166 lines
6.5 KiB
Markdown
166 lines
6.5 KiB
Markdown
<div align="center">
|
|
|
|
# haloq38flash
|
|
|
|
**qwen3.8-flash-next on amd strix halo — 91g quant, 56 tok/s, 262k context**
|
|
|
|
[](https://huggingface.co/julianmb/Qwen3.8-Flash-Next-IQ4_XS-GGUF)
|
|
[](https://github.com/Nathanw1014/llama.cpp/tree/strix-halo-vulkan)
|
|
[](https://huggingface.co/Qwen/Qwen3.8-Flash-Next/blob/main/LICENSE)
|
|
|
|
[](#results)
|
|
[](#results)
|
|
[](#the-converter-bug)
|
|
|
|
*every published quant byte-traced back to the official checkpoint*
|
|
|
|
</div>
|
|
|
|
---
|
|
|
|
## results
|
|
|
|
engine: [nathanw1014/llama.cpp `strix-halo-vulkan`](https://github.com/Nathanw1014/llama.cpp/tree/strix-halo-vulkan) · vulkan/radv · q8_0 kv · `-ub 2048` · temp 0
|
|
|
|
| depth | plain pp/tg | mtp pp/tg |
|
|
|------:|:-----------:|:---------:|
|
|
| 0 | 92.5 / 29.9 | 87.0 / **53.1** |
|
|
| 8k | 480 / 24.1 | 458 / **56.4** |
|
|
| 32k | 397 / 20.1 | 379 / **30.2** |
|
|
| 128k | 222 / 11.0 | 214 / 18.6 |
|
|
| 256k | 139 / 6.2 | — |
|
|
|
|
> [!NOTE]
|
|
> no collapse through 32k. the 128k+ falloff is context-mechanics
|
|
> (sparse-attention indexer), not quant size — see the reversal below.
|
|
|
|
<details>
|
|
<summary><b>the 128k reversal — the PLE quant loses under MTP at depth</b></summary>
|
|
|
|
at ≤32k the PLE quant wins everywhere. at 128k under mtp it *loses* to the
|
|
static 116g (18.6 vs 26.9 t/s). plausible mechanism: iq4_nl noise in the
|
|
n-gram table compounds over deep history and lowers draft acceptance.
|
|
single runs, n=1 caveat. pick your file by use case — see the table above.
|
|
|
|
</details>
|
|
|
|
---
|
|
|
|
## 📦 published quants
|
|
|
|
[huggingface.co/julianmb/Qwen3.8-Flash-Next-IQ4_XS-GGUF](https://huggingface.co/julianmb/Qwen3.8-Flash-Next-IQ4_XS-GGUF)
|
|
|
|
| file | size | pick it when |
|
|
|------|------|:------------:|
|
|
| `...-IQ4_XS-`**`PLE`**`.gguf` | 91 giB | ctx ≤ 32k — wins everywhere, mtp to 56 t/s |
|
|
| `...-IQ4_XS.gguf` | 116 giB | ctx ≥ 128k — faster mtp at depth, wider fork compat |
|
|
| `mtp-...-Q8_0.gguf` | 3.9 giB | mtp sidecar for nathanw1014-lineage engines |
|
|
|
|
<details>
|
|
<summary><b>the PLE cut — why the 51b n-gram table tolerates 4-bit</b></summary>
|
|
|
|
the PLE table is gathered 16 random rows per token via hash lookup — there is
|
|
no matmul on the table itself, and no two consecutive tokens hit the same rows.
|
|
the rows tolerate iq4_nl (4.25 bpw) with no measurable degradation across the
|
|
depth sweep. the cut: `--tensor-type "per_layer_token_embd=IQ4_XS"` on our
|
|
quantizer → 54g → 27g.
|
|
|
|
**fork caveat:** engines that feed gathered PLE rows straight into mul_mat as
|
|
quantized B operands assert (apepojken-class, ggml-vulkan.cpp:7794). verified
|
|
working on nathanw1014 strix-halo-vulkan and rocmfpx.
|
|
|
|
</details>
|
|
|
|
### pick your setup
|
|
|
|
| your use case | quant | engine | ctx | expect |
|
|
|---|---|---|---|---|
|
|
| coding agents, chat | 91g PLE | [nathanw1014 vulkan](https://github.com/Nathanw1014/llama.cpp/tree/strix-halo-vulkan) + mtp | ≤ 32k | 56 t/s |
|
|
| long documents | 116g static | same engine + mtp | 128k | 27 t/s |
|
|
| full rag / research | 116g static | [rocm 10 container](https://github.com/MorezMartin/engramhalo-rocm10) + ssd streaming | 262k | 14 t/s |
|
|
|
|
---
|
|
|
|
## 🐛 the converter bug
|
|
|
|
our first quant printed deterministic garbage at temp 0. bisect to root cause:
|
|
|
|
- experts, gdn reorder, ple scale, metadata: all innocent
|
|
- **97 of 388 f32 tensors differed by exactly 1.0** — every hyper-connection
|
|
norm shipped raw where the runtime expects `raw + 1`
|
|
- cause: the checkpoint nests hyper-connections under
|
|
`attn_hyper_connection` / `mlp_hyper_connection` / `hyper_connection_mixer`,
|
|
and those names hit early-return branches in the converter that bypass the
|
|
generic `norm.weight → +1` rule
|
|
|
|
> [!WARNING]
|
|
> **any fork rolling its own qwen4exp converter must fold `(1 + w)` into the
|
|
> hyper-connection gammas.** upstream runtime documents the contract at
|
|
> `qwen4exp.cpp:231`. miss it and every layer normalizes wrong — garbage
|
|
> from layer 0, all shapes correct, all shape-only tests pass.
|
|
|
|
fix + regression test: rocmfpx `port-qwen4exp` commit `61b6a3b48`
|
|
([pr charlie12345/ROCmFPX#98](https://github.com/charlie12345/ROCmFPX/pull/98))
|
|
|
|
---
|
|
|
|
## 🐳 docker
|
|
|
|
```bash
|
|
git clone https://github.com/julianmb/haloq38flash && cd haloq38flash
|
|
docker compose up --build
|
|
# serve on :8080 — vulkan/radv, no rocm install needed
|
|
```
|
|
|
|
add the mtp sidecar for 56 t/s:
|
|
|
|
```bash
|
|
docker compose run qwen38-flash-next /app/llama-server \
|
|
-md /models/mtp-Qwen3.8-Flash-Next-Q8_0.gguf \
|
|
--spec-type draft-mtp --spec-draft-n-max 6 --spec-draft-p-min 0.75
|
|
```
|
|
|
|
---
|
|
|
|
## 🔬 engine merge (paused)
|
|
|
|
we merged nathanw1014's branch (175 commits: vulkan perf stack, qwen4exp
|
|
runtime past the squash, lazy ple, spec-decode fixes) into ggml-org master —
|
|
one engine with lazy ple (262k on 66g resident) + depth fixes + master's
|
|
general improvements. paused at 53% build: the branches diverged semantically
|
|
in shared enum/base-class files.
|
|
|
|
- [`docs/engine-cherry-pick-plan.md`](docs/engine-cherry-pick-plan.md) — full
|
|
175-commit classification
|
|
- [`docs/engine-merge-status.md`](docs/engine-merge-status.md) — resume point
|
|
|
|
---
|
|
|
|
## 📁 layout
|
|
|
|
| path | what |
|
|
|------|------|
|
|
| `models/` | symlink farm to `/mnt/ssd2/models/` (never in git) |
|
|
| `results/` | dated receipts, one per investigation |
|
|
| `scripts/` | conversion pipeline, oracle, depth bench, resume helpers |
|
|
| `reddit/` | archived community threads that drove the investigation |
|
|
| `docs/` | engine merge plan, cherry-pick classification |
|
|
|
|
---
|
|
|
|
## ⚠️ operational gotchas (128g strix halo)
|
|
|
|
- always `-c 8192`-bounded ctx + `timeout` + `/usr/bin/time -v` — the gguf
|
|
default 262144 + full offload hard-hung this box once
|
|
- `vm.dirty_ratio=15 / dirty_background_ratio=5` — the 191g ple conversion
|
|
memmap wedges `balance_dirty_pages` for hours at kernel defaults
|
|
- conversion peak: ple scratch (191g) + f16 output (354g) coexist — budget
|
|
~560g free
|
|
- `pkill -x llama-cli`, never `-f` (matches your own wrapper shell)
|
|
- gpu memory is shared with everything else on the apu — two engines cannot
|
|
hold ~90g+ models simultaneously without an oom cascade
|
|
|
|
---
|
|
|
|
license: [qwen community license 1.0](https://huggingface.co/Qwen/Qwen3.8-Flash-Next/blob/main/LICENSE) · base model: [Qwen/Qwen3.8-Flash-Next](https://huggingface.co/Qwen/Qwen3.8-Flash-Next)
|