HNHacker News
TopNewBestAskShowJobs

y1n0

1,427 karma · joined March 10, 2024

submissionscomments
y1n0··on New 14‑inch MacBook Pro with M6
And you think 1 is better.
y1n0··on New 14‑inch MacBook Pro with M6
Are you kidding? When have they not? Apple II, II+, IIc, IIc Plus, IIe (also //e), IIe enhanced, IIGS. All sold concurrently.

They’ve always been this way.

y1n0··on I don't understand quantum physics (2018)
This has never not been true. There are always hand wavy axioms at the bottom.
y1n0··on LLMs are making formal software verification feasible at a much larger scale
I fully agree it’s harder to read. I do chip design and I feel like formal is more common in this area, but I generally restrict it to very simple properties that are easy to understand without having to mentally solve a sudoku puzzle.

It’s to easy to write an assertion that looks like it checks what you want it to, and gives you a false confidence when it passes, only to figure out after a field failure that it wasn’t checking what you thought at all.

But I’d pay for a commercial set of properties and assertions that had guarantees, and not worry about their complexity. For something standard like cache protocols and bus protocols.

Other cases it needs to be pretty straightforward to grasp by a peer reviewer in minutes, or it’s not worth it.

But when you can hit the sweet spot, a set of easily understood properties (or complex but trusted properties), formal can save massive amounts of time.

y1n0··on LLMs are making formal software verification feasible at a much larger scale
I think you still run the risk of common-mode failure. It might be more robust to pick a different model altogether to do either the VM or the spec.
y1n0··on LLMs are making formal software verification feasible at a much larger scale
I wouldn’t call them worthless. I guess it depends on the unit. But also having someone or something independently write tests is generally better than testing your own code.
y1n0··on LLMs are making formal software verification feasible at a much larger scale
It seems like it’s a good fit. Proving equivalence is much easier than writing a spec without a golden reference.
y1n0··on LLMs are making formal software verification feasible at a much larger scale
How do you know any test is checking what you want? You either review it, and/or you run the test on things you know are good and also things you know are bad.
y1n0··on LLMs are making formal software verification feasible at a much larger scale
You do it by comparing to something that is “known” to be good.

I didn’t read the whole article, just the appendix. There they talked about having Claude write a python interpreter. Then they had it write a set of properties to test and they vetted the properties against accepted existing python interpreters.

Proving two things are equivalent is a major component of formal.

For everything else, formal is only as good as the properties you specify. A long standing problem. But it’s a problem for all verification. If you write a shitty tests, you get shitty verification.

A key difference I think is that the same skills used for evaluating and reviewing application code apply to reviewing test code. That’s not really true for formal.

y1n0··on LLMs are making formal software verification feasible at a much larger scale
A little before that was:

> The most striking feature of agent proofs is that they are verbose, repeatedly spelling out details about the program’s execution: specifying the contents of CPU registers, explicitly writing out integer conversions, explicitly spelling out that many different addresses are all distinct from one another, writing out the exact contents of stack addresses, explicitly considering every possible branch in a function, etc.

> To be sure, these details are important to precisely reason about the execution of code. However, human proof engineers would have immediately introduced abstractions to factor out these types of repetitive specifications.

- - -

This matches my experience when I look at the tests LLMs write to check their own work when I’ve given them a task. The tests are almost universally terrible because they hard code everything in the most brittle way possible. They are nearly useless for regression tests because if I change anything about the source, even in non material ways, the tests all break in stupid ways that aren’t checking any important properties.

If I have the LLM make the then it also has to rewrite all the tests in a new episode of brittle concrete hyper-literal tests. The concept of a regression suite is non-existent.

It’s probably just as well. Having the implementer write the verification is generally a bad idea with people, maybe the problem of carrying over assumptions and blind spots and misconceptions applies to LLMs as well.

One might say it’s operator error, and I’m underspecifying things, but if I have to go to such elaborate lengths to get it to do what I need, I’ll just implement it myself.

y1n0··on Apple ML Research: Muse Spark Open with MLX on Mac Studio
The memory constraint isn’t design. It’s fab capacity. Apple has exactly 0 fabs. By the time they hired domain experts and built fabs and tuned to the point of reasonable yields, the crunch would have been over for 5 years, and apple would have wasted 20 billion with no return.
y1n0··on Tesla Roadster
The amount of time a ballistic object hovers with a speed of 0 is also 0. It literally never hovers. Look it up.
y1n0··on Tesla Roadster
Are you all so dense as to not grasp the difference between controlled flight vs ballistic?
y1n0··on Tesla Roadster
Controlled flight. Not ballistic.
y1n0··on AOSP Source Releases Are So Late Developers Disassemble Libraries to Report Bugs
A problem with legislation in general is that it is often very open to interpretation. Especially legislation around fast-moving technology fields. It’s very difficult to write both broadly applicable legislation that requires minimal interpretation. And in fact leaving the regulation open to interpretation can be a feature from the point of view of authority.

The only interpretation that matters is that of the Authority of enforcement and judgement. The implementer’s interpretation doesn’t hold much weight.

We can say what we earnestly believe the regulations mean, and how they can be met, but if the authorities say “not good enough” that’s all that matters.

y1n0··on GameBoy on E-Paper ESP32 at 30FPS [video]
E ink is a trademark name for electrophoretic displays (magnetic particles) by the company also called E ink.

E-paper is an informal category that anyone can use for any display type.

The display in this video/article is also electrophoretic, but not all displays called e-paper are and it’s aggravating. An example: https://kentdisplays.com/products/color-epaper-displays/

y1n0··on GameBoy on E-Paper ESP32 at 30FPS [video]
I don’t know if this particular display is magnetic particles or not, but epaper is not synonymous with eink. Something can be called epaper and still be lcd based. So I feel the video is misleading by declaring epaper is particle based.
y1n0··on OpenTPU – An open-source AI accelerator, developed by AI
I remember that movie. Gene Simmons from Kiss was the bad guy. He made a self propelled bullet of some sort that could target specific people.
y1n0··on An algorithmic failure beneath the secret ballot
I would never vote again if it wasn’t anonymous. Retaliation is real.
y1n0··on First Implantable Cochlear Implant Launches
I haven’t kept up to date with cochlear implants, but their capabilities used to be pretty crude. The bitdepth of sound being profoundly reduced to just 5 or 6 bits equivalent.

It wouldn’t take much to just sound like jumbled noise.

y1n0··on The first fully implanted cochlear implant reaches patients
I have heard of it, but other than giving something a name, what can be done about it?
y1n0··on The first fully implanted cochlear implant reaches patients
It’s pretty terrible. I started losing my hearing in my mid 30s. Best guess is genetics. My loss right now in my early 50s would be termed moderate, not quite severe but it continues to get worse.

Hearing aids have come a long way but all of the ones I’ve used still basically suck when it comes to understanding speech. Part of it, I think, is that my loss isn’t bad enough that fully closed ear pieces are recommended and as such background noise is still free to obfuscate conversation.

It basically makes conversation in anything but a quiet room extremely difficult. My brain works overtime trying to fit possible words to what I hear. Sometimes I fail with pretty humiliating results.

It doesn’t take much before you just stop trying to talk with people. Especially with the younger crowd because they tend to speak the quietest. It varies by individual, but at least older people who’ve had some loss themselves will speak louder to compensate.

The worst is restaurants. Usually I can guess what’s being said by the server, but not always. Accents increase difficulty.

I used to be pretty out going, joked a lot with people. Now I avoid most conversations when I can.

y1n0··on GPT-Synopsys: Frontier Intelligence to Revolutionize Chip Design
Compared to software formal methods have greater adoption. But there are a lot of things that fall under the “formal” umbrella.

The most common type that is used would probably be logical equivalence checking. Proving RTL and a netlist are equivalent is useful for catching synthesis bugs.

Or proving two netlists are equivalent after inserting test functions directly into a netlist, or some other netlist edit.

Property checking is what I use the most. You can check these during simulation which I wouldn’t call “formal” but you can also prove them using tools that use SAT solvers and whatnot to prove things mathematically.

As always, the tricky part of verification is writing the correct test or model. With formal we can use SystemVerilog assertions to write properties and sequences, but the difficulty in getting them right goes from trivial -> inscrutable very quickly.

It’s extreme easy to write assertions that pass and never realize your assertion was not doing what you thought and you weren’t proving what you meant to.

I haven’t used some of the more advanced tools so maybe they have ways to make this easier. But because of this I tend to just write assertions that are pretty easy to understand at a glance, and therefore closer to the trivial side of things.

If a peer has to solve a sudoku puzzle in their head to understand your work, then it’s unlikely the peer review will be worth anything. So I do what I can to make my work understandable at a glance (from a competent peer in the industry).

Of course making something simple can be quite challenging and often takes more time than leaving something complex and opaque.

I’ve never been involved in the foundry side of the work, and for ASICs, that is often half of the schedule.

y1n0··on Returning from vacation? The government can search your phone without a warrant
I’d look for a change in slope. A permanent change that adversely affected tourism would decrease the slope, but maybe not so much as to make it negative.
y1n0··on 10-year Treasury yield climbs above 5.3% to a level not seen in 24 years
If you look at the record, republicans out match democrats 3 to 1 for balanced budget years.

But it doesn’t matter. Those historical parties have nothing in common with their namesake parties today.

y1n0··on 10-year Treasury yield climbs above 5.3% to a level not seen in 24 years
There hasn’t been a balanced budget for 25 years. Today’s parties aren’t the same.
y1n0··on 56k.rip – the 1996 dial-up internet experience
I was so excited to get road runner, which was an isp from I think time warner, rolling out cable modem service in the early 2000s. I ended up taking a day of vacation just to ensure I was going to be home for the installation window.

I don’t remember if my computer at the time had built-in Ethernet or if I had to buy a card.

I have been always-online ever since. Wouldn’t have it any other way.

But I do still miss the BBS scene.

y1n0··on First Principles Thinking
Orthogonal means independent. One doesn’t inherently affect the other.
y1n0··on Samsung is expected to more than double output of its HBM4 and HBM4E DRAM
I’ve often wondered about die thinning. I mean just looking at sd cards and micro sd etc, it was clear that there had to be a thinning step (or many steps) and it seemed amazing to me that such a complex step could be economically viable. And yet it clearly is. All a matter of being able to operate at scale I guess.

It’s not a step I’ve seen discussed very much, so it was kind of neat to see it discussed front and center in a lay article like this.

y1n0··on Performance Improvements in .NET 11
It’s a spinal tap joke.
← PreviousPage 2 of 10Next →