evanreichard

It's quite stable. Been running NixOS on my M2 Max with basically no issues (HDMI working fine as well). However the lack of adequate sleep / no support for suspend to disk led to me buying a Framework 13 Pro, and I'm now selling the M2.

I don't think it's obvious. So, how does it not meet the criteria?

As opposed to loading it up in RAM + VRAM? Pretty much always better to split it up to multiple GPUs. My priorities are load up all available VRAM, then offload MoE experts to RAM (if possible), then offload other layers.

I use an RTX 3090 (24GB) and a GTX 1080ti (11GB). Just the 3090 for 3.8 I get maybe 60tg/s (UD-Q4 quant), for both I get around 40tg/s (UD-Q6). Not apples to apples though considering it's different quants.

Here's my config: https://gitea.va.reichard.io/evan/nix/src/branch/master/modu...

Not sure if it was the same bug, but I had a storage issue where System Data ballooned to like 200GB.

It had the most bizarre solution; airplane mode, set time to one year in the future, reboot, wait a few minutes, set time to 6mo in the future, reboot, wait a few minutes, set time to now, reboot. Went from 200GB to like 15GB. Was ridiculous.

(For anyone looking at this and considering doing it, you also need to ensure iMessage retention is forever, otherwise the iPhone will think it's a year old and delete the messages)

There's also https://github.com/ggml-org/llama.vim

Which I've been using with Qwen3 Coder. As long as infill is supported, that should work. I'll try later today.

How did you qualify a project as vibe coded?

It's an homage to MITRE ATT&CK.

Have you tried llama.vscode [0]? I use the vim equivalent, llama.vim [1] with Qwen3 Coder 30B and personally feel that it's better than Copilot. I have hot keys that allow me to quickly switch between the two and find myself always going back to local.

[0] https://github.com/ggml-org/llama.vscode

[1] https://github.com/ggml-org/llama.vim

At the end of the day it still has to open and close when you press a button. You can just wire a relay to complete the circuit on one of the remotes.

It's what I did when I got a new opener. Works fine in HomeAssistant.

It's wild. I bought 64GB (2x32) DDR4 SODIMM (CT2K32G4SFD832A) for $100 this April. Cheapest I can find it today is $270.

I picked up 32GB (2x16GB) DDR4 (CMK32GX4M2E3200C16) last September for $55. Now it's $155.

The video shows both directions.

This year I redid my whole lab as well and prioritized declarative everything so it'd be easy to go from scratch if needed. So I've got nix to create RKE2 images with cloud-init for XCP-ng templates, Terraform with XenOrchestra provider, and ArgoCD for the K8s cluster.

I've got two R640's so I can live migrate, and an R720XD with TrueNAS (democratic-csi for K8s persistence). QSFP (40Gb) for TrueNAS / R720XD, and SFP+ (10Gb) for R640's linked to a Brocade ICX 6610.

So I can update the hosts, and K8 nodes with 0 downtime. Do I need it? No, but I learned a lot and had / have fun deploying and maintaining it.

I love Garage. It just works. I have Garage running on a few older Odroid HC2's, primarily for k8s Velero backup, and it's just set and forget.

The current version of Bitwarden straight up doesn't work in Orion [0]. I ran into this yesterday when setting it up for the first time. Wasn't a great first experience as it's literally the only extension that is a deal breaker to me.

I'm still giving Orion a chance for now... I just installed a slightly older version that works.

[0] https://orionfeedback.org/d/10197-bitwarden-hangs-on-load

FWIW I've been running a jailbroken PW5 for a few years now always connected to the internet without any issues dealing with updates.

I did rename the ota binary. I'm aware that there's always the possibility of Amazon maybe having some other way to push an update, but I haven't had any issues so far.

I've got the same generation PW and have it jailbroken running KOReader. I've considered trying other readers out, not because of issues but rather shiny new thing reasons. But at least when it comes to KOReader, it seems like the PW are the best if you can jailbreak the version you're on.

(I want / need it to run KOReader because I wrote a small Lua plugin for it that syncs reading stats (words per minute, minutes read per year, etc) to a centralized server.)

As someone who used CrowdStike daily and worked as an MDR Analyst and Engineer at a top ranked MDR provider, CrowdStrike is a very capable piece of tech.

While the driver for purchase is almost always to pass audits, it's still a good product.

FYI I've been running multiple jailbroken Kindles connected to WiFi with KOReader for a few years now. Just install renameotabin and you're good to go. You can even register it like a normal Kindle.

The relicense doesn't bother me. That straight up lie, though.

All of my phones (except the iPhone, actually) had hard reboot issues at around the 2-3yr mark. Usually at under 30% battery charge under high load. HTC Incredible, Galaxy S4, and Nexus 6. Nexus 6 was the worst, but I also had that one the longest.

My iPhone X didn't have shutoff issues, but the battery did swell enough to push out the screen, which Apple replaced for free even with an expired warranty.

Yeah - I usually watch the network tab in debugging tools to figure out whats being blocked, then whitelist and try again.

I also realize that you shouldn't expect most people to do that, let alone know how to.

I am someone who is very aggressively anti-ad.

I can empathize with the sometimes aggressive blocking, and as you pointed out can be pretty block list dependent.

I generally will go in and whitelist things if a site breaks due to a DNS block, but of course putting your partner on the same VLAN can be problematic. I "got around" that by having a button in Home Assistant that will completely turn off Pi-hole (and now AdGuard). So my partner will go in and toggle that if there's a problem.

AdGuard Home does also have the ability to completely disable blocking for specific clients.

I use the bare WireGuard app on iOS. I just statically set the DNS server to the AdGuard Home IP (or Pi-hole IP) on my local network in the app.

I'm curious what issues you ran into with Pi-hole? I was running my instance for years without a single hiccup. I ended up moving to AdGuard Home about a year ago though because I wanted to run it on my OPNSense box.

I have an automatic WireGuard VPN set up on my devices to VPN into my home network when I'm not connected to my SSID, so my local DNS still works remotely.

That’s wild. I’m right outside DC and last month I used 2066kWh for a total bill of $258.85. So, 0.13c/kWh.

I’m all electric heat pump. That ran for about 300hrs last month, so accounts for around 800kWh.

I used to run a 5 node Ceph cluster on a bunch of ODROID-HC2's [0]. Was a royal pain to get installed (armhf processor). But once it was running it worked great. Just slow with the single 1Gb NIC.

Was just a learning experience at the time.

[0] https://www.hardkernel.com/shop/odroid-hc2-home-cloud-two/

Another data point - I'm running about 500w for my small rack. NAS (20 x 3.5HDD), ICX6610 (3 PoE APs), 2 x R630, and some small devices.

According to iDRAC, my R630's are drawing almost exactly 100w/each. Each with about 75 pods (k8s nodes).

> The outcome is considerably more time spent looking at the screen [...]

Quite the opposite. The wheel has discrete clicks while rotating so muscle memory becomes a thing. I've got half a dozen "workflows" memorized so I don't even need to look at the screen to do things like love a song, add it to my favorites, switch between apps, report an accident, etc.

Definitely recommend. We were at around 14 (pCi/l) as well. Peaks up to 20. Got an active mitigation system installed last year for around $1k. Hasn’t exceeded 2 since.

Unforeseen benefit of no longer having to run my dehumidifier this summer. So it’ll end up paying for itself in electricity cost after maybe a decade hah.

I have a similar (less polished) side project I threw together in a weekend [0][1]. I use ChatGPT's API and it costs ~0.5c / article generated. So IMO very reasonable.

[0] https://gitea.va.reichard.io/evan/VReader/src/branch/master/...

[1] https://vreader.va.reichard.io/