Arao AI
Back to search results

gemma-4-E4B-it · IQ3_S

Start a discussion
NVIDIA RTX 309024GBllama.cpp v0.42L2 Cross-framework verified

This page aggregates 1 real-world runs of gemma-4-E4B-it (IQ3_S) on NVIDIA RTX 3090 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

121.92 tok/s

Decode

Decode speed

447.63 tok/s

Prefill

Prefill speed

1.67 s

TTFT

Time to first token

13.19 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

12 days ago

Last verified

Performance

  1. llama.cpp · IQ3_S (current)Decode 121.92 · Prefill 447.63 ·
  2. vLLM · IQ4_XS Decode 102.62 · Prefill 342.57 ·

Core figures

Decode (avg)
121.92 tok/s
Prefill (avg)
447.63 tok/s
TTFT (avg)
1.67 s
VRAM (avg)
13.19 GB
MTP acceptance rate
—
TTFB
0.98 GB
Power draw
277.9 W

Configuration

Member-level fields are taken from the most recent run

Model
gemma-4-E4B-it
Quantization
IQ3_S
Framework
llama.cpp
Version
v0.42
Context length
32768 tokens
Batch size
512
GPU layers
28
Flash Attention
Off

Hardware

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

GPU
NVIDIA RTX 3090
Nominal VRAM
24 GB
Measured VRAM (avg)
13.19 GB
OS
Ubuntu 22.04
Driver
565.57
CUDA
12.8
Power draw
277.9 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-24

    v0.42 · Ubuntu 22.04 · CUDA 12.8 · 32768 ctx

    121.92 tok/s

    Decode

    447.63 tok/s

    Prefill

    1.67 s

    TTFT

    13.19 GB

    VRAM

    —

    MTP

    277.9 W

    Power draw

    L2 batch seed 2026-10-24: llama.cpp on hw_id=2 running model_id=67 at IQ3_S