Arao AI
Back to search results

Qwen3.8-9B · IQ3_XS

Start a discussion
NVIDIA H100 80GB HBM380GBllama.cpp v0.21L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.8-9B (IQ3_XS) on NVIDIA H100 80GB HBM3 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

324.47 tok/s

Decode

Decode speed

1,251.53 tok/s

Prefill

Prefill speed

2.14 s

TTFT

Time to first token

13.82 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

17 days ago

Last verified

Performance

  1. llama.cpp · IQ3_XS (current)Decode 324.47 · Prefill 1251.53 ·
  2. Ollama · IQ3_XS Decode 374.78 · Prefill 1868.44 ·

Core figures

Decode (avg)
324.47 tok/s
Prefill (avg)
1,251.53 tok/s
TTFT (avg)
2.14 s
VRAM (avg)
13.82 GB
MTP acceptance rate
—
TTFB
2.6 GB
Power draw
617.9 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.8-9B
Quantization
IQ3_XS
Framework
llama.cpp
Version
v0.21
Context length
2048 tokens
Batch size
512
GPU layers
33
Flash Attention
Off

Hardware

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

GPU
NVIDIA H100 80GB HBM3
Nominal VRAM
80 GB
Measured VRAM (avg)
13.82 GB
OS
Ubuntu 22.04
Driver
570.86
CUDA
12.8
Power draw
617.9 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-19

    v0.21 · Ubuntu 22.04 · CUDA 12.8 · 2048 ctx

    324.47 tok/s

    Decode

    1,251.53 tok/s

    Prefill

    2.14 s

    TTFT

    13.82 GB

    VRAM

    —

    MTP

    617.9 W

    Power draw

    L2 batch seed 2026-10-19: llama.cpp on hw_id=55 running model_id=52 at IQ3_XS