Arao AI
Back to search results

Hy-Memory-30B-A3B · IQ1_S

Start a discussion
AMD Ryzen AI Max+ 395128GBllama.cpp v0.23L2 Cross-framework verified

This page aggregates 1 real-world runs of Hy-Memory-30B-A3B (IQ1_S) on AMD Ryzen AI Max+ 395 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

83.71 tok/s

Decode

Decode speed

315.21 tok/s

Prefill

Prefill speed

1.37 s

TTFT

Time to first token

28.56 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 · IQ1_S (current)Decode 83.71 · Prefill 315.21 ·
  2. vLLM · Q5_K_S Decode 50.65 · Prefill 184.14 ·

Core figures

Decode (avg)
83.71 tok/s
Prefill (avg)
315.21 tok/s
TTFT (avg)
1.37 s
VRAM (avg)
28.56 GB
MTP acceptance rate
—
TTFB
0.67 GB
Power draw
48 W

Configuration

Member-level fields are taken from the most recent run

Model
Hy-Memory-30B-A3B
Quantization
IQ1_S
Framework
llama.cpp
Version
v0.23
Context length
2048 tokens
Batch size
256
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
AMD Ryzen AI Max+ 395
Nominal VRAM
128 GB
Measured VRAM (avg)
28.56 GB
OS
Windows 11 23H2
Driver
amdgpu 6.12
CUDA
12.6
Power draw
48 W

Get started

Original model resources

Base model resources may not match this quantization. Use the verified deployment weights above when available.

Sources & evidence

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

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

    v0.23 · Windows 11 23H2 · CUDA 12.6 · 2048 ctx

    83.71 tok/s

    Decode

    315.21 tok/s

    Prefill

    1.37 s

    TTFT

    28.56 GB

    VRAM

    —

    MTP

    48 W

    Power draw

    L2 batch seed 2026-10-19: llama.cpp on hw_id=24 running model_id=87 at IQ1_S