[{"data":1,"prerenderedAt":155},["ShallowReactive",2],{"project-37805810":3},{"id":4,"name":5,"line":6,"line_source":7,"title":8,"url":9,"domain":10,"maker":11,"posted_at":12,"year":13,"points":14,"comments":15,"rank_in_year":16,"shown_in_year":17,"top":18,"tier":19,"category":20,"subcategory":21,"format":22,"keywords":23,"reception":29,"sentiment":30,"stance_mix":31,"scored_comments":34,"aspects":35,"strip":37,"post_text":30,"what":40,"said":58,"maker_in_thread":93,"shape":99,"grid":102,"attempts":116,"same_week":117,"maker_shows":147,"maker_hidden":148,"reader_version":149,"card":150},"37805810","SimSIMD vs SciPy","How AVX-512 and SVE make SIMD nicer and ML 10x faster","title","Show HN: SimSIMD vs SciPy: How AVX-512 and SVE make SIMD nicer and ML 10x faster","https:\u002F\u002Fashvardanian.com\u002Fposts\u002Fsimsimd-faster-scipy\u002F","ashvardanian.com","ashvardanian",1696714899,2023,13,6,2112,3109,68,"shown","languages","simd numeric library","library",[24,25,26,27,28],"simd","vector similarity","avx-512","sve","scipy","quiet",null,{"positive":32,"neutral":32,"mixed":33,"critical":33},1,0,2,{"idea":36},{"positive":32,"critical":33},[33,38,33,39,33,33],5,3,{"problem":41,"audience":30,"pricing":30,"motivation":30,"time_to_build":30,"stack":48,"openness":57,"stage":57,"team_size":57},{"value":42,"source":43,"evidence":44},"Vector similarity and distance computations are slow in common libraries like NumPy and SciPy","maker",[45],{"id":46,"by":11,"text":47},"37805811","Over the years, the merits of Intel’s 512-bit Advanced Vector eXtensions (AVX-512) have been extensively debated. Introduced in 2014, CPUs took time to offer robust support. In…",[49],{"value":50,"source":51,"evidence":52},"C","inferred",[53],{"id":54,"by":55,"text":56},"37808036","westurner","I encourage one to merge into e.g. {NumPy, SciPy, }; are there PRs? Though SymPy.physics only yet supports X,Y,Z vectors and doesn't mention e.g. \"jaccard\"?, FWIW:…","unknown",{"praise":59,"objections":68,"requests":69,"questions":78,"alternatives":83},[60],{"value":61,"source":62,"evidence":63},"Interesting performance gains from SIMD on modern CPUs","commenter",[64,65],{"id":54,"by":55,"text":56},{"id":66,"by":55,"text":67},"37812219","Np. Thanks for the optimizations. From \"PostgresML is 8-40x faster than Python HTTP microservices\" (2023) https:\u002F\u002Fnews.ycombinator.com\u002Fitem?id=33270638 : > Apache Ballista and…",[],[70,74],{"value":71,"source":62,"evidence":72},"Merge the optimizations into NumPy or SciPy via PRs",[73],{"id":54,"by":55,"text":56},{"value":75,"source":62,"evidence":76},"Apply the optimizations to other projects such as CuPy, xtensor-fftw or llama.cpp",[77],{"id":54,"by":55,"text":56},[79],{"value":80,"source":62,"evidence":81},"Whether CuPy GPUs could benefit given they lack AVX-512 or SVE",[82],{"id":54,"by":55,"text":56},[84,89],{"value":85,"source":62,"evidence":86},"NumPy and SciPy",[87,88],{"id":46,"by":11,"text":47},{"id":54,"by":55,"text":56},{"value":90,"source":62,"evidence":91},"pgvector",[92],{"id":66,"by":55,"text":67},{"comments":34,"answered":32,"key":94},[95,96],{"id":46,"text":47},{"id":97,"text":98},"37808199","Hey, thanks for recommendations! Yes, I’m definitely open to contributing there.",{"commenters":34,"max_depth":100,"maker_replies":34,"first_comment_mins":33,"lifespan_hours":101},4,169,[103,105,107,109,111,114],[46,33,33,104,11,32],"project\u002Fpositive\u002Fperformance",[54,33,38,106,55,33],"project\u002Fpositive\u002Fidea",[97,33,33,108,11,32],"maker\u002Fneutral\u002Fnone",[66,33,39,110,55,33],"project\u002Fneutral\u002Fnone",[112,32,33,113,55,33],"37807941","other\u002Fneutral\u002Fnone",[115,34,33,113,55,33],"37885082",[],[118,123,128,132,137,142],{"id":119,"name":120,"points":121,"category":122},"37835995","eliot.blog",537,"creative",{"id":124,"name":125,"points":126,"category":127},"37782513","store",484,"web_saas",{"id":129,"name":130,"points":131,"category":127},"37818374","momo.coach",307,{"id":133,"name":134,"points":135,"category":136},"37792507","Rolepad",294,"productivity",{"id":138,"name":139,"points":140,"category":141},"37789371","extra-screen",281,"dev_tools",{"id":143,"name":144,"points":145,"category":146},"37848793","Obligator",273,"infrastructure",7,false,"v2",{"id":151,"name":5,"line":6,"ai":148,"cat":20,"year":13,"points":14,"comments":15,"rec":29,"s":30,"top":18,"kw":152,"maker":11,"domain":10,"tier":19,"attempts":33,"strip":153,"url":9,"mix":154,"praise":61,"objection":-1,"sub":21},37805810,[24,25,26,27,28],"050300",[32,32,33,33],1794036642979]