Arao AI
Back to search results

Llama-3-8B-Instruct · IQ2_M

Start a discussion
NVIDIA RTX PRO 6000 Blackwell 96GB96GBllama.cpp v0.24L2 Cross-framework verified

This page aggregates 1 real-world runs of Llama-3-8B-Instruct (IQ2_M) on NVIDIA RTX PRO 6000 Blackwell 96GB with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

345.66 tok/s

Decode

Decode speed

1,718.2 tok/s

Prefill

Prefill speed

1.46 s

TTFT

Time to first token

10.84 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

16 days ago

Last verified

Performance

  1. llama.cpp · IQ2_M (current)Decode 345.66 · Prefill 1718.2 ·
  2. llama.cpp · Q5_K_M Decode 194.41 · Prefill 877.5 ·
  3. Ollama · Q3_K_M Decode 249.19 · Prefill 1206.14 ·
  4. Ollama · Q6_K Decode 185.73 · Prefill 814.54 ·

Core figures

Decode (avg)
345.66 tok/s
Prefill (avg)
1,718.2 tok/s
TTFT (avg)
1.46 s
VRAM (avg)
10.84 GB
MTP acceptance rate
—
TTFB
2.77 GB
Power draw
398.5 W

Configuration

Member-level fields are taken from the most recent run

Model
Llama-3-8B-Instruct
Quantization
IQ2_M
Framework
llama.cpp
Version
v0.24
Context length
32768 tokens
Batch size
512
GPU layers
20
Flash Attention
On

Hardware

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

GPU
NVIDIA RTX PRO 6000 Blackwell 96GB
Nominal VRAM
96 GB
Measured VRAM (avg)
10.84 GB
OS
Windows 11 24H2
Driver
570.86
CUDA
12.8
Power draw
398.5 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

    v0.24 · Windows 11 24H2 · CUDA 12.8 · 32768 ctx

    345.66 tok/s

    Decode

    1,718.2 tok/s

    Prefill

    1.46 s

    TTFT

    10.84 GB

    VRAM

    —

    MTP

    398.5 W

    Power draw

    L2 batch seed 2026-10-19: llama.cpp on hw_id=68 running model_id=69 at IQ2_M