theblazehen

There is! https://valetudo.cloud/

Well that's on me for assuming :/ I assumed that it did, based on it being explained in the blog post

Not necessarily. The RTS/CTS pins on the serial port are there for the device to tell the host "stop sending data, I need to process the data already sent"

The reason this doesn't work is that USB adapters don't have those pins, and thus assume you can send it at 30kbaud - not respecting when the device says it must pause sending data.

By introducing a static flow rate below what the CPU can handle, it should theoretically keep the flow rate below the threshold that the CPU starts to choke at

I've implemented the same idea with OpenCL: https://github.com/theblazehen/vramblk

There is originally https://github.com/Overv/vramfs however that has the overhead of a FUSE filesystem + loop device when using as a swap device.

The performance is rather lackluster however, it's far from a miracle "now you effectively have more ram for a 90% performance drop" - it definitely feels like traditional swapping

For me personally, it's because it retains the same window position, and I get back what I had before when it closes. If GUIs worked more like they did with windows in plan9, I'd definitely use them more

Yeah, copilot is only my backup - Codex and Claude are my primary, where it's per-token. Just looked up my user prompt count from my opencode db

I think we use it differently then, I tend to go heavy on the back and forth.

For speccing things out, I have a back-and-forth with the grill-me skill, break things down into tickets, as well as kicking off subagents. That said, I significantly overestimated the number of human messages I send.

My daily 90th percentile wrt number of prompts sent is sitting at 160 queries / day and average at 97 queries / day.

Ran an analysis of my last 2000 messages, with the following breakdown

Task delegation / execution: 23% Investigation / diagnosis / “what’s going on?”: 21% Planning / architecture / brainstorming: 15% Testing / verification / release ops: 10% Review / cleanup / quality control: 9% Course-correction / constraints / preferences: 8% Agent / ticket / workflow orchestration: 6% Providing context / evidence / pasted material: 4% Social reactions / acknowledgements / vibes: 2% Other: 2%

300 prompts in a day isn't that unreasonable to achieve on a heavy day? And Opus has a significant multiplier as well

models are judged by GPT-4

An interesting choice

There's a hard limit of 4 displays on Nvidia, even if they're chained via MST - that is how I have a couple of them already. I don't have the PCI-e slots for a second GPU

HN second-chance bump just had to happen when I had a power outage... Back up now :)

I'm unironically working on a proxy that filters sites like reddit, hn, etc by using user provided LLM rules

A screenshot would still make it clear that that's all you're getting - and no extra chrome etc

Here's Claude playing Detroit: Become Human https://www.youtube.com/watch?v=Mcr7G1Cuzwk

Sure, however kids these days often can't socialize irl - should kids be isolated from friends because they're unable to have any private conversations at all?

During times in which I was unable to socialize irl (eg school holidays), and unable to talk to my friends online, I can confirm that the isolation was not good for my mental health.

I suspect you're hitting the page where they're running https://iocaine.madhouse-project.org/

Perhaps you got bot flagged or something

I enjoyed chatting to Opus 3 recently around recent world events, as well as more recent agentic development patterns etc

Ah, in my spaces (Involved in the proxy dev), most people have been using it for Opus. I suspect they may even have more of an issue with it, as they don't get the cost advantage of serving an in-house model

Most people have actually just been using Opus through antigravity

It can. For android, I have it dump screenshots and uiautomator xml dumps, and for web the playwright mcp.

I find that a critical step in agentic development is to close the loop for the LLM so it can get direct feedback without needing you to manually handle it.

To expand - This refers to the Ralph Wiggum loop, which keeps repeating a prompt to the agent until it responds with a completion promise

https://awesomeclaude.ai/ralph-wiggum has some tips and examples of it

That would require end users to install additional software though, which they do not want

I'm adhd as well, so I get the pain. I tend to try and do frontend / backend on a single project to at least stay within the same domain

Appreciate your explainer, and agreed with you. The way it was written came off to me as "don't worry about the pain you cause others for their sake, avoid causing pain because it'll be bad for yourself"

Try running multiple agents - more task switching overhead, but I find planning in one agent while another is executing is a good balance for me, and avoids the getting-distracted trap

If you're a man, one of your hardest battle may be not giving in to sexual urges that cause harm to others. History is littered with otherwise entirely brilliant men who succeeded at everything but this. You must succeed.

I'm not sure I like the framing of this

I just launched mine! https://blazelight.dev/

Not much content yet, but I found the gimmick really fun to design :)

Mullvad

Can even get it in a single rack if you use SSDs

I love your work, it's always very fascinating. Been reading your posts for years