Member Info for Phimx

Mickeydj :

No you have it wrong, Usersteve did post those statements

Usersteve: You posted

NDA #7: 8 April 2025

"[QBT] has signed a Non-Disclosure Agreement ("NDA") with a leading manufacturer of ASIC chips for bitcoin mining."

NDAs #8, #9, #10: 5 Dec 2025

"QBT has entered into three non-disclosure agreements ("NDA") with three separate ASIC manufacturers, each of which has developed its own Bitcoin mining rig."

Of course it's not clear if 8,9 and 10 include 7 and 8 or are they 5 distinct NDAs

Gardin is as confused as investors are

In the words of Gardin

" Should the AI Oracle calculate that the current SHA-256 input will not be successful in finding the winning hash, it skips that calculation and moves on to the next input."

The device performing the Oracle function is the ESP32 or whatever processors are being used in a mining rig running the Oracle software. They operate relatively slowly and low power, they just send instructions and one time header data to the awesomely fast ASIC. Actually the ASIC isn't that fast in terms of clock speed, its power comes from its two thousand or so identical SHA256 circuits each churning through thousands of different NONCEs each second.

Software in a microcontroller operates one instruction at a time and it may take tens or hundreds of individual instructions to perform quite simple programs I.e. blinking a LED, always a classic! If we ignore dual cores then the controller is a simple serial device churning away its program in micro-seconds, a billion times slower than the ASIC can process hashes.

Oldsharedog: You post

" Phimx - Maybe you're missing something - ever thought of that one ??

All the time, all the time.

I also know Gardin was at pains to point out several times that Method C had to be implemented on an ASIC. To discard 30% of perhaps 1 Tera inputs requires a process that process has to be repeated for each NONCE that takes time and an ESP32 is a slow device. How can it no magically be fast enough? How will it be fast enough for after market hashing boards containing many multiple ASICs hashing at 500 TH/s.

Commonsense is not in evidence.

Mrhanji: You post

" Ignore the Fud from Steve & Phim & all profit .. lol"

Why not explain to everyone how a slow ESP32 at 240 MHs can filter inputs to SHA256 circuits 𝗶𝗻 𝗿𝗲𝗮𝗹 𝘁𝗶𝗺𝗲 𝗻𝗼 𝗹𝗲𝘀𝘀 at the awesome rate of 1 ps I.e. one thousandth of a billionth of a second.

If you can explain that you're a very clever man.

But you can't you'll continue saying it's FUD without any idea or evidence of what you are talking about

Forfaiter: You post

.

" The old Nda's did not involve method C software incorporation which should be quicker to market."

Wrong. It depends how the NDA is worded. It could be in general terms I.e. information relating to all Methods and their derivatives or actual specific details relating to a specific Method.

Simply depends upon what the parties agree to.

Forfaiter:

The NDAs have provided plenty of mileage for Gardin, are we talking about the NDAs signed years ago or new ones. It's all spins, twirls and pivots and of course throw in plenty of seemingly endless discussion. Bobs your uncle nobody has a clue what he's achieving or what is the latest priority.

It's as confused as he is. He actually slipped in Method D which will save miners running methods on their miners. Another pivot into nonsense.

Allforprofit: You post

" Abandonment of he application (and write-ff of the significant sums spent on making it) seem inevitable. I expect the abandonment will be rewarded with a performance bonus for Gardin."

Whatever money was spent I'm sure was bettered by the credibility/kudos patent RNS brought to the company. In exactly the same way that adding strategic advisors, with credible CVs produced impressive RNS to leave the t naive open mouthed. There lots of intelligent fools in this world and AIM takes their money greedily.

Allforprofit:

Undoubtedly, if the patent is withdrawn before the 18 months is up then it won't be published. Consider a patent application with real value and substance, if it was withdrawn and the patent office still published it then the game is up for the applicant his valuable work is toast. The patent office has cost the applicant a potential fortune.

Allforprofit:

I know little about the ins and outs of patents. There is an inherent principle built into the system if your idea isn't patentable it's because it doesn't meet the requirements. That sounds so obvious as to be silly but an important concept is novelty and must not be prior art I.e. it must not be in the public domain. You can see how the system must work similar to proof of work in Bitcoin mining. You need to have taken step others haven't taken. You can't just paint pink dots on a pedal scooter and claim you invented it. There has to be something that can be actually protective.

Gardin's first patent, essentially what he paid Naik for, the optimisations was published way before the patent filing. Additionally at the end of his thesis he actually stated that he was making the work public for the Bitcoins industry's betterment or something similar. Both Gardin and his patent agent must have known this but they carried on, at least the application made great reading in a RNS and gave a dash of credibility to the scheme. I'm not sure decision trees will be patentable it's a bit like inventing a language that's possibly maybe but trying to patent all the variations of the alpha bet to make up words wouldn't be. So software is a tricky area the concept of decision trees was probably an "invention" but creating different inputs isn't.

Short answer is - if you can't patent your thing it won't be protected when you sell the first one, which will be ripped apart in a Chinese workshop. But the guy will be disappointed because your device didn't have anything of worth or novelty inside it. So divulging your plans in a patent doesn't necessarily tell someone how to make it he'll know already if he's skilled in that area.

Tricky to get over what I'm trying to say but it's a principle that's been around a while.

Allforprofit:

Well done for that interpretation, I found the transcript very difficult to follow, I had the sound of watching the latest sham interview. It is more like some form of therapy session or role play for dementia suffers. I find it embarrassing to listen to, can't the guy even spend a small amount of time to sharpen up his delivery, has he no shame?

Mrhanji: You post

" Hmm, Phimx - judging by the lack of responses, the mechanics chat is yesterday’s news. And honestly, it’s not that interesting anymore. "

Says everything then Mrhanji, the vast majority of QBT's don't understand a thing about the Methods. I note that you refrain from commenting also. You regurgitate the nonsense Gardin writes under the flickering candle in his sparse flat in Roviga. He is slowly running out of ideas.

No-one wants to crystallise their losses so likes of you and Mickeyare pretending to support this farce while desperately praying for a spike to ease your painfull losses. I predict you will soon be sat on bigger losses that's how the AIM game is played.

Ask Maidit he's a bit of an expert.

Ahhh yes the "weird nonce gaps"

Miss them out and you can claim an improvement in your hashing success compared to a system which includes those nonces. A read for tonight while I await Gardin's responses.

I'm claiming the Golden NONCE I believe

MICKEYDJ IS GARDIN

Mickeydj:

I like the story you made up but it seems like another garbled Gardin effort. batch prefiltering NONCES. But to hash an input requires 2 basic elements the header information for the current block to be hashed and NONCES.

How would pre-filtering the NONCES help. The ouput hash also depends on that input data. Unless we are harking back to the strange gaps that appeared in a query a few years ago !!!? Is Gardin using that to claim his 10% improvement I wonder.

Mickeydj: You post

" Method C Operation: It's batch pre-filtering (nonce ranges scored in ms on ESP32-S3)"

There has been no evidence or inkling that NONCE ranges are score in batches, in fact quite the opposite as posted below and repeated here.

"Should the AI Oracle calculate that the current SHA-256 input will not be successful in finding the winning hash, it skips that calculation and moves on to the next input."

You are inventing the batch NONCE pre-filtering nonsense. You are attempting to discredit my posts and hoodwink investors with makeup nonsense. I thought you were better than that!

Mickeydj: You post

" If this is impossible, why oh why oh why oh WHY is US ASIC testing (Dec 5 update)?"

Because the only saying ASIC testing is taking place, on the planet, is the chief mumbler himself.

Mickeydj:

At present, in the Method C hardware, the filtering is achieved by the QBT FPGA there is nothing known about anything upstream.

BUT that's irrelevant but the subject here is the software version of Method C running on an ESP32 Mictrocontroller.

Mickeydj: From 17 Ja. 25 RNS

he AI Oracle achieves these material advantages through being able to predict the likelihood of an input to SHA-256 to produce a winning hash above a certain level of difficulty. Should the AI Oracle calculate that the current SHA-256 input will not be successful in finding the winning hash, it skips that calculation and moves on to the next input."

It skips the calculation and moves onto the next input I.e. one after the other in otherworld serially. There is no batch upstream device.

Quote The date of the Jan 2025 RNS that indicated batches and upstream filtering please

Mickeydj:

Where is up-stream ?