dsv41-flash-cmp170hx-recipe
DeepSeek-V4.1-Flash on 4× NVIDIA CMP 170HX with vLLM: up to 319 tok/s for one user, up to 765 tok/s at eight, 262K context, one-command container install
Get picks like this daily. Today's launch numbers, the top launches, AI and tech news, and a daily opportunity pick, straight to your inbox.
Visit website ↗ Source listing ↗
GITHUB · launched Oct 24, 2026 · 16 upvotes (top 71%)
- Horizontal
- Model & infra
- Developer
- AI-native
- Hobby / open-source project
- ampere
- cmp-170hx
- deepseek
- exl3
- local-llm
- moe
- +2 more
Similar launches
See all 310 similar launches →Could you build this?
No Running DeepSeek-class models efficiently on NVIDIA CMP 170HX mining cards requires low-level CUDA and vLLM kernel work, driver-level compatibility fixes, and hardware-specific performance tuning that vibe coding alone won't supply.
What it would actually take: The stack is vLLM built against a patched or older CUDA setup, with custom kernels or attention backends and tensor-parallel configuration across four GPUs. The hard parts are getting the mining-card hardware to work at all (these cards have crippled compute and memory paths) and reaching the reported throughput. That takes ML systems engineers who understand GPU kernels, quantization, and inference serving, plus repeated benchmarking on the same hardware.
Competitors
Other products that read as similar to this one : 310 launches clear the similarity bar, closest 8 shown.
Attention rank: #237 of 311 (itself plus its competitors, highest first, normalized so YC and Product Hunt are compared fairly).
Launched 286 days after the earliest competitor.
- dsv41-flash-pp4-170hx · github · 2026-09-14 · 7 upvotes · similarity 0.83
- dsv41-flash-offload · github · 2026-10-02 · 97 upvotes · similarity 0.72
- DeepSeek-V4.1-Flash-vLLM-DGX-Spark · github · 2026-09-10 · 65 upvotes · similarity 0.70
- DeepSeek-v4.1-Flash-DGX-Sparks · github · 2026-09-11 · 118 upvotes · similarity 0.66
- deepseek-v41-flash-mac-mini · github · 2026-09-11 · 88 upvotes · similarity 0.66
- DeepSeek-V4.1-Flash-1017-FP8-abliterated-4x-RTX-PRO-6000 · github · 2026-10-23 · 27 upvotes · similarity 0.66
- deepseek-v41-flash-spark · github · 2026-09-10 · 92 upvotes · similarity 0.64
- deepseek-v4.1-flash-4x-rtx-pro-6000 · github · 2026-09-10 · 43 upvotes · similarity 0.64
Other launches for this product
- No other launches for this product.