Qwen3.8-27B · SGLang · NVFP4 + DSpark · conc-32
status: blocked
Configuration
| Model | Qwen/Qwen3.8-27B |
|---|
| Company | Alibaba |
|---|
| Family | Qwen |
|---|
| Parameters | 27B (dense, hybrid linear-attn) |
|---|
| Engine | SGLang + DSpark (speculative decoding) |
|---|
| Quant / precision | NVFP4 |
|---|
| Why this quant | MiaAI-Lab SGLang DGX Spark recipe (https://github.com/MiaAI-Lab/Qwen3.8-27B-SGLang-DGX-Spark). RadixArk NVFP4 W4A4 of Qwen3.8-27B on the cookbook-pinned lmsysorg/sglang:qwen38-27b image, DSpark block-7 / 8 draft tokens (RadixArk/Qwen3.8-27B-DSpark, unquant), mem-fraction-static 0.90, FP8 KV, GDN bf16, extra_buffer_lazy, native 262K, YaRN off. conc-32 point of the ShareGPT sweep (client conc 1/2/4/8/16/32). |
|---|
| Download | RadixArk/Qwen3.8-27B-NVFP4 |
|---|
| Context window | 262144 |
|---|
| Input modalities | text, image, video (served text-only here) |
|---|
Measured results
| Prefill tok/s | — |
|---|
| Decode tok/s | — |
|---|
| Peak memory (GB) | — |
|---|
| Completed | 2026-08-20 05:25 +0800 |
|---|
Full run command
# Not run — memory ceiling. Last attempted boot used MAX_CONCURRENT_REQUESTS=8.
cd ~/Desktop/Qwen3.8-27B-SGLang-DGX-Spark
MAX_CONCURRENT_REQUESTS=8 ./start-dspark.sh
python3 ~/Desktop/autobench/scripts/bench-serving.py \
--base-url http://127.0.0.1:8888 --model qwen3.8-27b-sglang \
--dataset ~/Desktop/autobench/benchmark_data/ShareGPT_V3_unfiltered_cleaned_split.json \
--num-prompts 1000 --max-seconds 900 --concurrency 32 --max-tokens 256
qwen3.8-27b
Alibaba
Qwen
NVFP4
16-40B
Spark recipe
conc-32
Notes
BLOCKED — memory ceiling, same wall as c16. DSpark never served 16 concurrent requests on this box at --mem-fraction-static 0.90; conc-32 is strictly harder (GDN pool 32×4 = 128 slots ≈ 28 GB mamba alone). Last good point is conc-8.