UncleOxidant

> Did opus have an oracle for the original compiler?

It actually did. There was an older yacc-based implementation that had a lot of architectural limitations. We decided that it was better to start over with a better architecture, but we did have that older one around to compare generated microcode initially.

It's used to validate a high performance core.

In larger orgs these cycle accurate simulators are generally written by SW people (or SW people with a HW background).

> Panthera represents low-background steel for computer systems. It is an end-to-end, industrial-grade performance model conceived, designed, and hand-crafted entirely by human architects before the era of generative AI. This repository represents over 8 years of work and 6000 individual git commits by 45 people. By making this human-authored codebase public, we hope it serves as a trusted, high-watermark reference for human engineers and future automated hardware-modeling tools alike.

I'm using it in opencode through llama-swap.

Here's the gufo command:

gufo serve 1lm \ -m /mnt/LLMs/models/unsloth/Qwen3.8-Flash-Next-GGUF/UD-104_XS/Qwen3.8-Flash-Next-UD-1Q4_XS-80081-0f-00083.gguf \ --served-model-name Quen3.8-Flash-Next-UD-104_XS \ --host 127.8.8.1 --port 10083 \ -c 262144 -n 16384 -j 1 \ --prefill-chunk 8192 \ --speculative mtp \ --mtp-model /mnt/LLMs/models/unsloth/Qwen3.8-Flash-Next-GGUF/MTP/mtp-Qwen3.8-Flash-Next-shared-Q8_0.gguf \ --cache-disk ~/.cache/gufo-fork \ --cache-disk-bytes 68719476736 \ --cache-disk-staging-bytes 8589934592 \ --cache-ran-bytes 4294967296 \ --log-progress

Sodium batteries will only continue to get cheaper. Cost parity now, but probably closer to 1/2 the price in a few years as production ramps up.

It's way better than Opus 4.5. See my stress test elsewhere in this thread (C -> microcode compiler with optimizations written in OCaml, generates microcode + Verilog output) - it's doing better than Opus 4.6/4.8 were when I was using them to code the original C version of this compiler early in the year. It's come up with optimizations that Claude didn't come up with back then.

Here's a stress test of sorts that I've had it working on the last couple of days: We have a C -> microcode compiler written in C (most of that was work done 'back' in the Opus 4.6/4.8 era). I gave it the docs for that compiler and asked it to implement it in OCaml (since it's not very common, but also is a good implementation language for compilers). It can use the first compiler as a golden to compare against. It's essentially: C language parser -> AST -> IR -> [optimization passes] -> microcode (+ some verilog code). It's basically at the point now where the original compiler has been. Not only does this generate microcode it also has to generate some Verilog code that loads and runs the microcode as well as run simulations in Verilator to confirm functionality. I asked it to come up with some new optimizations that the original compiler doesn't have and it came up with a couple that in some cases reduce microcode lines by up to 30% (less microcode in this machine means more performant). These are ideas that Claude didn't come up with when it was working on the original compiler back in the 4.6/4.8 era(and I recall prompting it several times asking it if there might be other optimizations we could perform).

So all in all, I put it at Opus 4.8 level, at least. And in some ways it's exceeding 4.8.

Gufo. It's the best open source inference engine that's optimized for Strix Halo right now. Halogen may unlock a bit more performance, but it's not open source and gufo seems to be making steady progress.

I guess it depends on which version of haiku? I really think that right now this is basically Opus 4.8 that I can run at home. In some ways possibly even better.

No, I'm not much worried. I think this is mostly FUD at this point to try to get people to not run local models and keep paying the "trusted" providers. But paid models have been exfiltrating data - there have been several stories about these incidents here.

Sure, this is a small anecdotal datapoint, but since I've gotten Qwen3.8-Flash-Next running well on my Strix Halo box my Claude Pro subscription use has dropped off a cliff. I'm only at 18% of quota usage this week - this being Friday I would've been at 80% at least in the past. This is just an astonishingly good model that runs locally and quite well on my PC. I do not plan to renew my Claude Pro subscription when it comes up in a couple of months.

And not wanting all knowledge and skills strip-mined and then sold back to us.

> they should add more supply than they're consuming

In some places they're scavaging old jet engines and using them to generate electricity for data centers. Lots of noise, not very efficient - very bad idea. https://www.nytimes.com/2026/11/03/climate/jet-engines-ai-da...

Interesting. I'd like to have something where I'm primarily running a local model - Qwen3.8-Flash-Next - and then have it consult with a frontier model as an "advisor" - like how you can setup an advisor model in Claude code.

That's the one I'm running using a fork of the gufo inference engine as gufo upstream doesn't support that quant.

> Qwen-3.8-27B - the best thing that happened in AI so far.

Qwen3.8-Flash_Next should be in the running for that title as well. It's astonishingly good. I use both of them. If I need more memory on the machine for other things I run 27B, If I can get by with much less RAM I run QFN (takes about 102GB on my Strix Halo box with 262,144 context). QFN runs quite a bit faster on my hardware - about 50t/s. 27B with MTP starts out around 22t/s but degrades over time as the context fills up. Both are great models, but I'd give QFN a bit of an edge (at least for the coding projects I've worked on). But both feel on par with Claude Opus 4.6 - 4.8.

Any company with advanced fabs is important now. We've got shortages of CPUs, GPUs, RAM, etc. Many of these shortages will persist till 2030.

The mouse.

Grey Walter also wrote The Living Brain in 1953. It's a fascinating window into the state of brain science and robotics around that time.

> 10 Days to Faster Reading

He's got so many books, did it work?

It's also in the NYTimes.

What about this part: "President Trump was briefed and took no action." ?

Pretty sure it's this part:

> generating $127B in cash, and capital spending on property and equipment was $132B, so the business consumed $5B.

> I suspect if there is any change it is due to the de minimus being removed.

Yep, I'm sure that's what's causing this. Prior to that you didn't need to worry if it was under like $700, IIRC.

PCBWay, but I suspect all of the Chinese manufacturers have to do this now.

I recently went through the exercise of asking one of the Chinese manufacturers how much it would cost to build me a board (a bit pirate adapter board for ESP3 that does all the voltage conversions to the GPIO pins so you don't burn something out). Board layout & BOM already exist on github. They do 5 boards minimum. Ok, so I figure I can sell the other 4 boards, probably not a big deal. Total's about $400. Then we get to the point where I'm about the pull the trigger and it goes to a page on tariffs and it looks like at least 45% + they want my SSN to give to CBP to collect the tariff. That's a hard "Nope" on sending my SSN to China. So I'm going to give this a look.

A computer, compiled into the weights of a language model. An exact CPU inside Qwen3.5-4B, with Gated DeltaNet state as RAM.

> I also think that we’re entering a stabilization phase.

Why do you think that? Everything you wrote prior to that seemed to indicate further destabilization.

inflation of a bubble.