Potentially faster method for mining on the CPU

[jlcooke]

Note: working on Quine-McClusky now.    Always wanted a high speed lib for logic reductions ...

Alright, got Quine-McCluskey implemented.  Also implements logred_set_equiv(a,b) which goes through all possible input permutations and confirms "a" behaves the same as "b".

Q-McC is an expensive operation (Q-McC on r05 takes a full second when compiled with -s -O6), so I have not included it in logred_set_reduce().

It reduces r05 (the 6th bit in 32bit addition) from

Code:

r05 (terms=  1120, symbols=   8842, uniqueSymbols=12)

to

Code:

r05 (terms=   156, symbols=   1080, uniqueSymbols=12)

It has no effect on c00,c01,..c_i

Still don't think you'll find what you're looking for.  But have fun!

logred_v7.zip

[1714867166]

1714867166

[1714867166]

1714867166

[jlcooke]

But I still think you haven't quite grasped the magnitude of the P/NP problem yet! Then again, if the impossible ever gets done, it'll be as always, by someone who didn't know it was impossible.

Also botnet - If you've got the crypto-bug - exploring which input bits effect which output bits (1, 2 or 3 at a time) might be interesting.  This kind of differential analysis is what took down MD5 and cost me $10,000 (http://en.wikipedia.org/wiki/MD5CRK)

Eg:

Code:

SHA256(000000...0000) = A
SHA256(000000...0001) = B
SHA256(000000...0011) = C
SHA256(000000...0010) = D
SHA256(000000...0110) = E
SHA256(000000...0111) = F
SHA256(000000...0101) = G
...

then

Code:

A^B = ?
A^C = ?
A^D = ?
...
B^C = ?
B^D = ?
...

[smolen]

exploring which input bits effect which output bits (1, 2 or 3 at a time) might be interesting

I don't expect SHA2 to be broken by the ways discussed in this thread, but it's a good thing that someone tries it. A good result from such approach would be not analytical solution (seems impossible, but there always is a hope that equations system will collapse ), but representation of double SHA2 better (by operations count or height) that ones currently used in ASICs. I think I know how to chop 10% (or so) off from ~120000 binary operations, but let's hope that this thread will bring some fresh ideas

[jlcooke]

exploring which input bits effect which output bits (1, 2 or 3 at a time) might be interesting

I don't expect SHA2 to be broken by the ways discussed in this thread, but it's a good thing that someone tries it. A good result from such approach would be not analytical solution (seems impossible, but there always is a hope that equations system will collapse ), but representation of double SHA2 better (by operations count or height) that ones currently used in ASICs. I think I know how to chop 10% (or so) off from ~120000 binary operations, but let's hope that this thread will bring some fresh ideas

logred_v8.zip ready - faster computation of !a which is the most expensive operation.

r_i seems to have almost exactly 2^i-1 terms.  So the last bit r₃₁ will have 2³⁰ terms.  I've confirmed this with r_i i=0..12 and it's bang on.  The Quine-McClumskey reduction sure does a number on the complexity but it's not enough.

botnet - to take this further, you need to demonstrate that one of the input bits of the sha256 compression function does not pass through the Least Significant Bit inputs of the adds.

Or show one of the compression function output bits do not depend on the Most Significant Bit outputs of the 32bit adds.

Or that chained 32-bit adds have some reductions.

Cheers

[organism17]

hi! im kind of new on the subject and brings me to this post and i think that this is a very intresting discussion, any news about this? everyone disapear because they found the way? haha

I recently found that an algebraic fault attack could be effective to get a input of a sha256 .
Here is the link if anyone want to check.
https://archive.org/details/algebraic-fault-attack-on-the-sha-256-compression-function/page/n4/mode/1up

I hope you found something intresting in the papper and i hope to hear some news about you guys.

[NotFuzzyWarm]

This thread and the link you gave are from the very early years of BTC and were long ago proven to be irrelevant when compared to using ASIC's to just brute-force the process.

[kano]

Wrote code to do the code generation, and optimise the results automatically at each step, long ago with 40 cores of CPU and 512GB of ram.
Ran out of ram real fast

[BitcoinSoloMiner]

I did recently see that a company with quantum in their name had figured a method to more efficiently brute force/ solve a block, i think they had 3 methods and had completed testing to prove that their nethods were up to something like 30% quicker than existing block solving

What ive said here isn't quite accurate as i cant remember exactly what they found out or if its even true but its interesting... This kind of tech could be worth alot of money and i bet its extremely well protected. Imagine if one of the big pools or players were using this. "30% more blocks? Damn!!!"

Wrote code to do the code generation, and optimise the results automatically at each step, long ago with 40 cores of CPU and 512GB of ram.
Ran out of ram real fast

Think those new 96 core threadrippers pro workstations with 8 channel ddr5 2TB ram could handle it?  the ram is ECC too

[NotFuzzyWarm]

Think those new 96 core threadrippers pro workstations with 8 channel ddr5 2TB ram could handle it?  Grin the ram is ECC too

No it can't. Period.
There are no 'complex equations' or patterns to solve therefore applying possible mathematical shortcuts is not possible because there are none. What part of 'random' do you folks not get?

[BitcoinSoloMiner]

I found the article


https://www.voxmarkets.co.uk/articles/quantum-blockchain-s-new-bitcoin-mining-method-promises-to-slash-energy-usage-8f70e57/

Think those new 96 core threadrippers pro workstations with 8 channel ddr5 2TB ram could handle it?  Grin the ram is ECC too

No it can't. Period.
There are no 'complex equations' or patterns to solve therefore applying possible mathematical shortcuts is not possible because there are none. What part of 'random' do you folks not get?

i would like to see your thoughts on this

Quantum announced a new proprietary "Method C" for more efficient BTC mining, based on machine learning and predictive AI technology.
QBT
 said the new method was yielding consistent results, and had favourably demonstrated predictive ability in c. 30% of cases if an input to SHA-256 would produce a winning hash.

Quantum Blockchain has announced a new proprietary method of Bitcoin mining, this time implemented in hardware, promising significant energy savings compared to traditional methods. The company's "Method B" made headlines last year as it promised to increase the rate of successful Bitcoin mining by 2.6 times - entirely achieved in software.

The new Method C uses predictive AI to accelerate the cumbersome BTC mining process. Specifically, the technology tries to predict whether an input to SHA-256, the core algorithm for Bitcoin mining, is likely to generate a winning hash, or not. The underlying assumption for Method C is that an "oracle" decision is materially less computationally demanding than the SHA-256 calculation for the same input string.

[NotFuzzyWarm]

Quantum announced a new proprietary "Method C" for more efficient BTC mining, based on machine learning and predictive AI technology.
QBT
 said the new method was yielding consistent results, and had favourably demonstrated predictive ability in c. 30% of cases if an input to SHA-256 would produce a winning hash.

Quantum Blockchain has announced a new proprietary method of Bitcoin mining, this time implemented in hardware, promising significant energy savings compared to traditional methods. The company's "Method B" made headlines last year as it promised to increase the rate of successful Bitcoin mining by 2.6 times - entirely achieved in software.

Well how about starting a new thread on this in the Development & Technical Discussion area where it belongs ? Hijacking an ancient thread with off-topic information is just - pointless - and does nothing to expose your information to those who are better qualified to look into it and assess its viability.

[BitcoinSoloMiner]

Quantum announced a new proprietary "Method C" for more efficient BTC mining, based on machine learning and predictive AI technology.
QBT
 said the new method was yielding consistent results, and had favourably demonstrated predictive ability in c. 30% of cases if an input to SHA-256 would produce a winning hash.

Quantum Blockchain has announced a new proprietary method of Bitcoin mining, this time implemented in hardware, promising significant energy savings compared to traditional methods. The company's "Method B" made headlines last year as it promised to increase the rate of successful Bitcoin mining by 2.6 times - entirely achieved in software.

Well how about starting a new thread on this in the Development & Technical Discussion area where it belongs ? Hijacking an ancient thread with off-topic information is just - pointless - and does nothing to expose your information to those who are better qualified to look into it and assess its viability.

Its not off topic, the topic is faster bitcoin mining (with CPU), the guy posted earlier this year mentioning quantum and asked for some more info on literature, unlike you instead of saying its not going to work i gave the info to the (maybe) 30% faster solution. Also, Kano said he was attempted something and ran out of ram quickly. Maybe faster RAM and more cores (from the next-gen threadripper workstation) can help solve this seemingly high-resource-demand software problem...

You are confusing what is being said with mining directly on the CPU. I think here is a software solution to feed potentially better input to ASICs.

[FP91G]

I found the article


https://www.voxmarkets.co.uk/articles/quantum-blockchain-s-new-bitcoin-mining-method-promises-to-slash-energy-usage-8f70e57/

I wouldn't believe all the announcements in the news. There were a lot of such announcements, including from Intel, but mining equipment from Bitmain is the most widespread in this market and therefore the best. I remember only 1 case when a private company released an ASIC for Ethereum, which was very good, but the supply was limited.

[BitcoinSoloMiner]

I found the article


https://www.voxmarkets.co.uk/articles/quantum-blockchain-s-new-bitcoin-mining-method-promises-to-slash-energy-usage-8f70e57/

I wouldn't believe all the announcements in the news. There were a lot of such announcements, including from Intel, but mining equipment from Bitmain is the most widespread in this market and therefore the best. I remember only 1 case when a private company released an ASIC for Ethereum, which was very good, but the supply was limited.

I think this is a software based solution rather than ASIC hardware. With the advancement of AI and ML over recent years including things like chat GPT its certainly possible there is software which can feed better inputs to an ASIC to solve a block

perhaps now with bitcoin ETF and bitcoin becoming a bit more mainstream we may see companies outside of china/russia which step up in the ASIC game, would be nice to have a competitor to bitmain, also one that doesnt totally scalp their prices