Member Info for Phimx

Awl007:

Bashing ?

From my post below I'm pointing out that a ESP32 Microcontroller running at 240 MHz will take days to sift through 1 Tera possible inputs to a BitAxe 1370 ASIC.

My figures underestimate the time of 2.33 days since owners are claiming actually hash rates of 1.5 TH/s using modifications but I'm being generous to Gardin.

If you wish to ignore this 𝗶𝗺𝗽𝗼𝘀𝘀𝗶𝗯𝗹𝗲 feat Gardin is claiming go ahead.

Allforprofit:

There is only one executive director ... guess who?

So campers we have a software Oracle running on a 240 Mhz processor (BitAxe 1370) now that's about 4 ns per cycle, bear that in mind. Now the software is going to run through the header information with a NONCE value and somehow decide if this input to a SHA256 engine is worth hashing and we know that 10% are rejected. If we make an absurd assumption that the processor takes only a single clock cycle to decide if an input is worth hashing then, also assuming an unmodified BitAxe hashes 1 TH/s, it takes 4,000 seconds to process the inputs for a standard BitAxe.

Now let's be sensible and add some realism for a microcontroller to run the Oracle would probably be in the hundreds of cycles but again we'll be kind and say it takes 50 cycles! Now 50 cycles x 4 ns x 1 T inputs to the SHA256 engines = 200,000 seconds or about 2.3 days

I'll stick to the standard BitAxe, it will be way faster.

Allforprofit: RNS 15 Dec 2021

" Professor Gardin has entered into the Agreement in order to fund the payment of a tax bill.

The Agreement provides for MCS to buy 5,000,000 shares from Professor Gardin on 16 December 2021 for 3p per share and for Professor Gardin to repurchase the same number of shares from MCS on 16 June 2022 for 3.185p per share.

I doubtvery much he will ever buy them back.

Mickeydj: You post

" Method C's Oracle filters junk before the ASIC processes them (~4,166 hashes per cycle, but prediction is microseconds via ML model, not full cycles)."

If the prediction is microseconds,as you state, how can it keep up with the ASIC. If the control board is supplying quality NONCES, as you state i.e filtered, then the 240 MHz controller has to be supplying a 𝗧𝗿𝗶𝗹𝗹𝗶𝗼𝗻 per second.

I'll take issue with your statement that NONCES are sent to the ASIC. You want me to believe the designers of all hashing ASICs had the foresight to build in a future feature especially for QBT to send individual NONCEs. Gardin couldn't get versatility from the ASICs that's why he stated he needed to make his own.

NONCEs aren't sent to ASICs they are simply incremented in a counter on-board the ASIC. Sending a NONCE wastes precious time.

Did you use AI for you reply, cos it's wrong

Oldsharedog:

Ponder that while having your breakfast

From the last RNS:

"This software-only version of Method C is integrated directly into CGMiner and similar operating systems running on the control boards of Bitcoin mining rigs."

Looking at the BitAxe Gamma QBT have acquired, it uses the following microcontroller,

"ESP32-S3 is a dual-core XTensa LX7 MCU, capable of running at 240 MHz"

This must surely be the slowest device on planet earth. Consider the ASIC, the BM1370 hashing at about 1.2 TH/s.

Now while a slow processor could conceivably run methods A and B, since the controller only has to tell the ASIC where to start looking for a winning hash, method C is a different beast we are to believe. Method C has to match the speed of the ASIC in real time, deciding whether to hash the next NONCE or not for about 1500 individual SHA engines on the ASIC. That itself is a fantastic task and I'm not sure that would be possible particularly at a 30% rejection rate.

With Method C software Gardin wants investors to believe that a 240 MHz processor can do the same job and communicate with the ASIC to organise the distribution of asymmetric NONCES.

That is pure make believe

1 TH/s is 1 Trillion. To put that into perspective in one clock cycle of the processor the ASIC has carried out 4,166 hashes. To run even the simplest of simple programs on a processor would take dozens of clock cycles. The whole thing is preposterous.

From AI:

Assembly Language (minimal): A highly optimized "hello world" program written directly in assembly language with minimal system calls might execute in just a few dozen instructions and potentially a few hundred or thousand clock cycles on a simple microcontroller.

Allforprofit/Oldsharedog:

New low boys!

Oldsharedog:

To answer this question " " Why do some posters try and deliberately hide important information that would transform the interview to that which has been posted ??"

I'd say ask the person posting it. I only know the reasoning and motivation behind my own posts. Contrary to what others post I am not in any conspiracy in any company's shares.

These conspiracy/shorting scares are promoted by posters out of desperation, particularly to damage the credibility of posters making valid points. If you can't play the ball go for the man !!

Mickedj:

Brews previous posts have nothing to do with what I posted this morning. I saw nothing which suggested Valsecchi was employed for work experience.

Oldsharedog: You post

" Phimx - so I take it you won’t or don’t want to answer the question.

VERY TELLING."

You are assuming I read your question. In your previous post I read as far as bit where you accused me of "defle ting" and I felt compelled to reply.

Oldsharedog" You post

" Phimx - deflecting from what I asked is not an answer to the question - I will ask again and keep asking until you or someone else answers the question"

Of course it's not an answer to your question nor is it deflection

𝗜𝗧 𝗜𝗦 𝗔 𝗡𝗘𝗪 𝗧𝗛𝗥𝗘𝗔𝗗 𝗪𝗜𝗧𝗛 𝗔 𝗡𝗘𝗪 𝗦𝗨𝗕𝗝𝗘𝗖𝗧 𝗛𝗔𝗩𝗜𝗡𝗚 𝗡𝗢𝗧𝗛𝗜𝗡𝗚 𝗪𝗛𝗔𝗧𝗦𝗢𝗘𝗩𝗘𝗥 𝗧𝗢 𝗗𝗢 𝗪𝗜𝗧𝗛 𝗬𝗢𝗨𝗥 𝗣𝗢𝗦𝗧𝗦

Oldsharedog:

Here is something new for you

In the disparaging FT article Valsecchi is identified as working for QBT on LinkedIn before removing the reference to QBT.

Valsecchi is a band member with a very close relative of Rita Pizzi. It would appear very likely he was introduced to Gardin by Pizzi. Prior to QBT he was building Web sites.

The renowned physicist Gardin seems fond of promoting the status of his team.

Edit:

That should have been Oldsharedog's line

" And whilst watching you may also understand that someone speaking, NOT, in their native language is also thinking of the , erm, translation whilst talking."

Oldsharedog: You post

" Why do some posters try and deliberately hide important information that would transform the interview to that which has been posted ??"

Gardin has been speaking English longer than I have. As I've written below, he interviewed far better a few years ago. Advancing years affects everyone. Perhaps that will be his get-out from this project - failing health - then a slow shuffle off into the twilight dragging a sack of money.

Usersteve:

There is a Sharetalk, I think, with Gardin in 2021, he interviews far more fluently and sensibly than his recent interviews. You would think, at the very least he would rehearse his lines and delivery far more. Why isn't he prerecording his answers to perfection and editing them in with the questions.

Why would anyone invest thousands on this one man band?

Llwelyn:

I think I've been accused of being everyone except the mumbler-mumbler!

Mrhanji:

Your faith is admirable but how many updates do you need before the weight of evidence drowns you.

Giving further consideration to Gardin's "revenue after summer" is that not another way of saying "revenue in 2025" ?

Again Gardin has slipped in a " revenue" statement without qualifying where it's coming from. Surely after last year's "revenue in 2024" fiasco he should have learned a valuable lesson - don't make a statement without some solid basis to deliver.

The mumbler appears to have no control of his mouth or perhaps he thinks he can make these claims with utter impunity, after all he still collects his wad come what may. It absolutely defies belief that an intelligent CEO would make the claim unless it was a certainty. Since he hasn't announced the results of testing by these NDAs then it unlikely they have placed orders or contracts.

If I had to bet I'd go for Dementia!

Mrhanji: You post

" Anyways.. for me nothing materially changed, Apart from method C is software version now - FG guidance since June stands - revenue after summer (Method A &B) unless we get an update otherwise"

What kind of guidance is "revenue after summer" It's just another statement without any actual meaning, it needs qualification. The year 2057 is after summer.

Mumbler-bumbler isn't keen on being precise now is he ? His strategy seems to be "keep the punters guessing"

Although it pains me to say this I have to side with QBT on the subject of hash rate, to a certain point. It is not the absolute hash rate that matters but the increase in rate compared to an identical rig without the methods!

Gardin provides no sensible data or details in particular comparative data and testing conditions. Note Bitdeers latest SEAL 04 performance states something like 6 to 7 J/TH but in low voltage ultra power save mode but even those conditions don't tell us what the hash rate is or what the efficiency is when hashing in different modes.

Specifics specifics Bob !

What can customers expect .... who knows!