode 185.41 · Prefill 664.26 ·
  2. llama.cpp · IQ4_XS Decode 145.51 · Prefill 650.25 ·
  3. Strata · IQ2_M (current)Decode 243.64 · Prefill 1152.92 ·
  4. Strata · IQ3_S Decode 220.16 · Prefill 1160.56 ·

Core figures

Decode (avg)
243.64 tok/s
Prefill (avg)
1,152.92 tok/s
TTFT (avg)
0.18 s
VRAM (avg)
11.93 GB
MTP acceptance rate
0.54
TTFB
1.95 GB
Power draw
82.8 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.8-9B
Quantization
IQ2_M
Framework
Strata
Version
v0.53
Context length
4096 tokens
Batch size
256
GPU layers
15
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)
11.93 GB
OS
macOS 15.0
Driver
560.94
CUDA
N/A
Power draw
82.8 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

    v0.53 · macOS 15.0 · CUDA N/A · 4096 ctx

    243.64 tok/s

    Decode

    1,152.92 tok/s

    Prefill

    0.18 s

    TTFT

    11.93 GB

    VRAM

    0.54

    MTP

    82.8 W

    Power draw

    L2 batch seed 2026-10-16: Strata on hw_id=88 running model_id=52 at IQ2_M