[{"data":1,"prerenderedAt":377},["ShallowReactive",2],{"project-33985550":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":28,"sentiment":29,"stance_mix":30,"scored_comments":35,"aspects":36,"strip":46,"post_text":47,"what":48,"said":58,"maker_in_thread":189,"shape":197,"grid":202,"attempts":340,"same_week":341,"maker_shows":41,"maker_hidden":369,"reader_version":370,"card":371},"33985550","Juice Labs","Software for Remote GPU-over-IP","title","Show HN: Software for Remote GPU-over-IP","https:\u002F\u002Fgithub.com\u002FJuice-Labs\u002FJuice-Labs\u002Fwiki","github.com","stevegolik",1671034249,2022,138,56,436,2570,17,"notable","infrastructure","gpu virtualization","other",[24,25,26,27],"remote gpu","gpu over ip","gpu sharing","cuda remoting","mixed",-0.062,{"positive":31,"neutral":32,"mixed":33,"critical":34},7,16,2,12,37,{"performance":37,"docs":39,"idea":42,"execution":45},{"positive":33,"critical":38},5,{"positive":40,"critical":41},0,1,{"positive":43,"critical":44},3,6,{"positive":40,"critical":40},[41,41,41,40,43,41,40,41,43,43,41,40,40,40,40,43,40,43,40,43,40,38,43,38,38,43,41,40,41,43,38,43,43,41,43,43,40,43,38,43],"We built installable software for Windows & Linux that makes any remote Nvidia GPU accessible to, and shareable across, any number of remote clients running local applications, all over standard networking.",{"problem":49,"audience":55,"pricing":55,"motivation":55,"time_to_build":55,"stack":56,"openness":57,"stage":57,"team_size":57},{"value":50,"source":51,"evidence":52},"Lets remote clients use a remote Nvidia GPU over standard networking","maker",[53],{"id":54},"post",null,[],"unknown",{"praise":59,"objections":72,"requests":114,"questions":132,"alternatives":163},[60],{"value":61,"source":62,"evidence":63},"Impressive idea, similar to bitfusion","commenter",[64,68],{"id":65,"by":66,"text":67},"33989704","zamadatix","That's really awesome. I'm not sure what I'd use it for but just being able to makes me want to find an excuse! What's impressive is this seems to have more capabilities than…",{"id":69,"by":70,"text":71},"33993238","lmeyerov","Neato, sounds like bitfusion in their early days! Definitely of interest to us, even w\u002F latency limits, both for ai dev & investigations and occasional full runs I do have to…",[73,88,103],{"value":74,"source":62,"evidence":75},"Not open source; only Dockerfiles and binaries visible",[76,80,84],{"id":77,"by":78,"text":79},"33995875","lopkeny12ko","I'm confused, where is the actual source code? This repo only has some Dockerfiles that, as far as I can tell, are pulling precompiled opaque binaries and have some convenience…",{"id":81,"by":82,"text":83},"33997097","pifm_guy","I don't think it's opensource. I assume it cost quite some $$$ to produce this because you kinda have to cut nvidias binary drivers in half, which is going to require quite a lot…",{"id":85,"by":86,"text":87},"33992314","sworley","wait is the code actually FOSS or is this just freeware. I only see docker files.",{"value":89,"source":62,"evidence":90},"Network latency likely too high for gaming or interactive use",[91,95,99],{"id":92,"by":93,"text":94},"33990419","allanrbo","It surprises me that this works well enough to be useful. I would have thought that network latency, being orders of magnitude higher than memory latency, would be a huge…",{"id":96,"by":97,"text":98},"33990891","cobertos","I'd be surprised if this works for anything latency sensitive over anything more than a LAN. Even just the time it takes speed of light between NY and LA (410^6m\u002F310^8m\u002Fs=1\u002F75s)…",{"id":100,"by":101,"text":102},"33994246","Avlin67","no point without rdma enabled gpus…",{"value":104,"source":62,"evidence":105},"Inference servers already cover the use case better",[106,110],{"id":107,"by":108,"text":109},"33993432","kkielhofner","For anything involving inference you’re much better off with one of the many inference model servers such as TensorFlow serving, Triton Inference Server, etc.",{"id":111,"by":112,"text":113},"33994333","en4bz","That's the biggest problem with this model. With inference it's better to just use a dedicated model server. For training it's better to deploy on a massive dedicated machine.…",[115,126],{"value":116,"source":62,"evidence":117},"Add benchmarks and demos to the GitHub repo",[118,122],{"id":119,"by":120,"text":121},"33996009","fire","I'm at the server limit and can't join additional servers without leaving something else. Can you add this stuff to the github repo?",{"id":123,"by":124,"text":125},"33993313","JayStavis","Serverless GPU is all of the rage in the past month - I'd love to see a review of this from someone who knows how to benchmark a GPU workload. In particular: - Autoscaling Stable…",{"value":127,"source":62,"evidence":128},"Support graphical applications on Linux servers",[129],{"id":130,"by":120,"text":131},"33996053","> *Basically, we aren't targeting support for graphical applications running on Linux because there is very little demand for this - but we cover everything else. You CAN run…",[133,143,152],{"value":134,"source":62,"evidence":135},"Does it support multi-server setups or only one machine?",[136,139],{"id":137,"by":112,"text":138},"33994301","I doubt this does multi-server. All the GPUs probably have to be on the same machine.",{"id":140,"by":141,"text":142},"33991715","xrd","I see lots of comments in various ML repositores about trouble running on multiple GPUs. This seems like a great way to run across multiple low VRAM GPUs instead of buying a huge…",{"value":144,"source":62,"evidence":145},"CUDA Driver API or Runtime API remoting?",[146,148],{"id":147,"by":112,"text":144},"33994683",{"id":149,"by":150,"text":151},"34005921","cwbaker400","Driver.",{"value":153,"source":62,"evidence":154},"Can it be used for video decode or VM GPU access?",[155,159],{"id":156,"by":157,"text":158},"34000004","ridgered4","Can this be used to accelerate video decode in a linux machine\u002Fvirtual machine? It sounds like it is not for graphics on linux but it unclear to me where decode falls.",{"id":160,"by":161,"text":162},"34026440","ApolloRising","Would this allow a VMware Workstation Linux VM use the GPU from a Windows Host with an Nvidia Video Card for ML usage?",[164,171,175,179],{"value":165,"source":62,"evidence":166},"Virtual-GL",[167],{"id":168,"by":169,"text":170},"33990438","fock","Didn't we have those things already? Virtual-GL and Co. say hi. Also for most real GPU applications, you need to get the data in and out. I don't think splitting compute across a…",{"value":172,"source":62,"evidence":173},"Nvidia Bitfusion",[174],{"id":69,"by":70,"text":71},{"value":176,"source":62,"evidence":177},"Triton Inference Server",[178],{"id":107,"by":108,"text":109},{"value":180,"source":62,"evidence":181},"GeForce Now and Shadow cloud gaming",[182,186],{"id":183,"by":184,"text":185},"33992002","johanvts","The datacenter is probably not thousands, but hundreds of kilometers away so there is room to deliver 60fps. I was surprised how well GeForce Now works.",{"id":187,"by":97,"text":188},"33998606","Yeah, I've tried with shadow.tech and you can feel the latency. There's enough throughput to get a quality video stream through but there's enough latency to feel annoyed. I only…",{"comments":33,"answered":41,"key":190},[191,194],{"id":192,"text":193},"34002972","We started it on GitHub because we're planning to open source part of our code in the next month or two - namely the control plane business logic to determine which clients…",{"id":195,"text":196},"33994040","For some use cases we're already at parity - YMMV.",{"commenters":198,"max_depth":199,"maker_replies":33,"first_comment_mins":200,"lifespan_hours":201},39,4,234,66,[203,205,207,209,211,213,216,218,219,222,225,226,229,230,231,232,236,238,239,242,245,249,252,256,258,260,263,267,270,274,278,280,282,285,289,291,292,294,296,298,299,300,301,305,307,310,312,314,318,322,326,328,330,335,336],[130,40,41,204,120,40],"project\u002Fcritical\u002Fperformance",[77,41,41,206,78,40],"project\u002Fcritical\u002Fdocs",[81,41,41,208,82,40],"project\u002Fcritical\u002Fidea",[210,41,41,208,78,40],"34001062",[192,41,40,212,11,41],"maker\u002Fneutral\u002Fidea",[214,41,43,212,215,40],"34004685","canadacow",[123,33,43,217,124,40],"project\u002Fneutral\u002Fperformance",[107,33,41,208,108,40],[220,33,40,221,124,40],"33993610","commenter\u002Fneutral\u002Fidea",[223,33,40,224,108,40],"33994413","other\u002Fneutral\u002Fnone",[111,33,41,208,112,40],[227,33,43,228,215,40],"33993912","maker\u002Fneutral\u002Fdocs",[119,33,43,228,120,40],[92,43,43,217,93,40],[96,43,41,204,97,40],[233,43,40,234,235,40],"33991710","commenter\u002Fneutral\u002Fperformance","denkmoon",[187,43,40,237,97,40],"other\u002Fneutral\u002Fperformance",[183,43,40,234,184,40],[240,43,40,224,241,40],"33993791","miovoid",[243,43,40,224,244,40],"33994139","nixpulvis",[246,43,40,247,248,40],"33995378","commenter\u002Fneutral\u002Fnone","stingraycharles",[250,43,43,217,251,40],"33991056","delijati",[253,43,40,254,255,40],"33991200","commenter\u002Fcritical\u002Fperformance","AnIdiotOnTheNet",[257,43,43,217,241,40],"33993800",[259,43,40,224,215,40],"33993974",[261,43,40,224,262,40],"33999088","bbarnett",[264,43,43,265,266,40],"33991249","project\u002Fmixed\u002Fperformance","capableweb",[268,43,40,237,269,40],"33991638","Melatonic",[271,43,40,272,273,40],"33999273","other\u002Fneutral\u002Fpricing","wolrah",[275,43,38,276,277,40],"33991833","project\u002Fpositive\u002Fperformance","bob1029",[279,43,43,217,241,40],"33993678",[281,43,38,276,215,40],"33993837",[283,43,43,217,284,40],"33995236","kevingadd",[286,43,38,287,288,40],"33993128","project\u002Fpositive\u002Fidea","taf2",[69,199,43,290,70,40],"project\u002Fmixed\u002Fidea",[168,38,41,208,169,40],[293,38,40,234,269,40],"33991868",[295,38,40,234,112,40],"33994264",[297,38,41,204,241,40],"33993908",[140,44,38,287,141,40],[137,44,41,204,112,40],[65,31,38,287,66,40],[302,303,43,217,304,40],"33992021",8,"nimitt",[195,303,40,306,11,41],"maker\u002Fpositive\u002Fperformance",[156,308,43,309,157,40],9,"project\u002Fneutral\u002Fnone",[85,311,41,208,86,40],10,[160,313,43,309,161,40],11,[315,34,43,316,317,40],"33990267","project\u002Fneutral\u002Fexecution","dezmou",[319,34,43,320,321,40],"33990732","project\u002Fneutral\u002Fdocs","neuronexmachina",[323,324,40,224,325,40],"33996140",13,"imhoguy",[147,327,43,316,112,40],14,[149,327,43,329,150,40],"maker\u002Fneutral\u002Fexecution",[331,332,38,333,334,40],"33989480",15,"maker\u002Fpositive\u002Fnone","Mo3",[100,32,41,204,101,40],[337,18,38,338,339,40],"33990397","project\u002Fpositive\u002Fnone","yangikan",[],[342,347,351,356,360,365],{"id":343,"name":344,"points":345,"category":346},"33923907","LearnGPT",352,"ai",{"id":348,"name":349,"points":350,"category":346},"33910863","Hello (sayhello.so)",318,{"id":352,"name":353,"points":354,"category":355},"33931971","Pg_CRDT",296,"data",{"id":357,"name":358,"points":359,"category":355},"33969908","XetHub",279,{"id":361,"name":362,"points":363,"category":364},"33922754","Pynecone",209,"dev_tools",{"id":366,"name":367,"points":368,"category":364},"33906520","Ezy",155,false,"v2",{"id":372,"name":5,"line":6,"ai":369,"cat":20,"year":13,"points":14,"comments":15,"rec":28,"s":373,"top":18,"kw":374,"maker":11,"domain":10,"tier":19,"attempts":40,"strip":375,"url":9,"mix":376,"praise":61,"objection":74,"sub":21},33985550,-0.06,[24,25,26,27],"1110310133100003030305355310135331330353",[31,32,33,34],1794036628469]