[{"data":1,"prerenderedAt":271},["ShallowReactive",2],{"project-43835948":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":34,"aspects":35,"strip":41,"post_text":44,"what":45,"said":94,"maker_in_thread":176,"shape":186,"grid":189,"attempts":234,"same_week":235,"maker_shows":32,"maker_hidden":263,"reader_version":264,"card":265},"43835948","Neurox","GPU Observability for AI Infra","title","Show HN: Neurox – GPU Observability for AI Infra","https:\u002F\u002Fgithub.com\u002Fneuroxhq\u002Fhelm-chart-neurox-control","github.com","leeab",1745949723,2025,25,22,1816,3837,48,"shown","infrastructure","gpu observability","web_app",[24,25,26,27],"gpu monitoring","kubernetes observability","gpu cost tracking","multi-cloud gpu","mixed",-0.273,{"positive":31,"neutral":32,"mixed":33,"critical":32},1,4,0,9,{"privacy":36,"docs":37,"design":38,"idea":39},{"positive":33,"critical":31},{"positive":31,"critical":33},{"positive":33,"critical":33},{"positive":33,"critical":40},2,[33,33,33,33,42,33,42,33,42,33,31,33,43,42,33,31,33,31,33,31,33,33],3,5,null,{"problem":46,"audience":53,"pricing":62,"motivation":71,"time_to_build":44,"stack":75,"openness":92,"stage":93,"team_size":92},{"value":47,"source":48,"evidence":49},"Lack of visibility into GPU utilization, workload ownership and cost across multi-cloud Kubernetes clusters","maker",[50],{"id":51,"by":11,"text":52},"43835949","GPU observability is broken, so we built Neurox. When I co-founded Mezmo (a Series D observability platform), we obsessed over logs, metrics, and traces. I learned firsthand how…",{"value":54,"source":55,"evidence":56},"Teams running GPU workloads on Kubernetes across multiple clouds or on-prem","inferred",[57,58],{"id":51,"by":11,"text":52},{"id":59,"by":60,"text":61},"43837026","nicoslepicos","I've heard a few folks at events mention curiosity about stuff like this. Given you decided to start self-hosted, are you planning on a cloud version in the next while too? I'm…",{"value":63,"source":48,"evidence":64},"Free self-hosted tier up to 64 GPUs",[65,68],{"id":66,"by":11,"text":67},"43837084","We have considered this and may go down this route in the future. One thing we asked ourselves was what open sourcing provides. Usually it's a desire for privacy or cost in the…",{"id":69,"by":11,"text":70},"43837722","Someone asked about this earlier. We have considered this and may go down this route in the future. Was there something specific that you were looking for with open source? (ie:…",{"value":72,"source":48,"evidence":73},"GPU monitoring across multi-cloud environments is insufficient; metrics alone do not explain cost or cause",[74],{"id":51,"by":11,"text":52},[76,82,86],{"value":77,"source":48,"evidence":78},"Prometheus",[79],{"id":80,"by":11,"text":81},"43836334","Haha...there is some truth to that. We do use Prometheus under the hood to collect metrics. However, our thesis is that metrics alone isn’t enough. We marry metrics + kube state…",{"value":83,"source":48,"evidence":84},"Kubernetes",[85],{"id":51,"by":11,"text":52},{"value":87,"source":48,"evidence":88},"NVIDIA SMI",[89],{"id":90,"by":11,"text":91},"43838039","We collect a handful of metrics, but coming from our previous lives in DevOps, we only collect just what's needed to avoid unnecessary metrics bloat. The main 3 are: - GPU…","unknown","launched",{"praise":95,"objections":105,"requests":135,"questions":145,"alternatives":160},[96],{"value":97,"source":98,"evidence":99},"Interest in GPU observability tooling from people in the field","commenter",[100,101],{"id":59,"by":60,"text":61},{"id":102,"by":103,"text":104},"43836824","freeatnet","Interesting! A friend recently asked me if I knew of any tools to improve GPU observability across their deployments (primarily for cost tracking purposes, I think), but he was…",[106,114,121,128],{"value":107,"source":98,"evidence":108},"Not open source",[109,113],{"id":110,"by":111,"text":112},"43837654","zekrioca","* Not open-source.",{"id":102,"by":103,"text":104},{"value":115,"source":98,"evidence":116},"Requires persistent storage on GPU cluster, a blocker for bare metal",[117],{"id":118,"by":119,"text":120},"43837388","fustercluck","Your Github repo says you need 120 GB of persistent storage, but our bare metal GPU clusters only have local storage. Would like to try your thing, but hosting the data with the…",{"value":122,"source":98,"evidence":123},"Looks like Grafana with extra steps",[124],{"id":125,"by":126,"text":127},"43836236","firgrove","this feels like grafana with extra steps",{"value":129,"source":98,"evidence":130},"Name is similar to AWS Neuron\u002FNeuronX",[131],{"id":132,"by":133,"text":134},"43840182","nickysielicki","Are you concerned at all that your name is one letter away from neuronx? https:\u002F\u002Fawsdocs-neuron.readthedocs-hosted.com\u002Fen\u002Flatest\u002Fneuron-runtime\u002Findex.html",[136,141],{"value":137,"source":98,"evidence":138},"Open source the project",[139,140],{"id":102,"by":103,"text":104},{"id":110,"by":111,"text":112},{"value":142,"source":98,"evidence":143},"Cloud-hosted version",[144],{"id":59,"by":60,"text":61},[146,153],{"value":147,"source":98,"evidence":148},"What metrics and Kubernetes data are collected and how customizable are dashboards by role",[149],{"id":150,"by":151,"text":152},"43837826","badmonster","What metrics and Kubernetes runtime data does Neurox collect to provide its AI workload monitoring dashboards, and how customizable are these dashboards for different user roles…",{"value":154,"source":98,"evidence":155},"Is Kubernetes a good abstraction for GPU compute given shortages",[156],{"id":157,"by":158,"text":159},"43839268","mountainriver","Kubernetes is now kinda a bad abstraction for accelerated compute with the GPU shortages",[161,166],{"value":162,"source":98,"evidence":163},"Prometheus and Grafana",[164,165],{"id":125,"by":126,"text":127},{"id":51,"by":11,"text":52},{"value":167,"source":98,"evidence":168},"Slurm",[169,172],{"id":170,"by":133,"text":171},"43840179","> Everyone we talked to runs their compute in multi-cloud and uses Kubes as the unifier across all environments. I categorically support any company willing to take a strong…",{"id":173,"by":174,"text":175},"43840560","dharmab","Is your comment pro-SLURM or anti-SLURM? I took a serious look at SLURM for my problem space and among my conclusions were: - Hiring people who know Kubernetes is going to be far…",{"comments":177,"answered":43,"key":178},10,[179,180,183],{"id":51,"text":52},{"id":181,"text":182},"43841838","We're not for nor against Slurm. I do believe it has use cases in HPC, scientific and academic settings. We think our web UI is a bit easier to use and we do offer a competing…",{"id":184,"text":185},"43841873","I've heard of AWS Neuron...didn't realize they had a package called NeuronX. Tbh, I feel like many AI companies have similar names. Maybe the guys who own neuronx.io might be…",{"commenters":187,"max_depth":43,"maker_replies":177,"first_comment_mins":33,"lifespan_hours":188},11,12,[190,192,194,195,197,199,201,202,205,206,207,209,212,215,216,217,219,222,225,226,228,230],[51,33,33,191,11,31],"project\u002Fpositive\u002Fidea",[170,33,33,193,133,33],"other\u002Fneutral\u002Fnone",[173,33,33,193,174,33],[181,33,33,196,11,31],"project\u002Fneutral\u002Fdesign",[132,31,42,198,133,33],"project\u002Fneutral\u002Fnone",[184,31,33,200,11,31],"maker\u002Fneutral\u002Fnone",[59,40,42,198,60,33],[203,40,33,204,11,31],"43837235","project\u002Fneutral\u002Fprivacy",[102,42,42,198,103,33],[66,42,33,204,11,31],[118,42,31,208,119,33],"project\u002Fcritical\u002Fprivacy",[210,42,33,211,11,31],"43837500","project\u002Fneutral\u002Fdocs",[213,42,43,214,119,33],"43837850","project\u002Fpositive\u002Fdocs",[150,32,42,196,151,33],[90,32,33,196,11,31],[157,43,31,218,158,33],"project\u002Fcritical\u002Fidea",[220,43,33,221,11,31],"43839481","maker\u002Fneutral\u002Fidea",[110,223,31,224,111,33],6,"project\u002Fcritical\u002Fnone",[69,223,33,204,11,31],[125,227,31,218,126,33],7,[80,227,33,229,11,31],"project\u002Fneutral\u002Fidea",[231,232,33,193,233,33],"43837798",8,"28374654",[],[236,241,245,250,255,259],{"id":237,"name":238,"points":239,"category":240},"43820228","PyXL",983,"hardware",{"id":242,"name":243,"points":244,"category":20},"43803148","RetrOS-32",575,{"id":246,"name":247,"points":248,"category":249},"43816419","screenrecorder.me",462,"dev_tools",{"id":251,"name":252,"points":253,"category":254},"43835584","Beatsync",439,"creative",{"id":256,"name":257,"points":258,"category":20},"43778081","TacOS",320,{"id":260,"name":261,"points":262,"category":254},"43791992","Clever Coloring Book",316,false,"v2",{"id":266,"name":5,"line":6,"ai":263,"cat":20,"year":13,"points":14,"comments":15,"rec":28,"s":267,"top":18,"kw":268,"maker":11,"domain":10,"tier":19,"attempts":33,"strip":269,"url":9,"mix":270,"praise":97,"objection":107,"sub":21},43835948,-0.27,[24,25,26,27],"0000303030105301010100",[31,32,33,32],1794036673899]