[{"data":1,"prerenderedAt":210},["ShallowReactive",2],{"project-10555692":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":35,"aspects":36,"strip":41,"post_text":43,"what":44,"said":78,"maker_in_thread":136,"shape":143,"grid":147,"attempts":173,"same_week":174,"maker_shows":144,"maker_hidden":202,"reader_version":203,"card":204},"10555692","TensorFlow","TensorFlow for AWS on a Real GPU","title","Show HN: TensorFlow for AWS on a Real GPU","https:\u002F\u002Fgithub.com\u002Fpavlovml\u002Ftensorflow\u002Fblob\u002Fmaster\u002FREADME.md","github.com","alexkern",1447358532,2015,77,12,347,2565,14,"notable","infrastructure","cloud gpu deployment","other",[24,25,26,27,28],"tensorflow","aws","gpu","cuda","machine learning","roasted",-0.375,{"positive":32,"neutral":33,"mixed":33,"critical":34},0,2,5,9,{"execution":37,"performance":38,"docs":40},{"positive":32,"critical":33},{"positive":32,"critical":39},1,{"positive":32,"critical":33},[39,32,42,32,42,39,39,42,39,42,32,39],3,null,{"problem":45,"audience":43,"pricing":43,"motivation":43,"time_to_build":43,"stack":55,"openness":75,"stage":76,"team_size":77},{"value":46,"source":47,"evidence":48},"Running TensorFlow on AWS GPU instances","inferred",[49,51],{"id":50},"post",{"id":52,"by":53,"text":54},"10556189","tacos","Top link points to a non-existent AMI so this \"recipe\" -- like almost every \"here's how to ...\" recipe on GitHub that gets posted on HN -- doesn't work. I swear, it's like 100%…",[56,60,68],{"value":5,"source":57,"evidence":58},"maker",[59],{"id":50},{"value":61,"source":62,"evidence":63},"CUDA","commenter",[64],{"id":65,"by":66,"text":67},"10556302","Smerity","Honestly, there are enough issues with TensorFlow right now due to CUDA 3.0 that using it with AWS is highly problematic. I appreciate the author's attempt, but there's no way…",{"value":69,"source":62,"evidence":70},"Bazel",[71],{"id":72,"by":73,"text":74},"10556893","erikbern","I think the nan issue you are referring to was caused by some weird stuff with Bazel. I put together an AMI in virginia: ami-cf5028a5 and if you have a masochistic streak, here…","open_source","beta","unknown",{"praise":79,"objections":89,"requests":118,"questions":123,"alternatives":131},[80],{"value":81,"source":62,"evidence":82},"Maker responsive to feedback and updated the README and build",[83,86],{"id":84,"by":11,"text":85},"10556488","Thanks for the feedback! I've added a note to the README that support is still experimental. I'll be tracking the issue and updating the repo + AMI as it develops. Will be…",{"id":87,"by":11,"text":88},"10577544","Updated to the latest commit, works for me! :)",[90,98,110,114],{"value":91,"source":62,"evidence":92},"CUDA 3.0 support causes NaNs and unreliable training",[93,94],{"id":65,"by":66,"text":67},{"id":95,"by":96,"text":97},"10556600","vrv","Thanks, would be good to have multiple sources of verification that HEAD now supports this natively without issues like unexpected NaNs.…",{"value":99,"source":62,"evidence":100},"Recipe failed: AMI was region-specific and the link was broken",[101,102,106],{"id":52,"by":53,"text":54},{"id":103,"by":104,"text":105},"10556270","droque","I tried with the Oregon region and it failed. I changed to the N. California region and it worked. (No comment about the GPU though.)",{"id":107,"by":108,"text":109},"10556385","fred256","An AMI id is region-specific, the original poster should probably have mentioned what region he created the AMI in.",{"value":111,"source":62,"evidence":112},"Multi-GPU g2.8xlarge not faster than g2.2xlarge",[113],{"id":72,"by":73,"text":74},{"value":115,"source":62,"evidence":116},"Google docs list GPU as unsupported",[117],{"id":52,"by":53,"text":54},[119],{"value":120,"source":62,"evidence":121},"Verification that HEAD supports CUDA without NaNs",[122],{"id":95,"by":96,"text":97},[124],{"value":125,"source":62,"evidence":126},"Why multi-GPU instances do not scale under Xen virtualization",[127],{"id":128,"by":129,"text":130},"10557123","varelse","Xen virtualization disables P2P copies ergo GPUs have what we call a \"failure to communicate and some GPUs you just can't reach (without going through the CPU that is).\"",[132],{"value":133,"source":62,"evidence":134},"Running TensorFlow locally on a laptop or phone",[135],{"id":65,"by":66,"text":67},{"comments":42,"answered":32,"key":137},[138,139,140],{"id":84,"text":85},{"id":87,"text":88},{"id":141,"text":142},"10556460","I've added a note that the AMI's region must be us-west-1. Thanks for the heads up!",{"commenters":35,"max_depth":144,"maker_replies":42,"first_comment_mins":145,"lifespan_hours":146},4,75,98,[148,150,152,154,156,158,160,161,163,166,168,170],[65,32,39,149,66,32],"project\u002Fcritical\u002Fexecution",[84,32,32,151,11,39],"maker\u002Fneutral\u002Fdocs",[95,32,42,153,96,32],"project\u002Fneutral\u002Fexecution",[87,32,32,155,11,39],"project\u002Fpositive\u002Fexecution",[72,32,42,157,73,32],"project\u002Fmixed\u002Fexecution",[128,32,39,159,129,32],"project\u002Fcritical\u002Fperformance",[52,39,39,149,53,32],[103,39,42,162,104,32],"project\u002Fmixed\u002Fdocs",[164,39,39,165,53,32],"10556372","project\u002Fcritical\u002Fdocs",[107,39,42,167,108,32],"project\u002Fneutral\u002Fdocs",[141,39,32,169,11,39],"maker\u002Fpositive\u002Fdocs",[171,39,39,165,172,32],"10557245","michaelZejoop",[],[175,180,185,190,194,198],{"id":176,"name":177,"points":178,"category":179},"10570666","Nixar",130,"dev_tools",{"id":181,"name":182,"points":183,"category":184},"10581735","Twitch Paints",109,"creative",{"id":186,"name":187,"points":188,"category":189},"10588371","Interviews can be stressful",96,"web_saas",{"id":191,"name":192,"points":193,"category":179},"10574045","react-yearly-calendar",94,{"id":195,"name":196,"points":197,"category":179},"10571065","ied",86,{"id":199,"name":200,"points":201,"category":179},"10569618","cljs4excel",83,false,"v2",{"id":205,"name":5,"line":6,"ai":202,"cat":20,"year":13,"points":14,"comments":15,"rec":29,"s":206,"top":18,"kw":207,"maker":11,"domain":10,"tier":19,"attempts":32,"strip":208,"url":9,"mix":209,"praise":81,"objection":91,"sub":21},10555692,-0.37,[24,25,26,27,28],"103031131301",[32,33,33,34],1794036537651]