Member Info for Phimx

Further evidence of ASIC optimisations!

In the paper "The Unreasonable Fundamental Incertitudes Behind Bitcoin Mining by Courtois, Grajek, Naik.

The footnote on page 32 states

" We call it improvement X and we do not give it a number in this paper as this improvement also concerns full SHA256 implementations in ASIC and has already been applied by most ASIC designers.

This paper c. 2014

So can the Oracle actually improve the efficiency of the latest ASICs. We know ASIC optimisations have been applied for a decade at least ! Gardin should be providing sensible figures instead of childish racing car analogies!

Try typing in negative analysis of QBT??

"Quantum Blockchain Technologies Plc (QBT) faces significant negative sentiment and performance indicators. Investor discussions reveal skepticism about the company's products and marketing, particularly on ADVFN. Additionally, QBT's earnings have declined by 33.4% per year over the past five years, and it holds less than one year of cash, indicating a weak financial position. "

At the end of the day my/Hexam observation that Gardin's irrefutable lab testing is incorrect has not been disputed by anyone. The result is glaringly simple to be refute.

It shows very simply Gardin elaborates what little is going on behind the scenesHe's been talking to the largest North Americal players for 2 years and not a single LOI with anyone. Nothing apart from NDA he says!

All show and no go

Koolkat:

You believe in AI generated drivel if you wish.

Throughout RNS 17 Jan 2025 3o% improvement is bandied about e.g.

"The Company’s board of directors believe this is a significant milestone, since this proprietary technology has been used in trials to mine Bitcoin with a competitive advantage against the same hardware without the AI Oracle implementation. The material competitive advantage in mining enabled by the AI Oracle may be achieved either by (i) reducing the energy cost of mining by approximately 30%; or (ii), accelerating the mining speed at current energy consumption and costs with approximately a 30% greater hash rate."

𝗕𝗨𝗧

We know this is incorrect. The Oracle, from what is claimed by Gardin, throws away 30% of inputs, so you can either save 30% on your energy bill by hashing the same number of inputs as an ASIC without the Oracle 𝗢𝗥 hash 43% more inputs for the same energy.

Gardin after all his elaborate and exhaustive lab testing by his world class experts, so we are led to believe, messed up this incredibly simple calculation. Thus I believe his team of experts don't exist or thorough scientific lab testing didn't.

As most AIM RNS what you see is what you want to see

I'm not saying they are a scam. I'm saying Gardjn isn't telling you how effective/efficient they are compared to current optimised ASIC solutions. This is seen in many AIM companies they trumpet certain characteristics without disclosing the negatives. Inspirit energy claimed Kw outputs for their waste heat recovery system without ever disclosing if it was electricity or heat kW.

These companies generally don't lie, that would be fraud, but on the other hand they either don't disclose negative news or delay it as long as possible. Look at UKOG for similar patterns.

Dooogle:

The only thing you Koolkat and Mickdj can respond with is claiming I write lies. Claim whatever you want, you seem to forget reasonable people here will read my comments in a different light compared to the childish drivel of your group.

What you have to remember is SHA-256 wasn't created for Bitcoin mining ! Therefore and not surprisingly optimisations can be made to speed up the hashing algorithm specifically for the Bitcoin input. Since the NIST implementation we could describe as general purpose then again unsurprisingly we should expect to find very specific cut down versions will be faster.

To belabour the issue, why is the focus on redesigning third party ASICs ? The Oracle is the heart of Method C, we know it would be about 4% of the logic in the hashing chip therefore a small ASIC in a slower process and obviously cheap could eliminate losing inputs before being sent to the ASIC. The hashing ASIC just sees a stream of inputs, it doesn't know they have been pre-sorted. We know from Foggy' assertions that the Oracle is small and low power and should be capable of feeding the faster hashing ASIC because it has less to do. Only a changes to a PCB and software would be required.

More lies to ponder !

More lies

The implementation of the SHA-256 hash algorithm in ASICs has been the subject of several studies. Xu et al. [10] presented a hybrid complementary metal–oxide–semiconductor (CMOS) hardware accelerator for both SHA-256 and SM3 hash algorithms designed in Semiconductor Manufacturing International Corporation (SMIC) 65 nm process node. Their system achieved a maximum clock frequency of 472 MHz by reutilizing logic blocks to reduce area and shorten the critical path. Dadda et al. [11] proposed a novel compressor architecture combining delay-balancing and pipelining techniques synthesized in ST 180 nm CMOS technology. The final design operated at 819 MHz and had an area of 164,856 µm2. Bai and Li [12] described an alternative compressor topology that used 7-3-2 adder arrays to reduce the circuit’s critical path by half. The final ASIC was synthesized in SMIC 180 nm CMOS process library and could operate up to 208 MHz with a 211,955 µm2 total core area.

Wu et al. [13] presented a highly parallelized SHA-256 hardware implementation designed in Intel 14 nm CMOS technology. Their circuit used three pipelines to shorten the timing path and divide the computation chain into independent steps. The final design operated at a maximum clock frequency of 1530 MHz and had a 14,273 µm2 core area. Li et al. [14] proposed an asynchronous SHA-256 accelerator synthesized in SMIC 40 nm CMOS process technology. The system utilized a cascade hash structure and operated at 227 MHz with an overall area of 38,885 µm2. Zhang et al. [15] described a new message expansion structure for full pipeline computation implemented in 28 nm CMOS technology. Their ASIC presented an area of 209,773 µm2 and could operate up to 446 MHz

From the Custom ASIC Design for SHA-256 link I posted recently

" Several optimisations configurations were tested and analysed during synthesis phase to enhance the performance of the final design"

I guess that's more lies which support my view that current day hashing ASICs are well optimised and Gardin's project has little to offer the big boys.

Dooogle: You posted

"They know whats coming , they cant bear the thought of QBT success .... and they dont like it up 'em Captain ! "

If I thought QBT had a commercial project that was heading for success I would be well invested. Alas, all the signs and evidence tell me its another perpetual development destined for eventual collapse. Already it has reached the stage where the spikes are too small for the gullible to get out.

The desperation is proportional to the insults posted !

Koolkat: You posted

" QBT have a clear vision and mission."

I think that is very debatable, I haven't called Gardin - flip-flop randomly, he's made up the story as he's gone along. He didn't appear to realise ASICs cost a lot of money to develop early on in the process. I think he gets most of his content from Google searches.

Koolkat:

I understand your anxiety regarding this poor investment choice you've made but there is no need to vent your lack of knowledge and frustration on me. If you cannot understand simple mathematical operations or more complex concepts then perhaps you should attempt some self improvement.

Having the good grace to admit your failings would be a promising start. My explanation of the 43% improvement will be obvious to many here and calling me a liar will simply present you in a very poor light.

Calling QBT is an opinion not a lie, it seems its not only maths skills you lack.

On a lighter note !

This document is interesting reading and shows how the man-in-the-street, given the wherewithal, could make his own ASIC for hashing Bitcoins, albeit not in the fastest of processes. Of particular interest is the small number of the logic gate used - a mere 27,863 gate equivalents. Compare to the mighty TMSC - 31,550 GE in a 7 nm process.

https://www.mdpi.com/2073-431X/13/1/9

Koolkat:

You simply cut and paste the QBT propaganda and seem unable to or question for yourself, that's not good for making money.

Maybe I've got this all wrong but from what I've seen and read my money is on this being a lifestyle company only. I'll happily let you make your own choice but you've been given a lot to ponder over many months I would seriously consider every post with an equal weight.

Insults don't make you money careful consideration may !

Koolkat: You posted

" QBT's Method C improves Bitcoin mining efficiency. (FACT!)"

But as I've said earlier 𝗯𝘆 𝗵𝗼𝘄 𝗺𝘂𝗰𝗵 𝗰𝗼𝗺𝗽𝗮𝗿𝗲𝗱 𝘁𝗼 𝗰𝘂𝗿𝗿𝗲𝗻𝘁 𝗔𝗦𝗜𝗖𝘀. That is a very important point upon which the Company lives or dies.

Gardin isn't saying and I think because it is with reference to the standard NIST algorithm and not the latest optimised ASIC hashers.

Koolkat: You wrote

" Phimx, If you believe your sums are correct that's great, but the program I am using is giving me a different answer. My advice to you is redo your DD. I cannot do it for you."

A program !

That is ... sorry I have no words.

As Hexam has stated 30/70. Let's put it another way. You have 2 identical FPGAs but one is having its inputs fed from the Oracle. The other FPGA I feed in 100 possible inputs and I hash them all. The other FPGA I feed in 143 possible contenders but the Oracle throws away 30% therefore 100 are passed to the FPGA to be hashed.

Both FPGAs hashed 100 headers in the same time and using the same energy but the FPGA + Oracle sorted 143 headers in 𝘁𝗼𝘁𝗮𝗹

You're in good company Gardin's team of world class experts had trouble with that stupendously simple concept.

Koolkat:

I'm beginning to suspect you are having trouble with the sum!

Stop Googling for an answer you can cut and paste and think man.

Koolkat:

It increases mining speed by 43% - allegedly!

If you can't calculate the exact figure say so and I'll get a 10 year old to explain it

Dooogle:

Just to be clear where I stand!

As with many AIM companies, I believe there is a project of sorts behind the hype 𝗕𝗨𝗧

The project has no hope of success, it is a pay packet for the directors only and add that to the fees Gardin is paid, a very nice pay packet. With all these project companies it's more about what is not said than said.

What is the gain of QBT's ASIC optimisations versus the latest ASICs. A very simple question, Gardin should be able to publish without giving away his trade secrets.