Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

YOU DON'T NEED A SERVER RACK. SIX USED MINI PCS AND A $60 CISCO SWITCH WIRE INTO A HOME CLUSTER ON YOUR DESK. that clip is six tiny mini-pcs stacked and patched into one cisco business 110 switch, red cables, a single power strip. each box is a full...

RetroChainer's profile picture

RetroChainer

6,117 subscribers

95,267 Aufrufe • vor 2 Monaten •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

10 Kommentare

Profilbild von Chris Thompson
Chris Thompsonvor 2 Monaten

I have done this with mini elitedesks. I recommend building a little rack for them because they get quite hot when stacked

Profilbild von ViceSol
ViceSolvor 2 Monaten

This is one of the coolest things about homelabs. You don't need enterprise hardware to learn enterprise concepts - just a few inexpensive machines and curiosity

Profilbild von tech is cool
tech is coolvor 2 Monaten

Yes! My old surface book pro is running as an appliance for Ubuntu, synapse OS + Application, and hosting Ollama. Is a local home AI studio

Profilbild von Dilara Guller
Dilara Gullervor 2 Monaten

😀😀😀

Profilbild von darknessblade
darknessbladevor 2 Monaten

One thing you want to do is, open them up, and take out the WIFI card/antenna's. That way you have less hardware faults to deal with. And set it up so it can boot by LAN.

Profilbild von Nekt0
Nekt0vor 2 Monaten

pull is better than persuasion

Profilbild von Ashar Rai Mujeeb
Ashar Rai Mujeebvor 2 Monaten

Fun fact: this is how Steven Spielberg did the CGI for titanic.

Profilbild von Loweffortusername
Loweffortusernamevor 2 Monaten

I actually got a used server for pretty cheap, they're not always as bad as you'd think, but I do like this idea for a project!

Profilbild von Elon Orbit
Elon Orbitvor 2 Monaten

I love the setup. Real SPOF is that power strip though. Redundant nodes mean squat when the strip dies. Still beats cloud rent.

Profilbild von Akshay Ravirala | Data & AI Systems
Akshay Ravirala | Data & AI Systemsvor 2 Monaten

The $60 switch is never the bottleneck on a cluster like this — it's the power draw at 24/7 uptime that changes the math. Still a great way to hit k8s failure modes (node death, split-brain, noisy-neighbor) you'd never trigger on purpose in prod.

Ähnliche Videos

THIS WHOLE 3-NODE CLUSTER RUNS PROXMOX, CEPH AND KUBERNETES - ON 10 WATTS TOTAL. that clip is three zima boards wired to one little switch, and the power meter reads 10 watts for the entire thing. on those three tiny fanless boards: proxmox running a real cluster, three nodes as one datacenter ceph spreading storage across all three, so a node can die and your data doesn't kubernetes on top, kube1/2/3, plus docker and high-availability failover this is the exact stack that runs in datacenters, shrunk onto a shelf. kill one node and the vms migrate to the other two. it heals itself. and it pulls less power than a light bulb. no server rack, no cooling, no power bill you'd ever notice. the uncomfortable part isn't that it's small. it's that "enterprise infrastructure" was mostly the rack and the noise, and none of that was the actual capability. follow RetroChainer for more ways enterprise gear shrinks to a shelf, and bookmark this before you rent another vps.
0:28

THIS WHOLE 3-NODE CLUSTER RUNS PROXMOX, CEPH AND KUBERNETES - ON 10 WATTS TOTAL. that clip is three zima boards wired to one little switch, and the power meter reads 10 watts for the entire thing. on those three tiny fanless boards: proxmox running a real cluster, three nodes as one datacenter ceph spreading storage across all three, so a node can die and your data doesn't kubernetes on top, kube1/2/3, plus docker and high-availability failover this is the exact stack that runs in datacenters, shrunk onto a shelf. kill one node and the vms migrate to the other two. it heals itself. and it pulls less power than a light bulb. no server rack, no cooling, no power bill you'd ever notice. the uncomfortable part isn't that it's small. it's that "enterprise infrastructure" was mostly the rack and the noise, and none of that was the actual capability. follow RetroChainer for more ways enterprise gear shrinks to a shelf, and bookmark this before you rent another vps.

RetroChainer

42,348 Aufrufe • vor 2 Monaten

100 MAC MINIS ON ONE METAL SHELF. IN THE CLOUD, EACH OF THOSE RENTS FOR $120 TO $500 A MONTH. that clip is a homemade rack: wire shelving packed with mac minis, each running macos, wired into one cluster. why macs and not a normal server? two reasons. you can only build, sign, and notarize ios and mac apps on apple hardware. cloud providers know it and charge for it. and apple silicon sips power and runs near-silent, so you can stack a lot of them on a shelf in a room, not a datacenter. the rent they're escaping: a single mac in the cloud runs ~$120 to $500 a month (macstadium up to aws) one engineer documented saving $4,000+ a month by self-hosting mac minis instead of cloud ci runners a used or new mini is a few hundred dollars once, then a few dollars of electricity what a shelf like this actually runs: ios and mac build farms, render jobs, app and device testing automation, and increasingly local ai inference. own the boxes, rent nothing. the uncomfortable part: the cloud was never the only option. it was the convenient one. the people who did the math bought the shelf. the honest caveat: it's real capex up front, real heat and power, and you become the sysadmin. no rented rack means no one to call at 3am. worth it at scale, overkill for one build a week. no rack rental, no per-hour metering, no fleet you don't own. save this before your cloud invoice renews again.
0:25

100 MAC MINIS ON ONE METAL SHELF. IN THE CLOUD, EACH OF THOSE RENTS FOR $120 TO $500 A MONTH. that clip is a homemade rack: wire shelving packed with mac minis, each running macos, wired into one cluster. why macs and not a normal server? two reasons. you can only build, sign, and notarize ios and mac apps on apple hardware. cloud providers know it and charge for it. and apple silicon sips power and runs near-silent, so you can stack a lot of them on a shelf in a room, not a datacenter. the rent they're escaping: a single mac in the cloud runs ~$120 to $500 a month (macstadium up to aws) one engineer documented saving $4,000+ a month by self-hosting mac minis instead of cloud ci runners a used or new mini is a few hundred dollars once, then a few dollars of electricity what a shelf like this actually runs: ios and mac build farms, render jobs, app and device testing automation, and increasingly local ai inference. own the boxes, rent nothing. the uncomfortable part: the cloud was never the only option. it was the convenient one. the people who did the math bought the shelf. the honest caveat: it's real capex up front, real heat and power, and you become the sysadmin. no rented rack means no one to call at 3am. worth it at scale, overkill for one build a week. no rack rental, no per-hour metering, no fleet you don't own. save this before your cloud invoice renews again.

RetroChainer

21,055 Aufrufe • vor 2 Monaten

YOUR HOMELAB IS USELESS IF YOU CAN'T REACH IT FROM ANYWHERE - AND YOU SHOULD NEVER OPEN A PORT TO DO IT. that clip is a homelab in a tiny rack: cheap used mini-pcs running a proxmox cluster with ollama for local ai. but the hardware isn't the hard part. reaching it safely when you're not home is. the wrong way: forward a port on your router. now your services face the whole internet, and every bot on earth starts knocking. the right way, and what he's running: a mesh vpn like twingate, netbird or tailscale. it builds a private encrypted network between your devices and the lab. no open ports, no public ip, nothing exposed. your phone in a cafe talks to your proxmox box at home like they're on the same wifi, and the internet sees nothing. that's the piece that turns a pile of mini-pcs into infrastructure you actually use. no port forwarding, no exposed dashboard, no public attack surface. the uncomfortable part isn't building the lab. it's that most people either can't reach it, or reach it by leaving the front door wide open. follow RetroChainer for the homelab step everyone skips, and bookmark this before you forward a single port.
0:59

YOUR HOMELAB IS USELESS IF YOU CAN'T REACH IT FROM ANYWHERE - AND YOU SHOULD NEVER OPEN A PORT TO DO IT. that clip is a homelab in a tiny rack: cheap used mini-pcs running a proxmox cluster with ollama for local ai. but the hardware isn't the hard part. reaching it safely when you're not home is. the wrong way: forward a port on your router. now your services face the whole internet, and every bot on earth starts knocking. the right way, and what he's running: a mesh vpn like twingate, netbird or tailscale. it builds a private encrypted network between your devices and the lab. no open ports, no public ip, nothing exposed. your phone in a cafe talks to your proxmox box at home like they're on the same wifi, and the internet sees nothing. that's the piece that turns a pile of mini-pcs into infrastructure you actually use. no port forwarding, no exposed dashboard, no public attack surface. the uncomfortable part isn't building the lab. it's that most people either can't reach it, or reach it by leaving the front door wide open. follow RetroChainer for the homelab step everyone skips, and bookmark this before you forward a single port.

RetroChainer

58,500 Aufrufe • vor 2 Monaten

THIS IS WHERE "THE CLOUD" ACTUALLY LIVES - A ROOM OF DELL STORAGE ARRAYS YOU RENT A SLIVER OF EVERY MONTH. that clip is a real datacenter aisle, rack after rack of dell powerstore arrays, every hexagon panel packed with drives. this is "the cloud." not a metaphor, this room, plus the power, cooling and staff keeping it alive. when you pay for cloud storage, you rent a few gigabytes of these arrays. forever. the bill grows with every photo. here's the part they'd rather you not connect. these enterprise drives run on a schedule. data centers swap them out long before they fail, by the pallet. those decommissioned drives get tested and resold as recertified, the same 20tb enterprise disk for a fraction of new. so the exact hardware humming in this room ends up in a nas on your desk, minus the monthly rent. no cloud subscription, no data on someone else's array, no bill that grows with your library. the uncomfortable part isn't that the cloud is huge. it's that "the cloud" was always just this room, and you can buy a piece of it once instead of renting it forever. follow RetroChainer for what the cloud doesn't want you to picture, and bookmark this before your next storage renewal.
1:10

THIS IS WHERE "THE CLOUD" ACTUALLY LIVES - A ROOM OF DELL STORAGE ARRAYS YOU RENT A SLIVER OF EVERY MONTH. that clip is a real datacenter aisle, rack after rack of dell powerstore arrays, every hexagon panel packed with drives. this is "the cloud." not a metaphor, this room, plus the power, cooling and staff keeping it alive. when you pay for cloud storage, you rent a few gigabytes of these arrays. forever. the bill grows with every photo. here's the part they'd rather you not connect. these enterprise drives run on a schedule. data centers swap them out long before they fail, by the pallet. those decommissioned drives get tested and resold as recertified, the same 20tb enterprise disk for a fraction of new. so the exact hardware humming in this room ends up in a nas on your desk, minus the monthly rent. no cloud subscription, no data on someone else's array, no bill that grows with your library. the uncomfortable part isn't that the cloud is huge. it's that "the cloud" was always just this room, and you can buy a piece of it once instead of renting it forever. follow RetroChainer for what the cloud doesn't want you to picture, and bookmark this before your next storage renewal.

RetroChainer

34,236 Aufrufe • vor 2 Monaten

A USED $700 RTX 3090 HAS MORE MEMORY AND FAR MORE SPEED THAN THE ~$1,000 A2 GOING INTO THIS SERVER. THE A2 STILL WINS - FOR ONE REASON THAT ISN'T PERFORMANCE. that clip is a dell poweredge r760 - a 2u rack server - getting an nvidia a2 dropped in through a riser. this is the enterprise route to local ai. the rung the desktop ladder skips entirely. the a2, verified: 16gb of memory, single-slot, low-profile, and just 40 to 60 watts it's an inference card, built to sit in a server that has no room or spare power for a real gpu so why not a 3090? a used rtx 3090 gives you 24gb and many times the compute for around $700. the a2 gives 16gb, much slower, for roughly a thousand. on raw local-llm value, the desktop card wins, and it isn't close. the a2's whole reason to exist is the thing you can't see on a spec sheet: it fits. a 2u server has no spare gpu power and no room for a triple-slot furnace. the a2 slips into one low-profile slot on 60 watts and lets an existing server do inference without a rebuild. the honest caveats: the r760 is not a desk machine. loaded, it idles at hundreds of watts, and its fans sound like a hair dryer that never turns off. it belongs in a rack or a closet 16gb is still 16gb. the same 7b-to-32b ceiling as the cheap cards. the enterprise badge doesn't buy you a bigger model what it does buy: ecc memory, redundant power, remote management, hot-swap everything. reliability, not speed so who this is for: someone who already runs a rack and wants to add private inference without touching the power or cooling budget. for anyone starting from a desk, the mac mini or the 3090 wins on every axis that matters. the real point: the "best" local ai box depends entirely on what you already own. on a desk, efficiency wins. in a rack, fit and reliability win. the a2 isn't a bad card - it's a card for a constraint the ladder never mentions. no fastest gpu here, no desk-friendly box, no bigger model than the cheap cards already run. save this before you buy enterprise gear for a desktop job.
0:28

A USED $700 RTX 3090 HAS MORE MEMORY AND FAR MORE SPEED THAN THE ~$1,000 A2 GOING INTO THIS SERVER. THE A2 STILL WINS - FOR ONE REASON THAT ISN'T PERFORMANCE. that clip is a dell poweredge r760 - a 2u rack server - getting an nvidia a2 dropped in through a riser. this is the enterprise route to local ai. the rung the desktop ladder skips entirely. the a2, verified: 16gb of memory, single-slot, low-profile, and just 40 to 60 watts it's an inference card, built to sit in a server that has no room or spare power for a real gpu so why not a 3090? a used rtx 3090 gives you 24gb and many times the compute for around $700. the a2 gives 16gb, much slower, for roughly a thousand. on raw local-llm value, the desktop card wins, and it isn't close. the a2's whole reason to exist is the thing you can't see on a spec sheet: it fits. a 2u server has no spare gpu power and no room for a triple-slot furnace. the a2 slips into one low-profile slot on 60 watts and lets an existing server do inference without a rebuild. the honest caveats: the r760 is not a desk machine. loaded, it idles at hundreds of watts, and its fans sound like a hair dryer that never turns off. it belongs in a rack or a closet 16gb is still 16gb. the same 7b-to-32b ceiling as the cheap cards. the enterprise badge doesn't buy you a bigger model what it does buy: ecc memory, redundant power, remote management, hot-swap everything. reliability, not speed so who this is for: someone who already runs a rack and wants to add private inference without touching the power or cooling budget. for anyone starting from a desk, the mac mini or the 3090 wins on every axis that matters. the real point: the "best" local ai box depends entirely on what you already own. on a desk, efficiency wins. in a rack, fit and reliability win. the a2 isn't a bad card - it's a card for a constraint the ladder never mentions. no fastest gpu here, no desk-friendly box, no bigger model than the cheap cards already run. save this before you buy enterprise gear for a desktop job.

Grimmer

24,287 Aufrufe • vor 2 Monaten

EVERYONE'S BUYING THE M4 MAC MINI AS A CHEAP DESKTOP. HERE'S WHY IT ENDS UP UNUSED. the pitch is irresistible: a tiny mac that runs cool, quiet, and cheap. then the receipt shows up. the mac mini ships with no screen, no keyboard, no mouse. add a monitor, a decent keyboard and mouse, and more storage, and the "$799 mac" is past $1,000 fast. and macos isn't windows. if your world is gaming, specific windows apps, or the setup you already know, the mini fights you the whole way. that's how so many end up back on a pc, the mini in a drawer. so what is the mac mini actually great at? not being your main desktop. it's the box you never look at. headless in a corner, it's a near-silent always-on server: run local ai models on it, host your automations, keep your files, and reach it from the laptop you already use. it earns its keep by disappearing. the honest turn: the mac mini isn't a cheap desktop. it's a cheap server pretending to be one. buy it for the job it's quietly brilliant at, not the one the ads imply. want a main computer? price the whole setup first. want a silent little box that runs ai and never bothers you? this is the one. no monitor in the box, no keyboard, no "just $599". save this before you buy the sticker price instead of the real one.
0:34

EVERYONE'S BUYING THE M4 MAC MINI AS A CHEAP DESKTOP. HERE'S WHY IT ENDS UP UNUSED. the pitch is irresistible: a tiny mac that runs cool, quiet, and cheap. then the receipt shows up. the mac mini ships with no screen, no keyboard, no mouse. add a monitor, a decent keyboard and mouse, and more storage, and the "$799 mac" is past $1,000 fast. and macos isn't windows. if your world is gaming, specific windows apps, or the setup you already know, the mini fights you the whole way. that's how so many end up back on a pc, the mini in a drawer. so what is the mac mini actually great at? not being your main desktop. it's the box you never look at. headless in a corner, it's a near-silent always-on server: run local ai models on it, host your automations, keep your files, and reach it from the laptop you already use. it earns its keep by disappearing. the honest turn: the mac mini isn't a cheap desktop. it's a cheap server pretending to be one. buy it for the job it's quietly brilliant at, not the one the ads imply. want a main computer? price the whole setup first. want a silent little box that runs ai and never bothers you? this is the one. no monitor in the box, no keyboard, no "just $599". save this before you buy the sticker price instead of the real one.

RetroChainer

14,184 Aufrufe • vor 2 Monaten

AMD JUST KILLED THE $200 - A - MONTH AI SUBSCRIPTION - A 65-WATT BOX WITH 128GB OF UNIFIED MEMORY THAT RUNS 70B MODELS LOCALLY. that clip is amd's ryzen ai max+ 395 mini box, held up on stage. the whole pitch is one spec: 128gb of high-speed unified memory. why it matters: cpu, gpu and npu share the same 128gb pool, so a model that used to need a rack of vram just fits. what it actually runs: llama 70b in full precision on the built-in gpu open models up to ~100b parameters, locally, no cloud all of it in a mini pc pulling around 65 watts that's the part that ends the subscription: your prompts, your data, your model, on a box on your desk, not metered by the token. honest line: this runs open models, not gpt-5 or claude opus. for chat, coding help and agents on your own data it's more than enough. for the absolute frontier, the cloud still wins. and it isn't free: the box runs ~$1,500 and up. but that's a few months of a $200 sub, then it's yours forever. no monthly ai bill, no tokens metered, no prompt leaving the room. the uncomfortable part isn't that amd made a fast box. it's that "you need the cloud for ai" was a pricing choice, and unified memory just made it optional. follow RetroChainer for the hardware that turns rent into ownership, and bookmark this before your next ai renewal.
0:41

AMD JUST KILLED THE $200 - A - MONTH AI SUBSCRIPTION - A 65-WATT BOX WITH 128GB OF UNIFIED MEMORY THAT RUNS 70B MODELS LOCALLY. that clip is amd's ryzen ai max+ 395 mini box, held up on stage. the whole pitch is one spec: 128gb of high-speed unified memory. why it matters: cpu, gpu and npu share the same 128gb pool, so a model that used to need a rack of vram just fits. what it actually runs: llama 70b in full precision on the built-in gpu open models up to ~100b parameters, locally, no cloud all of it in a mini pc pulling around 65 watts that's the part that ends the subscription: your prompts, your data, your model, on a box on your desk, not metered by the token. honest line: this runs open models, not gpt-5 or claude opus. for chat, coding help and agents on your own data it's more than enough. for the absolute frontier, the cloud still wins. and it isn't free: the box runs ~$1,500 and up. but that's a few months of a $200 sub, then it's yours forever. no monthly ai bill, no tokens metered, no prompt leaving the room. the uncomfortable part isn't that amd made a fast box. it's that "you need the cloud for ai" was a pricing choice, and unified memory just made it optional. follow RetroChainer for the hardware that turns rent into ownership, and bookmark this before your next ai renewal.

RetroChainer

16,995 Aufrufe • vor 1 Monat

BOLT A GPU ONTO THIS $200 FANLESS BOARD AND IT DOES THE TWO THINGS A RASPBERRY PI PHYSICALLY CAN'T: LOCAL AI AND 4K TRANSCODING. that clip is a zimaboard with a graphics card plugged straight into its pcie slot. that slot is the whole point. a pi has gpio and usb, but no real pcie, so it tops out where this one starts. drop a gpu in and the tiny box suddenly does: ollama and local models with actual gpu acceleration, not cpu crawl plex or jellyfin transcoding several 4k streams at once, in hardware anything else that was "you need a real pc for that", on a fanless board the rest of the board backs it up: quad-core intel up to 3.6, dual sata, dual 2.5gbe, runs zimaos, linux, even windows. so one silent box becomes a nas, a media server, and a local ai node at the same time. no tower, no raspberry-pi ceiling, no cloud gpu rented by the hour. the uncomfortable part isn't that it takes a gpu. it's that "a pi can do it" quietly stopped being true the moment the workload needed a real pcie lane. follow RetroChainer for the small boxes that punch way above their size, and bookmark this before you buy a whole tower for one gpu job.
0:21

BOLT A GPU ONTO THIS $200 FANLESS BOARD AND IT DOES THE TWO THINGS A RASPBERRY PI PHYSICALLY CAN'T: LOCAL AI AND 4K TRANSCODING. that clip is a zimaboard with a graphics card plugged straight into its pcie slot. that slot is the whole point. a pi has gpio and usb, but no real pcie, so it tops out where this one starts. drop a gpu in and the tiny box suddenly does: ollama and local models with actual gpu acceleration, not cpu crawl plex or jellyfin transcoding several 4k streams at once, in hardware anything else that was "you need a real pc for that", on a fanless board the rest of the board backs it up: quad-core intel up to 3.6, dual sata, dual 2.5gbe, runs zimaos, linux, even windows. so one silent box becomes a nas, a media server, and a local ai node at the same time. no tower, no raspberry-pi ceiling, no cloud gpu rented by the hour. the uncomfortable part isn't that it takes a gpu. it's that "a pi can do it" quietly stopped being true the moment the workload needed a real pcie lane. follow RetroChainer for the small boxes that punch way above their size, and bookmark this before you buy a whole tower for one gpu job.

RetroChainer

24,813 Aufrufe • vor 1 Monat

FOUR BOXES ON A DESK ARE NOW A FRONTIER AI CLUSTER This is four NVIDIA DGX Spark units wired together, sitting on a single desk and running Qwen3-235B end to end. Each box holds 128GB of unified memory. Stacked together, that is half a terabyte of memory pooled across the cluster, all from hardware you could fit in a backpack. A year ago a setup like this lived in a data center behind a badge reader. Now it runs next to a monitor and a coffee cup. Here is the part people get wrong. The four boxes do not magically merge into one giant GPU. The model is sharded across all of them, and they pass data over a fast link while every box does its slice of the work. That is the whole trick. No single box could hold a 235B model, but split four ways, it runs. The machines stay quiet and barely warm while doing it, sipping a fraction of the power a rack of server GPUs would burn. The wall between you and a frontier-class model used to be a cloud bill. Now it is just desk space. Bookmark this & Watch it run ↓
1:42

FOUR BOXES ON A DESK ARE NOW A FRONTIER AI CLUSTER This is four NVIDIA DGX Spark units wired together, sitting on a single desk and running Qwen3-235B end to end. Each box holds 128GB of unified memory. Stacked together, that is half a terabyte of memory pooled across the cluster, all from hardware you could fit in a backpack. A year ago a setup like this lived in a data center behind a badge reader. Now it runs next to a monitor and a coffee cup. Here is the part people get wrong. The four boxes do not magically merge into one giant GPU. The model is sharded across all of them, and they pass data over a fast link while every box does its slice of the work. That is the whole trick. No single box could hold a 235B model, but split four ways, it runs. The machines stay quiet and barely warm while doing it, sipping a fraction of the power a rack of server GPUs would burn. The wall between you and a frontier-class model used to be a cloud bill. Now it is just desk space. Bookmark this & Watch it run ↓

slash1s

19,473 Aufrufe • vor 3 Monaten

FOUR DIFFERENT VENDORS ARE NOW SHIPPING GB10 MINI PCs WITH 128GB UNIFIED MEMORY, AND ONE MICROTIK CRS 804 SWITCH CAN CONNECT UP TO EIGHT OF THEM INTO A 1 TERABYTE LOCAL AI CLUSTER 00:00 he points at the MikroTik CRS 804, "you need some kind of switch that'll handle QSFP56 ports like these", the interconnect that makes the whole cluster possible the GB10 ecosystem is no longer just Nvidia. Dell Pro Max GB10, ASUS Ascent GX10, and MSI Edge Expert all ship the same Grace Blackwell Superchip with 128GB of coherent memory. same silicon, different cases, same 200 gigabit ports on the back the CRS 804 is what connects them at prosumer prices. four 400 gigabit QSFP56 ports on one 1U chassis, breakout cables that split each port into two 200 gigabit lanes. one switch drives eight GB10 units in parallel do the math. eight nodes at 128GB each equals 1024GB of pooled unified memory across the cluster. run vLLM, shard a frontier model across all eight, and inference happens locally on hardware that fits in half a rack the real limiter revealed in the stress test was never throttling. it was interconnect topology, exactly the layer this switch fixes at a fraction of enterprise switch pricing $400 a month for combined chatgpt pro and claude code max hits $4,800 a year per developer. a small team of five running through this cluster pays back inside eight months and never expires the article covers the buying ladder for a single desk. this post is proof of the cluster ladder that starts where the desk one ends save this before the GB10 lineup grows past four vendors and prosumer cluster switches move upmarket
1:05

FOUR DIFFERENT VENDORS ARE NOW SHIPPING GB10 MINI PCs WITH 128GB UNIFIED MEMORY, AND ONE MICROTIK CRS 804 SWITCH CAN CONNECT UP TO EIGHT OF THEM INTO A 1 TERABYTE LOCAL AI CLUSTER 00:00 he points at the MikroTik CRS 804, "you need some kind of switch that'll handle QSFP56 ports like these", the interconnect that makes the whole cluster possible the GB10 ecosystem is no longer just Nvidia. Dell Pro Max GB10, ASUS Ascent GX10, and MSI Edge Expert all ship the same Grace Blackwell Superchip with 128GB of coherent memory. same silicon, different cases, same 200 gigabit ports on the back the CRS 804 is what connects them at prosumer prices. four 400 gigabit QSFP56 ports on one 1U chassis, breakout cables that split each port into two 200 gigabit lanes. one switch drives eight GB10 units in parallel do the math. eight nodes at 128GB each equals 1024GB of pooled unified memory across the cluster. run vLLM, shard a frontier model across all eight, and inference happens locally on hardware that fits in half a rack the real limiter revealed in the stress test was never throttling. it was interconnect topology, exactly the layer this switch fixes at a fraction of enterprise switch pricing $400 a month for combined chatgpt pro and claude code max hits $4,800 a year per developer. a small team of five running through this cluster pays back inside eight months and never expires the article covers the buying ladder for a single desk. this post is proof of the cluster ladder that starts where the desk one ends save this before the GB10 lineup grows past four vendors and prosumer cluster switches move upmarket

NO1ennn

59,725 Aufrufe • vor 3 Monaten

6 dead office PCs at $50 each run a Kubernetes cluster that AWS would bill $10,000 a year for. Most home labs die before they boot. Someone specs a Xeon. Prices 128GB of RAM. Opens 40 tabs. Buys nothing. The build that actually runs is uglier than that. HP EliteDesk 800 G3 Mini. Corporate castoffs, dumped on eBay by the pallet every time an office refreshes. $50 a unit, sold stripped. 16GB of RAM and a 256GB NVMe put each one back to work. 35W chips. All 6 idle under 100W. k3s on top, not kubeadm. 1 binary, 512MB of RAM. node1 starts the cluster. node2 and node3 join as servers. That is 3 etcd members, so losing one kills nothing. 6 nodes. 17 namespaces. 173 pods. CPU idling at 4.6%. The same cluster on EKS: 6 t3.xlarge at $121 a month, plus $73 for the control plane. $9,636 a year. $300 once. And it did not start there. The first server was a retired gaming PC headed for a closet. The right first machine is the one already collecting dust. Stop planning the rack. Start having fun.
1:19

6 dead office PCs at $50 each run a Kubernetes cluster that AWS would bill $10,000 a year for. Most home labs die before they boot. Someone specs a Xeon. Prices 128GB of RAM. Opens 40 tabs. Buys nothing. The build that actually runs is uglier than that. HP EliteDesk 800 G3 Mini. Corporate castoffs, dumped on eBay by the pallet every time an office refreshes. $50 a unit, sold stripped. 16GB of RAM and a 256GB NVMe put each one back to work. 35W chips. All 6 idle under 100W. k3s on top, not kubeadm. 1 binary, 512MB of RAM. node1 starts the cluster. node2 and node3 join as servers. That is 3 etcd members, so losing one kills nothing. 6 nodes. 17 namespaces. 173 pods. CPU idling at 4.6%. The same cluster on EKS: 6 t3.xlarge at $121 a month, plus $73 for the control plane. $9,636 a year. $300 once. And it did not start there. The first server was a retired gaming PC headed for a closet. The right first machine is the one already collecting dust. Stop planning the rack. Start having fun.

AiMind

47,863 Aufrufe • vor 1 Monat

THE $599 MAC MINI GETS A $49 DOCK AND TURNS FROM A DESK COMPUTER INTO A TINY 24/7 AI WORKFLOW SERVER 00:19 he clips the dock onto the back of the mac mini, adds extra ports, ethernet, storage and turns one small silver box into a workflow hub that can stay online for 700+ hours a month a base mac mini can sit online all day, pulling light power while scripts watch folders, move files, process notes and keep boring workflows running in the background paired with claude, it becomes a small home automation box: 2-hour lectures get transcribed, 30-page articles get summarized, reports get drafted, and research gets cleaned before you even open the laptop the difference is the loop. trigger, process, verify, save, repeat. the same workflow that costs $40-100/month across cloud tools can run locally from one machine on your desk that is why this tiny upgrade matters. the dock is cheap, but it gives the mac mini enough expansion to act like a quiet server instead of another computer waiting for you to use it bookmark this before your desk setup realizes it became infrastructure.
0:29

THE $599 MAC MINI GETS A $49 DOCK AND TURNS FROM A DESK COMPUTER INTO A TINY 24/7 AI WORKFLOW SERVER 00:19 he clips the dock onto the back of the mac mini, adds extra ports, ethernet, storage and turns one small silver box into a workflow hub that can stay online for 700+ hours a month a base mac mini can sit online all day, pulling light power while scripts watch folders, move files, process notes and keep boring workflows running in the background paired with claude, it becomes a small home automation box: 2-hour lectures get transcribed, 30-page articles get summarized, reports get drafted, and research gets cleaned before you even open the laptop the difference is the loop. trigger, process, verify, save, repeat. the same workflow that costs $40-100/month across cloud tools can run locally from one machine on your desk that is why this tiny upgrade matters. the dock is cheap, but it gives the mac mini enough expansion to act like a quiet server instead of another computer waiting for you to use it bookmark this before your desk setup realizes it became infrastructure.

Gipp 🦅

73,919 Aufrufe • vor 3 Monaten

THIS $1,300 CARD HAS THE SAME 32GB OF VRAM AS A $2,000+ RTX 5090 - FOR LOCAL AI, THAT'S THE SPEC THAT DECIDES EVERYTHING. that clip is the gigabyte radeon ai pro r9700 ai top, 32gb of vram, built for local ai. the pitch is simple: the same memory capacity as an rtx 5090, for roughly a third of what a 5090 actually sells for. here's why that matters, and where it doesn't. for ai, vram is the gate. a model either fits in memory or it doesn't. 32gb loads models a 16gb card can't touch, full stop. so for running local llms and image models, this card does the one thing that counts, cheap. honest line: this is not a 5090 for gaming. the 5090 is far faster at raster and ray tracing. they only match on how much vram they carry. if you want frames, buy the 5090. if you want to fit big models locally, you just found the same 32gb for a third of the money. no $2,000+ price tag, no cloud gpu rental, no model too big for 16gb. the uncomfortable part isn't that amd is cheaper. it's that for local ai you were about to overpay for gaming power you'd never use. follow RetroChainer for more on paying for the spec that matters, and bookmark this before your next gpu.
1:52

THIS $1,300 CARD HAS THE SAME 32GB OF VRAM AS A $2,000+ RTX 5090 - FOR LOCAL AI, THAT'S THE SPEC THAT DECIDES EVERYTHING. that clip is the gigabyte radeon ai pro r9700 ai top, 32gb of vram, built for local ai. the pitch is simple: the same memory capacity as an rtx 5090, for roughly a third of what a 5090 actually sells for. here's why that matters, and where it doesn't. for ai, vram is the gate. a model either fits in memory or it doesn't. 32gb loads models a 16gb card can't touch, full stop. so for running local llms and image models, this card does the one thing that counts, cheap. honest line: this is not a 5090 for gaming. the 5090 is far faster at raster and ray tracing. they only match on how much vram they carry. if you want frames, buy the 5090. if you want to fit big models locally, you just found the same 32gb for a third of the money. no $2,000+ price tag, no cloud gpu rental, no model too big for 16gb. the uncomfortable part isn't that amd is cheaper. it's that for local ai you were about to overpay for gaming power you'd never use. follow RetroChainer for more on paying for the spec that matters, and bookmark this before your next gpu.

RetroChainer

38,844 Aufrufe • vor 2 Monaten

TWO BOXES THE SIZE OF A MAC MINI JUST RAN A 235 BILLION PARAMETER MODEL ON A DESK It is two NVIDIA DGX Spark units linked by a single cable. A year ago a model this size meant renting a GPU cluster by the hour. Now it sits next to your monitor for around $8,000. Here is the twist most people miss. Linking them does not create one shared 256GB memory pool. The model is split across both boxes, and that is the only reason a 235B model fits at all. It answers at roughly 10 tokens per second, and both chips sit at just 74 degrees while sipping around 50 watts. Every token stays on the desk. Nothing touches a cloud, and nothing leaves the room. The ceiling for what you can run at home just jumped from 70B to 235B. Bookmark this & Watch it run ↓
1:21

TWO BOXES THE SIZE OF A MAC MINI JUST RAN A 235 BILLION PARAMETER MODEL ON A DESK It is two NVIDIA DGX Spark units linked by a single cable. A year ago a model this size meant renting a GPU cluster by the hour. Now it sits next to your monitor for around $8,000. Here is the twist most people miss. Linking them does not create one shared 256GB memory pool. The model is split across both boxes, and that is the only reason a 235B model fits at all. It answers at roughly 10 tokens per second, and both chips sit at just 74 degrees while sipping around 50 watts. Every token stays on the desk. Nothing touches a cloud, and nothing leaves the room. The ceiling for what you can run at home just jumped from 70B to 235B. Bookmark this & Watch it run ↓

slash1s

101,317 Aufrufe • vor 3 Monaten

THE $4,000 DGX SPARK BENCHMARKS AT 342 T/S ON PROMPT PROCESSING - THEN DROPS TO 73 WHEN IT ACTUALLY GENERATES. that clip is a dgx spark running llama-bench on qwen3-coder 30b, side by side with a mac. two numbers decide everything here: pp512, prompt processing = 342 tokens/sec tg128, token generation = 73 tokens/sec that gap is the whole lesson. local ai isn't one speed, it's two. prompt processing is reading your input: your files, your long context. the spark is compute-heavy, so it eats that fast. token generation is writing the answer, and it's bound by memory bandwidth, where the spark is modest and unified-memory macs stay close. so "which box is faster" has no single winner. feeding huge context or running agents that read a lot? prompt processing wins, the spark shines. long chat replies, one token at a time? generation matters more, and a mac holds its own. no single benchmark number, no universal "fastest box", no upgrade that wins every workload. the uncomfortable part isn't that the spark is fast. it's that "tokens per second" was always two numbers, and the hype only ever quotes the flattering one. follow RetroChainer for the benchmark truth nobody screenshots, and bookmark this before you buy on one number.
1:14

THE $4,000 DGX SPARK BENCHMARKS AT 342 T/S ON PROMPT PROCESSING - THEN DROPS TO 73 WHEN IT ACTUALLY GENERATES. that clip is a dgx spark running llama-bench on qwen3-coder 30b, side by side with a mac. two numbers decide everything here: pp512, prompt processing = 342 tokens/sec tg128, token generation = 73 tokens/sec that gap is the whole lesson. local ai isn't one speed, it's two. prompt processing is reading your input: your files, your long context. the spark is compute-heavy, so it eats that fast. token generation is writing the answer, and it's bound by memory bandwidth, where the spark is modest and unified-memory macs stay close. so "which box is faster" has no single winner. feeding huge context or running agents that read a lot? prompt processing wins, the spark shines. long chat replies, one token at a time? generation matters more, and a mac holds its own. no single benchmark number, no universal "fastest box", no upgrade that wins every workload. the uncomfortable part isn't that the spark is fast. it's that "tokens per second" was always two numbers, and the hype only ever quotes the flattering one. follow RetroChainer for the benchmark truth nobody screenshots, and bookmark this before you buy on one number.

RetroChainer

42,451 Aufrufe • vor 2 Monaten

CHINESE AI TEACHER STACKED 4 NVIDIA DGX SPARK CHIPS INTO A CLUSTER - AND NOW RUNS A BUSINESS ON IT FROM HIS DESK 4 chips at $2,999 each - one time investment - and now he has a cluster that handles client inference requests around the clock while he does something else he's running a Qwen 32 35B model with 4 concurrent requests at the same time and the speed barely drops compared to a single request what used to require a $10,000+/month cloud setup now sits on his desk and costs him $40-60/month in electricity 4 chips at $2,999 each - one time investment - and now he has a cluster that handles client inference requests around the clock while he does something else the people still paying per token are financing someone else's infrastructure
1:42

CHINESE AI TEACHER STACKED 4 NVIDIA DGX SPARK CHIPS INTO A CLUSTER - AND NOW RUNS A BUSINESS ON IT FROM HIS DESK 4 chips at $2,999 each - one time investment - and now he has a cluster that handles client inference requests around the clock while he does something else he's running a Qwen 32 35B model with 4 concurrent requests at the same time and the speed barely drops compared to a single request what used to require a $10,000+/month cloud setup now sits on his desk and costs him $40-60/month in electricity 4 chips at $2,999 each - one time investment - and now he has a cluster that handles client inference requests around the clock while he does something else the people still paying per token are financing someone else's infrastructure

Sprytix

51,132 Aufrufe • vor 4 Monaten

Someone just posted the full blueprint for an AI swarm that does the job of a 200-person quant research team. Six agents. Running 24/7. Finding brand-new alpha while you sleep. Citadel needs 100 PhDs to do this. Two Sigma needs 200. This does it with six bots and one laptop. Two ways to play this - spend a weekend building your own swarm, or copy the wallet of one that's already up $2M: Boris Cherny runs Claude Code at Anthropic. Two weeks ago he said: "I don't prompt Claude anymore. I have loops running that prompt Claude. My job is to write loops" Alpha research is just a pipeline. So instead of sitting in it, you hand each stage to its own agent: > one reads every new research paper overnight and pulls out the trade idea > one builds the features and cleans the data > one backtests it over 20 years, costs and slippage included > one runs the hard stats and kills anything overfit > one checks it still works in every market regime > one strips out plain momentum and value to see if any real edge is left Each of those six is a job a fund pays a $600,000-a-year quant to do. He runs all six for the price of an API bill. The rule that makes it work: the agent that builds a signal never gets to approve it. A separate, stronger agent tries to kill it first. Whatever survives all six by morning is real, new alpha. One trader's already running this exact swarm on Polymarket. That $2M wallet is public, every trade on-chain. The full build is in the post below - six agents, the tool that runs them, and the five mistakes that kill most people. Bookmark & read this before it's buried.
0:41

Someone just posted the full blueprint for an AI swarm that does the job of a 200-person quant research team. Six agents. Running 24/7. Finding brand-new alpha while you sleep. Citadel needs 100 PhDs to do this. Two Sigma needs 200. This does it with six bots and one laptop. Two ways to play this - spend a weekend building your own swarm, or copy the wallet of one that's already up $2M: Boris Cherny runs Claude Code at Anthropic. Two weeks ago he said: "I don't prompt Claude anymore. I have loops running that prompt Claude. My job is to write loops" Alpha research is just a pipeline. So instead of sitting in it, you hand each stage to its own agent: > one reads every new research paper overnight and pulls out the trade idea > one builds the features and cleans the data > one backtests it over 20 years, costs and slippage included > one runs the hard stats and kills anything overfit > one checks it still works in every market regime > one strips out plain momentum and value to see if any real edge is left Each of those six is a job a fund pays a $600,000-a-year quant to do. He runs all six for the price of an API bill. The rule that makes it work: the agent that builds a signal never gets to approve it. A separate, stronger agent tries to kill it first. Whatever survives all six by morning is real, new alpha. One trader's already running this exact swarm on Polymarket. That $2M wallet is public, every trade on-chain. The full build is in the post below - six agents, the tool that runs them, and the five mistakes that kill most people. Bookmark & read this before it's buried.

cvxv666

104,237 Aufrufe • vor 3 Monaten

ONE BUILDER WIRED 4 RASPBERRY PI MODULES INTO A €450 AI CLUSTER WITH 64GB OF RAM AND KILLED HIS $200/MONTH CHATGPT SUBSCRIPTION 00:37 he points at the terminal, "loaded, loaded, loaded - 4 nodes synced over gigabit, 64 gigabytes total, llama 3.1 partitioned across the cluster" he stacked 4 raspberry pi CM5 lite modules with 16 gigabytes each onto a sipeed nanocluster board, the whole rig runs off a single 65 watt power supply and a gigabit internal switch distributed llama partitions the model across the nodes with synchronized workloads handled over the internal fabric, no quantization needed, the cluster hits about 30 tokens per second on small models a heavy developer pays $200 a month for chatgpt pro and another $200 for claude code, this rig cost €450 in parts and breaks even on the stack in month 2 most people will keep paying anthropic and openai forever, a few will spend a weekend wiring 4 raspberry pi modules and never see a subscription invoice again the window is open, follow and bookmark before it closes
0:37

ONE BUILDER WIRED 4 RASPBERRY PI MODULES INTO A €450 AI CLUSTER WITH 64GB OF RAM AND KILLED HIS $200/MONTH CHATGPT SUBSCRIPTION 00:37 he points at the terminal, "loaded, loaded, loaded - 4 nodes synced over gigabit, 64 gigabytes total, llama 3.1 partitioned across the cluster" he stacked 4 raspberry pi CM5 lite modules with 16 gigabytes each onto a sipeed nanocluster board, the whole rig runs off a single 65 watt power supply and a gigabit internal switch distributed llama partitions the model across the nodes with synchronized workloads handled over the internal fabric, no quantization needed, the cluster hits about 30 tokens per second on small models a heavy developer pays $200 a month for chatgpt pro and another $200 for claude code, this rig cost €450 in parts and breaks even on the stack in month 2 most people will keep paying anthropic and openai forever, a few will spend a weekend wiring 4 raspberry pi modules and never see a subscription invoice again the window is open, follow and bookmark before it closes

Antid

64,291 Aufrufe • vor 3 Monaten

THAT'S F**KING CRAZY, MY GROK BOT CONTENT DESK COSTS ME $20 A MONTH. I GOT QUOTED OVER $15,000 FOR THE HUMAN VERSION. I actually priced it out before I built it. Researcher, writer, designer, analyst, scheduler, editor – six roles, market rates, five figures a month before anyone ships a single post. Freelance instead of full-time? Cheaper, and then you own six calendars and six inboxes. → Seven agents replaced that line item. Here's the part that actually explains why. One narrow job each. A model asked to be six people is mediocre at six things. Asked to be one, it's one. Six worse prompts beat one perfect one. A fixed handoff order. The bottleneck was never ideas – it was the six hand-offs between having one and shipping it. Every hand-off is where a human team loses a day. Here it's a function call. A seventh agent that only routes. Six agents without a chief of staff is a group chat. With one, it's a desk. Nobody sleeps. Six people go home at six. The chain doesn't. Same week of work, compressed into an evening. Zero code. Seven prompts. Ten minutes to stand up. I still approve everything before it publishes – that gate I kept on purpose. And this hasn't been measured against months of results yet. When it is, the misses get written up too. I wrote it all down so you don't have to guess at the prompts. Seven agents, one article ↓
0:30

THAT'S F**KING CRAZY, MY GROK BOT CONTENT DESK COSTS ME $20 A MONTH. I GOT QUOTED OVER $15,000 FOR THE HUMAN VERSION. I actually priced it out before I built it. Researcher, writer, designer, analyst, scheduler, editor – six roles, market rates, five figures a month before anyone ships a single post. Freelance instead of full-time? Cheaper, and then you own six calendars and six inboxes. → Seven agents replaced that line item. Here's the part that actually explains why. One narrow job each. A model asked to be six people is mediocre at six things. Asked to be one, it's one. Six worse prompts beat one perfect one. A fixed handoff order. The bottleneck was never ideas – it was the six hand-offs between having one and shipping it. Every hand-off is where a human team loses a day. Here it's a function call. A seventh agent that only routes. Six agents without a chief of staff is a group chat. With one, it's a desk. Nobody sleeps. Six people go home at six. The chain doesn't. Same week of work, compressed into an evening. Zero code. Seven prompts. Ten minutes to stand up. I still approve everything before it publishes – that gate I kept on purpose. And this hasn't been measured against months of results yet. When it is, the misses get written up too. I wrote it all down so you don't have to guess at the prompts. Seven agents, one article ↓

SCOTTY BEAM

19,221 Aufrufe • vor 29 Tagen

HOLY...THE BOTS SCOOPED BLOOMBERG SIX GROK AGENTS PUBLISHED A HEADLINE 11 MINUTES BEFORE THE TERMINAL AND TRADED IT FOR $2,860 WHILE I SLEPT I turned Grok Bot into a newsroom. Not a chatbot that summarizes news. Six desks that find it, check it, write it, trade it and sell it, in that order, with an empty chair at the top. Why it works: Grok sees every post on this app the moment it exists. The terminal sees it after a human picks up the phone. That gap is eleven minutes on a good night, and a newsroom that never sleeps can live inside it. > MARA reads 1,900 posts a minute -> flags anything from a real place with a real job title. > OTIS needs two independent sources -> one post is a rumor, a post plus a document is a story. > JUNE writes 40 words, zero adjectives -> the headline the terminal will write later. > REX trades the gap -> 4% of the desk, entry 0.31, no approval step. > PIP fires the alert to 212 paying readers -> $19 a month each. > CHIEF holds the kill switch -> the only human decision is whether the desk keeps running. One night, off the log. 04:12 a crane operator in Rotterdam posts that the east quay went dark. 04:16 two sources. 04:19 headline. 04:21 position. 04:23 alert out. 04:34 the terminal catches up. 04:41 the market sits at 0.74. $2,860 on one story. $4,028 a month from readers who want the next one eleven minutes early. One evening to set up, no code. Six bots on one cloud computer, one skill each, one group chat where the story moves down the building. A newsroom is not writers. It is a rule about sources and a chair that stays empty. What should the desk cover next?
0:33

HOLY...THE BOTS SCOOPED BLOOMBERG SIX GROK AGENTS PUBLISHED A HEADLINE 11 MINUTES BEFORE THE TERMINAL AND TRADED IT FOR $2,860 WHILE I SLEPT I turned Grok Bot into a newsroom. Not a chatbot that summarizes news. Six desks that find it, check it, write it, trade it and sell it, in that order, with an empty chair at the top. Why it works: Grok sees every post on this app the moment it exists. The terminal sees it after a human picks up the phone. That gap is eleven minutes on a good night, and a newsroom that never sleeps can live inside it. > MARA reads 1,900 posts a minute -> flags anything from a real place with a real job title. > OTIS needs two independent sources -> one post is a rumor, a post plus a document is a story. > JUNE writes 40 words, zero adjectives -> the headline the terminal will write later. > REX trades the gap -> 4% of the desk, entry 0.31, no approval step. > PIP fires the alert to 212 paying readers -> $19 a month each. > CHIEF holds the kill switch -> the only human decision is whether the desk keeps running. One night, off the log. 04:12 a crane operator in Rotterdam posts that the east quay went dark. 04:16 two sources. 04:19 headline. 04:21 position. 04:23 alert out. 04:34 the terminal catches up. 04:41 the market sits at 0.74. $2,860 on one story. $4,028 a month from readers who want the next one eleven minutes early. One evening to set up, no code. Six bots on one cloud computer, one skill each, one group chat where the story moves down the building. A newsroom is not writers. It is a rule about sources and a chair that stays empty. What should the desk cover next?

slash1s

12,091 Aufrufe • vor 1 Monat