Native fp8 E4M3 GGUF of Qwen3-8B for AMD RDNA4 (gfx1201 - Radeon AI PRO R9700 / RX 9070 / 9070 XT / W-series), quantized with AMD Quark from the full-precision BF16 weights by The Rock8.

This is not a dequant-to-f16 shim: The Rock8's llama.cpp fork runs fp8 on RDNA4's native WMMA fp8 tensor cores for prefill and v_dot4_f32_fp8_fp8 for decode. No-compromise quality-per-byte 8-bit.

What it is

Format: fp8 E4M3 (F8E4M3), block-scaled, produced by AMD Quark from BF16.
Target: AMD RDNA4 / gfx1201 (Radeon AI PRO R9700, RX 9070 / 9070 XT).
Runtime: The Rock8 (llama.cpp fork with native RDNA4 fp8 kernels) on TheRock ROCm 7.13.
File: Qwen3-8B-Quark-F8E4M3.gguf (9.2 GB).

Source model + license

Source: Qwen/Qwen3-8B.
License: Apache-2.0 (inherited from the source model; redistribution of this quantized derivative is permitted with attribution). This is a derivative work.

Validation (real gfx1201 hardware)

Metric	Value
Perplexity (wikitext, 20 chunks, n_ctx=512)	10.93
Prefill `pp512`	4688 t/s
Decode `tg128`	58.0 t/s
Coherence check ("dried grape" -> "raisin")	Pass

Benched on a single R9700 (gfx1201).

Run it

llama.cpp (The Rock8 fork)

bash

# chat
llama-cli   -m Qwen3-8B-Quark-F8E4M3.gguf -ngl 99 -p "What do you call a dried grape? Answer in one word."
# bench
llama-bench -m Qwen3-8B-Quark-F8E4M3.gguf -ngl 99 -p 512 -n 128

Lemonade appliance (container)

bash

podman run -d --rm --runtime crun --name lemonade \
  --device /dev/kfd --device /dev/dri \
  --group-add keep-groups --security-opt seccomp=unconfined \
  -v /path/to/quacken-8b:/models:ro \
  -e MODEL=/models/Qwen3-8B-Quark-F8E4M3.gguf -e MODEL_NAME=Quacken-8B-FP8 \
  -e HIP_VISIBLE_DEVICES=0 -p 13305:13305 \
  ghcr.io/the-monk/the-rock8:rdna4-tr713 serve

curl -s http://localhost:13305/api/v1/chat/completions -H 'Content-Type: application/json' \
  -d '{"model":"user.Quacken-8B-FP8","messages":[{"role":"user","content":"What do you call a dried grape? Answer in one word. /no_think"}],"max_tokens":16}'

Container (same image on each registry; --runtime crun is required for GPU): ghcr.io/the-monk/the-rock8:rdna4-tr713 - docker.io/gorilla4x/the-rock8:rdna4-tr713 - quay.io/the-monk/the-rock8:rdna4-tr713 (images may not be pushed to every registry yet).

The Rock8 - RDNA4 fp8 (links)

GitHub: The-Rock8 - kernels, patch series, appliance recipe, full feature doc.
Collection: The Rock8 - RDNA4 fp8.
Sibling model: Quacken-R1-14B-FP8.
3 more coming: Quacken-27B-FP8 (Qwen3.6-27B), Qwen3.6-35B-A3B (MoE) fp8, and Ornith-1.0-35B fp8 (in final validation).

Every artifact links to the others - land on any one, reach them all.

Gorilla4X/Quacken-8B-FP8

Quacken-8B-FP8

The Rock8 - Got any weights? 💪🦆