Nicheloom visits: 58,069 all-time · 7,637 in the last 7 days

The opportunity tracker for new startups.

Today: 1,953 launches · Niche of the day: specialized calculation tools · See what launched today →

elpis-on-7900xtx

Qwen3.8-27B EXL3 + DFlash2 speculative decoding on an AMD RX 7900 XTX (exllamav3-rocm): recipe, patches, benchmarks

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.

Want to reach these readers? Sponsor an issue.

GITHUB · launched Nov 1, 2026 · 18 upvotes (top 91%) · 0 views

Niche: llm inference engines · 133 launches, 4 in the last 7 days ▼ -33% · Explore this niche →

Category and tags
  • Horizontal
  • Model & infra
  • Developer
  • AI-native
  • Hobby / open-source project
  • amd
  • exllamav3
  • llm-inference
  • qwen
  • rocm
  • rx-7900-xtx
  • +1 more

Visit website ↗ Source listing ↗

More like this: all 133 launches in llm inference engines →

Could you build this?

Partial The docs, benchmarks, and recipe are easy to produce, but the core value is inference-engine patches for ROCm and speculative decoding on a consumer AMD GPU, which needs low-level GPU and ML systems expertise that vibe coding alone won't supply.

What it would actually take: Start from the existing exllamav3 codebase and its ROCm port, then add the Qwen3 model support, the EXL3 quantization path, and a draft-model speculative decoding loop. The hard parts are HIP kernel debugging and performance tuning on RDNA3 hardware, plus validating correctness of the quantized and speculative outputs. This takes an engineer comfortable with CUDA/HIP kernels, quantization formats, and a 7900 XTX for benchmarking.

Competitors

Other products that read as similar to this one : 1341 launches clear the similarity bar, closest 8 shown.

Attention rank: #1229 of 1342 (itself plus its competitors, highest first, normalized so YC and Product Hunt are compared fairly).

Launched 361 days after the earliest competitor.

  • qwen38-exl3-dflash2 · github · 2026-09-15 · 18 upvotes · similarity 0.77
  • exl3xpu · github · 2026-09-22 · 10 upvotes · similarity 0.65
  • Infernix · github · 2026-10-08 · 98 upvotes · similarity 0.57
  • genpark-speculative-decoding-verifier-skill · github · 2026-09-28 · 8 upvotes · similarity 0.57
  • local-llm-tuning-guide · github · 2026-10-16 · 21 upvotes · similarity 0.56
  • d4r · github · 2026-09-27 · 133 upvotes · similarity 0.56
  • ASpec · github · 2026-11-05 · 48 upvotes · similarity 0.55
  • strixite · github · 2026-10-04 · 11 upvotes · similarity 0.55

Related data

The niche, labels, vibe code rating and competitor comparison on this page are analysis by Nicheloom. If you use them, please credit nicheloom.com. How you may use our data.