Quantum Blockchain Techn... Share Discussion Threads

[iceagefarmer] 09/12/2024 09:04

i see RKBEEKEEPER up early today on LSE..multiple posts..suggests news coming,just hope its not a pump

[anddownitgoes] 08/12/2024 21:30

As per the QBT website:

Method C has been designed as a SaaS (Software as a Service) client-server application. The QBT server, where the Method C engine runs, updates the key parameters on the client side of the AI model.

So will be a SaaS. Will have to be as near zero latency as possible, on top of making an assessment of the header as instantaneous as possible. It is a big ask. Will presumably have to have the servers sited onsite with a miner or as close to their data centre as possible as well as being top of the range hardware.

Wonder if this is even part of their testing atm.

[ellipses] 08/12/2024 21:20

Not a problem. They are writing a software version. Also they are using an ESP32...




That's got oodles of RAM and storage and goes really fast.

ROFL

[anddownitgoes] 08/12/2024 21:13

While I'm sticking the knife in their Method C product description bothers me:

Method C performs a real time assessment of different header’s parameters before submitting them to SHA-256 and decides whether they are worth processing, or can be skipped. This approach avoids processing a large number of hashes, obtaining the target result, u.e. the a winning hash, in less time.

So as I suspected they need to assess each header and then hash the promising ones. So time taken for every 100 inputs is 100 assessments + xx hashes. So time taken to assess is critical and is a time overhead for the promising inputs.

[anddownitgoes] 08/12/2024 21:05

Yeah it is quite significant imo. Significant as someone has taken the time to reword their claims in such a way so they are not so definitive.

Presumably with the Bitaxe they have now started to do tests with a real rig and even on lower difficulty their claims are being rolled back.

Depending on what sort of figures they are getting now perhaps their focus will be purely on Method C going forward.

[ellipses] 08/12/2024 21:00

That's an interesting one in as much as they no longer give guesses as to numbers. Presumably Method C will similarly lose its 30% to 50% improvement.

What chance they found their original guesses were wrong because they had not managed to get it working on commercial hardware but now they have done so on a single BitAXE ASIC, AKA *Real Mining Hardware* they are no longer so sure.

Materially Improve might be doing a lot of hard work here.

[anddownitgoes] 08/12/2024 20:48

Checking the last modified date I get early September. That is when the change to the IT Infra page was made.

Really it's a bit pointless reading into the meaning behind the words used. Whoever updates the website does not have English as their first language or access to a spellchecker.

Interesting though that on the products page they have rolled back on specific claims for Methods A and B.

Method A was 'can improve the ability to find a winning hash up to 10%' and is now 'can improve the ability to find a winning hash, compared to current blind search'

Method B was 'can improve the ability to find a winning hash by 2.6 times, compared to standard search, while saving up to 4.3% of energy' and is now 'Method B, can materially improve the ability to find a winning hash compared to standard search, while saving, at the same time, energy'

[1knocker] 08/12/2024 16:43

Iceage, I see you are now taunting QBT with quotes from FG's infamous Miami interview!

Welcome aboard the bear faction. All heaven rejoices when a sinner repents.

[iceagefarmer] 08/12/2024 08:36

27/6/2024 Interview:

"We announced last year that we are heading for revenues this year [2024]. We still havn't announced the opposite. In the Accounts, one has to be prudent. Any word has to be agreed with the Auditors...[but]...we still havn't announced that we will not generate revenues this year [2024]. So I will make it clear - with all the contracts we have, with all the discussions we have, with all the contacts we have, with the development of our technology, how it performed from in the lab into products, what is next for products - we [are] still on track for this our end goal."

[anddownitgoes] 07/12/2024 21:45

This bit is good as well:

The IBM QISKIT quantum computing facility, available on the Cloud, and on our IT platform, allows for the simulation of several quits, for the initial testing of the proprietary quantum algorithm.

ChatGPT clears it up:

I believe there might be a misunderstanding. "Quit" is not a common term in the context of quantum computing. It's possible that you may have meant "qubit," which is the fundamental unit of quantum information.

Good to see QBT have a good eye for detail

[anddownitgoes] 07/12/2024 21:42

Bit misleading Ellipsis, the page clearly says they have commerncial Bitcoin miners with hundred of ASIC mining rigs.

Can't be clearer than that.

[ellipses] 07/12/2024 20:26

Has LSE stopped taking commentary or have the Bulls abandoned their facile dream?

Maybe QBT need to update their "We've got lots of CPU and GPU cores" page to mention how many MDK boards they are running.