{"type":"post","id":"lfortran-compiled-fortran-qwen3-8-27b-inference-benchmarked-68fa99","slug":"lfortran-compiled-fortran-qwen3-8-27b-inference-benchmarked-68fa99","title":"LFortran-compiled Fortran Qwen3.8-27B inference benchmarked against vLLM and llama.cpp on H200","description":"A Fortran Discourse author reports running Qwen3.8-27B 8-bit inference with an LFortran-compiled GPU implementation at 102 tokens per second on an NVIDIA H200.","retrieval_nugget":"A Fortran Discourse author reports running Qwen3.8-27B 8-bit inference with an LFortran-compiled GPU implementation at 102 tokens per second on an NVIDIA H200. The comparison is a single-machine experiment, not an independent benchmark, but it makes a concrete case for testing compiler-based GPU offload beyond established inference runtimes.","published_at":"2026-11-16","updated_at":"2026-11-16","record_date":"2026-11-16","date_kind":"published_at","topics":["inference"],"entities":["LFortran","Qwen3.8-27B","vLLM","llama.cpp","NVIDIA H200","Fortran Discourse"],"related_patterns":["agent-economics-moves-to-completed-work"],"novelty_label":"new_fact","what_changed":"A new H200 benchmark reports LFortran-compiled Fortran Qwen3.8-27B inference at 102 tokens per second, ahead of vLLM and llama.cpp but below a custom C/CUDA implementation.","prior_refs":["https://newruntime.com/posts/pli-4-0-0-pre-release-open-source-6d6a03/"],"source_url":"https://fortran-lang.discourse.group/t/performance-portable-llm-inference-in-fortran-beating-llama-cpp-vllm-on-h200/11309","source_urls":["https://fortran-lang.discourse.group/t/performance-portable-llm-inference-in-fortran-beating-llama-cpp-vllm-on-h200/11309"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:lfortran-compiled-fortran-qwen3-8-27b-inference-benchmarked-68fa99","status":"published","visuals":[],"editor":{"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]},"routes":{"html":"https://newruntime.com/posts/lfortran-compiled-fortran-qwen3-8-27b-inference-benchmarked-68fa99/","markdown":"https://newruntime.com/posts/lfortran-compiled-fortran-qwen3-8-27b-inference-benchmarked-68fa99.md","json":"https://newruntime.com/posts/lfortran-compiled-fortran-qwen3-8-27b-inference-benchmarked-68fa99.json"},"markdown_url":"https://newruntime.com/posts/lfortran-compiled-fortran-qwen3-8-27b-inference-benchmarked-68fa99.md","next_reads":[{"type":"topic","path":"/topics/inference/","url":"https://newruntime.com/topics/inference/","title":"Inference","reason":"Explore the inference topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/inference.md","json_url":"https://newruntime.com/topics/inference.json"},{"type":"collection","path":"/posts/","url":"https://newruntime.com/posts/","title":"Field Notes","reason":"Open the latest editorial analysis.","media_type":"text/html","json_url":"https://newruntime.com/posts.json"},{"type":"collection","path":"/sources/","url":"https://newruntime.com/sources/","title":"Sources","reason":"See the original sources this archive cites.","media_type":"text/html"}]}
