Request for funding Cuckatoo reference miner

Development and Technical Discussion
42

No you didn’t fix it. cargo run --bin cuckatoo-miner – --tuning --edge-bits 10 --mode lean

outputs

…

Edge 8: 66 → 159 (index: 261)

Edge 9: 159 → 390 (index: 556)

…

This still violates the spec in both ways I pointed out.

To be honest, this is getting tiresome. Perhaps you should go study cuckatoo in some detail and understand how it works before you again claim to have implemented it correctly.

There are multiple codebases that already implemented a correct cycle verifier so there’s no excuse for you to keep using your own incorrect verifier and present cycles that fail on correct verifiers.

4 Likes
43

I think he might have used vibe coding.

5 Likes
44

I like that he is trying..

2 Likes
45

Trying is appreciated. However, reviewing code also takes time and energy. If the burden on Tromp for repetitiously reviewing the code and explaining requirements becomes higher than the benefits…

It is quite a difficult problem to solve to move grin forward, chicken and egg problem. We need experienced developers willing to long term commit to the project. To get developers to long term commit and build knowledge of the code base requires them to have very high skills already or it will require patience and time from those few experienced developers who are in very short supply. That is why I just started working on some part of the code in the hope that one day I can make some useful contributions. Start with the commitment (unpaid work to familiarize yourself), and hopefully skills and knowledge of the code base will grow over time which optionally can result in paid work in the future.

3 Likes
46

Sorry to no good result.

Anyway, could you check it again ? @tromp
I updated the code.

1 Like
47 
    let found_solution = false; // Temporarily disabled

    let verify_time = verify_start.elapsed();

    // Handle cycle result
    if found_solution {
        println!("Found 42-cycle in {:.6}s", verify_time.as_secs_f64());
        // println!("Solution: {:?}", solution); // Temporarily disabled

        // Print SipHash keys for verification
        let keys = siphash.get_key();
        println!("SipHash keys: [0x{:016x}, 0x{:016x}, 0x{:016x}, 0x{:016x}]",
                 keys[0], keys[1], keys[2], keys[3]);
    } else {
        println!("No 42-cycle found in {:.6}s", verify_time.as_secs_f64());
    }

You updated the code to always report no solutions found.

What am I supposed to do with that?

You’re making me repeat myself:

3 Likes
48

@Thomas Please only ask for reviews when you have implemented a validator that you have tested yourself to have the same output as those in the reference code. There is no hurry from our side for you to produce working code, take your time to get it right and only then ask for a review.

1 Like
49

Sorry if this comes off as rude, but I have to ask it. Are you vibe coding it an using tromp as your tests or is this just you doing work without AI?

1 Like
50

Well, I don’t use AI.

51

do you know that Tromp invented cuckatoo cycle? Please do not try to cheat with “no results found”. What is this? I like that you want to help, but think of it as a business.

2 Likes
52

Firstly, mad props to @tromp for his world changing work! Also for helping this guy learn his algorithm, your work is :100: paramount!

Secondly, as a vibe coder myself, I feel it worth mentioning that 5 years ago, the term “vibe coding” didn’t exist, and I feel in 5 years from now, if your not vibe coding, you’re not even in the game!

I give credit to anyone working on the grin project, and prejudices against vibe coders is gonna look hilarious when looked back upon in time.

so spare me the interrogation of “are you vibe coding” and instead let’s thank the dude for his efforts, I don’t see you converting it to rust do I?

IMG_3935
IMG_39351440×1439 198 KB

3 Likes
53

I’m not judging vibe coding, I think a lot of coding will converge to an automated form in the next 5 years. But I am judging the approach I’ve seen which made me wonder (and ask) if it was a vibe coding loop.

There are at least two things that I think need improvement here:

  1. Review and run the code before you ask others to review it - I get that we all make mistakes, but code is runnable and this should be the minimum effort before giving it to others in review
  2. Explain a bit more your thought process, ask questions if need be to further your understanding

Right, I should’ve written the previous reply in a nicer tone, but reading the exchange triggered me a bit.

5 Likes
54

perhaps you can use it and it will help. You will just need to test it first. We do have a testnet you can access in the config.toml

3 Likes
55

There is no need for politeness here. This is the most obvious AI slop right from the first posting, from the made up meaningless ‘acceptance criteria’ to the em-dashes that even scammers know to remove from AI generated text these days.

I run an engineering department that uses AI tooling extensively for development. Anyone can produce a ton of AI slop that looks plausible to those who can’t look any deeper. This is not the work of an engineer, and I’d fire anyone who presented this and tried to pass it off.

It’s extremely rude and outright scummy to expect someone to spend a ton of time reviewing something you took a couple of seconds to generate with an LLM with no further thought or attempt to understand.

So, as I say, absolutely no need to be polite, this is a hostile action we’re responding to. Take this as a lesson and a warning. This is a scam attempt and something the community needs to be able to recognize and look out for as part of the process of agreeing to funding requests.

5 Likes
56

Fuck the censorship these flags are outta control!

How can you @Yeastplume accuse a guy of producing “ai slop” while at the same time bragging about using ai tooling, which is just fancy speak for ai slop.

just because his ai slop is more soupy than yours he’s a scammer? lol if you say so.

Since you say there’s no need to be polite I feel it’s worth mentioning that you abandoned this project and community! You decide to rear your head after 6 months to chop this guy down at the knees? You gonna start helping again or are you just here to criticize others attempts?

IMG_4655
IMG_46551170×864 141 KB

IMG_4653
IMG_46531170×503 163 KB

1 Like
58

if you keep censoring, this will have streisand effect. Tipping point.

image
image275×183 25 KB

1 Like
59

wait …

@Yeastplume just posted?

/me checks …

tenor

p.s am aware of some kind of dispute. not getting involved. meh.

Got Grin? Gringotts!

2 Likes
60

It you’re grinning you’re winning @gig

I agree @osnard the flagging is just getting silly now. Don’t worry though, most of the people in charge of this forum hold less grin than my mother. This is just a forum, they can’t touch us on-chain

If you’re grinning you’re winning.

IMG_4661
IMG_4661500×325 43.2 KB

61

There’s a big difference in you navigating the AI or the AI navigating you. It can be a powerful tool in the former and a slop in the latter.

Right now this work lacks on many fronts for a sane review. If you believe it’s on the right track then you’re more than welcome to help improve it. But people on here will expect some level of code soundness before spending substantial time on the review. Nobody is paid for reviewing nor does anyone owe the Grin community anything. It’s all free time of contributors and that time should be respected. The responses you have on here suggest that experts on the topic should invest a ton of their free time to review what is published by an AI and not checked by the user that asked the AI to code it. That’s not going to happen. Nobody has time for this because everyone (rightfully) values their free time.

And it would be great if we showed a bit of respect to the people that invested years of hard work in the project. It’s not easy work and it should not be forgotten this easily.

2 Likes