[flagged]
A specific women? Women in general? Is vague-posting back in vogue?
I regret building this $3000 Pi AI cluster
141–150 of 377 posts
Re: I regret building this $3000 Pi AI cluster
#142Earlier quoted context omitted.
The cost effective way to do it is in the cloud. Because there's a very good chance you'll learn everything you intended to learn and then get bored with it long before your cloud compute bill reaches the price of a desktop with even fairly modest specs for this purpose.
I can get a Xeon E5-2690V4 with 28 threads and 64GB of RAM for about $150. If you need cores and memory to make a lot of VMs you can do it extremely cheaply
Power and noise - old server hardware is not something you want in your home.
Commodity desktop cpus with 32 or 64GB RAM can do all of this in a low-power and quiet way without a lot more expense.
Re: I regret building this $3000 Pi AI cluster
#143Earlier quoted context omitted.
The cost effective way to do it is in the cloud. Because there's a very good chance you'll learn everything you intended to learn and then get bored with it long before your cloud compute bill reaches the price of a desktop with even fairly modest specs for this purpose.
I can get a Xeon E5-2690V4 with 28 threads and 64GB of RAM for about $150. If you need cores and memory to make a lot of VMs you can do it extremely cheaply
For $3000 you can get 3x used Epyc servers with a total of 144 cores and 384 GB memory, with dual-port 25Gbe networking so you can run them in a fully connected cluster without a switch. It will have >20x better perf/$ and ~3x better perf/W.
That combo gives you the better part of a gigabyte of L3 cache and an aggregate memory bandwidth of 600 GB/s, while still below 1000W total running at full speed. Plus your NICs are the fancy kind that let you play around with RoCEv2 and such nifty stuff.
It would also be relevant to then also learn how to do stuff properly with SLURM and Warewulf etc. instead of a poor mans solution with Ansible playbooks like in these blog posts.
Re: I regret building this $3000 Pi AI cluster
#144I thought the conclusion should have been obvious: A cluster of Raspberry Pi units is an expensive nerd indulgence for fun, not an actual pathway to high performance compute. I don’t know if anyone building a Pi cluster actually goes into it thinking it’s going to be a cost effective endeavor, do they? Maybe this is just YouTube-style headline writing spilling over to the blog for the clicks. If your goal is to play…
The cost effective way to do it is in the cloud. Because there's a very good chance you'll learn everything you intended to learn and then get bored with it long before your cloud compute bill reaches the price of a desktop with even fairly modest specs for this purpose.
It's good for the soul to have your cluster running in your home somewhere.
Re: I regret building this $3000 Pi AI cluster
#145Earlier quoted context omitted.
Isn't your home lab supposed to make you learn stuff? Why would you leave it idle?
You wouldn’t, it’s given as a lower bound, it costs more than that when not idling
but then you’d turn it off, if you don’t then cloud is much more expensive too.
Also $150 for 100w is crazy, thats like $1.70 per kWh; it would cost about $150 a year at the (high) rates of southern Sweden.
Re: I regret building this $3000 Pi AI cluster
#146Earlier quoted context omitted.
I use mine for a plex server. I don't need to transcode + I need something I can leave on that draws little power. I have a powerful rig, but the one time I get to turn it off is when I'd need the media server lol. There's a lot of scenarios where power usage comes into play. These clusters don't make much sense to me though.
That’s insane, drawing very little power from an always on server is a solved problem.
Not to mention there's more cost, effort, and energy effective compute from old laptops.
Unless you have a robot body for your potential RPi, don't buy one.
Re: I regret building this $3000 Pi AI cluster
#147Earlier quoted context omitted.
The cost effective way to do it is in the cloud. Because there's a very good chance you'll learn everything you intended to learn and then get bored with it long before your cloud compute bill reaches the price of a desktop with even fairly modest specs for this purpose.
That really depends on what you want to learn and how deep. If you're automating things before the hypervisor comes online or there's an OS running (e.g: working on datacenter automation, bare metal as a service) you will have many gaps
If you want to run something like GNS3 network simulation on a hosting service's hardware you'll either have to deal with hiring a bare-metal server or deal with nested virtualisation on other people's VM setups. Network simulation absolutely drinks RAM, too, so just filling an old Xeon with RAM starts to look very attractive in comparison to cloud providers who treat it an expensive upsell.
Re: I regret building this $3000 Pi AI cluster
#148[flagged]
Can you elaborate? I don't know what you are referring to.
Re: I regret building this $3000 Pi AI cluster
#149Earlier quoted context omitted.
I can get a Xeon E5-2690V4 with 28 threads and 64GB of RAM for about $150. If you need cores and memory to make a lot of VMs you can do it extremely cheaply
On Aliexpress those Xeon+mobo+ram kits are really cheap.
1. Not in the US with tariffs now.
2. I would not trust complicated electronics from Aliexpress from a safety and security perspective.
Re: I regret building this $3000 Pi AI cluster
#150I thought the conclusion should have been obvious: A cluster of Raspberry Pi units is an expensive nerd indulgence for fun, not an actual pathway to high performance compute. I don’t know if anyone building a Pi cluster actually goes into it thinking it’s going to be a cost effective endeavor, do they? Maybe this is just YouTube-style headline writing spilling over to the blog for the clicks. If your goal is to play…
The author, Jeff Geerling, is a very intelligent person. He has more experience with using niche hardware than almost anyone on earth. If he does something, there's usually a good a priori rationale for it.