Topic: THE FASTEST MINER IN THE WORLD!!! (Read 498 times)

Sigh - we are talking about 2 different things here.

Bitcoin block header hashing is a double SHA256, but the ASIC internally doesn't do the first half of the first SHA256, and doesn't complete the second SHA256.
So saying that the ASIC does a double SHA256 is not correct.

I will admit that the way I wrote it the first time is a little misleading when I now reread it now, since I didn't mention the 2nd part of the process:
Then you are not talking about the double SHA256, you are talking about the internal rounds though the 64 step loop, that depend on the size of the input data, to get each of the 2 single SHA256 results, to give the final (incomplete) double SHA256 result (the 2nd one).

So in this case the ASIC does (almost) 2 times though the 64 step loop.

You verilog file says
So if what you meant by that was not 2 times SHA-256 hashing, but 2 times though the internal 64 step loop, that is only a part of the SHA256 hash, then yes that's (almost) correct (it's a few % less than 2 times), but if you mean 2 times though SHA256, then no that's not correct.

To explain it in simple terms:
You have a hash called SHA256
The code calculates the SHA256 hash (of course) of any input (of any length) you give it and it gives a single fixed size output.

But the process of doing that hash involves multiple steps, including one or more times though a part of the hashing code, an internal 64 step loop, and the number of times depends on the size of the input.
That internal loop is not a SHA256 hash, it's just a part of the SHA256 hash code.

If you can't stand a troll or two then you have no place in the Internet. Whats next, comments disabled videos in Youtube? Telegram Channel rather than group?

If you just rage quit when challenged, you are simply projecting a lack of online experience, which directly affects the image of your project and fundraising. If someone made a comment you didn't like, instead of degrading into personal attacks, you should have simply proved your arguments by pointing to the code and design merits. As you correctly said, open source talks by itself with code.

For what particular reason did you decide to attack members of the community instead of calmly prove your arguments? Did you read your own topic thread? You are claiming something extraordinary (for an industry with millions of dollars involved) that doesn't even exist beyond the design stage.

By the way, seeing the nature of your content, it starts to feels like it belongs more to the Announcements Tokens area, it is an ICO after all... If your project is legit, i see no reason for you to lose your cool in the mining speculation area, where anything goes regarding of "vaporware" products that might never even exist in the first place. Yes i know you started in hardware, but that is a mistake to do before actual hardware exists, and design doesn't count, at least a prototype reviewed by a third party should be expected to get serious attention from the community. If you have seen crowdfunding (both success and failures) you should know well.

If you leave the forum for good, you are just likely to be forgotten. And if your design has merits, the giants will "make it theirs" in no time, something the Chinese loves to do.

I added a nicely flavored garlic butter for you

The shame of this is very simply if he has a good idea that can work miners simply don't want to pre-fund  it.

His best move is talking to people like steve from Canaan or to the whatsminer crew of builders.

Both are good companies that want to succeed and if he has a good idea he could work it out with them.

As a longterm miner  I can say most pre funded asic projects failed for the miner/investor. So most miners don't want to pre-fund the gear.

As for good pre-funds with a new company

hmm

1 set of KNC worked out
1 set of Avalon did well

any others ?  maybe a Spondoolies can't remember

as for bad miner prefunds  the list is long.

And I just refilled my tub of popcorn

Sorry to see you go.
Did you ever get in contact with Steve of Canaan As he would truly be the person to talk to about your code.
Good luck finding a partner for your build.

Kano, you are a brazen troll.
At first you accuse me of my code doing two SHA256 rounds, and at the same time claim that it is not necessary.
Here is your quote:


And then, when I told you directly that you are ignorant about this, you went to check the information and realized that you could lose your reputation. And now you already claim that the miners perform 2 rounds. At the same time, again accuse me of being wrong?
Here is your new statement:


In addition - you are blind and do not know how to read. I wrote to you that Bitcoin hashing consists of 3 rounds of SHA256. But I did not write that the miners perform all 3 rounds. As you yourself saw in my code, which speaks about the level of my knowledge much better than me, there are only 2 rounds of SHA256. However, they are not complete. In my chips, hashing begins with the 65th cycle (step), because at this step in the message comes the timestump, and ends at the 188th cycle (step), because further hashing does not make sense. The reason that hashing in my chips does not start at the 67th step where nonce comes, but at 65th, because my chip goes through all the nonce in the specified timestump range to reduce the number of data transfers between the chip and the control device. That is, the number of iterations is determined by the formula: F = 2 ^ 32 * (x - y), where x is the end point of the timestump, and y is the starting point of the timestump.

From all of the above, I conclude that you are a brazen troll and a fool. You do not like the truth and do not admit your mistakes, but consider it your duty to call a "shit" anything that your brains cannot understand.

TO ALL.
Trolling and rude attitude of users of this forum got me. I don't want to be here anymore. If this message will be read by a reasonable person who is interested in our developments, in the topic there is a link to our website, and on our website in down is a link to our telegram-channel. There you can always find us.

I'm leaving. You can continue to transfer feces here between each other, which many of you love so much. I definitely have nothing to do here.

I would never invest my money to a guy like him.

Telling Kano he doesn't know how mining works is like telling Sakharov he doesn't know how fusion works.

It also needs a source of work from a bitcoind to hash, convert that source of work, by doing the first hash cycle of the first 64 bytes of the 80 byte block header (64 steps) to generate a midstate/work item, and when it finds a share meeting the requirements of the work sent to it, it needs to send the output to a bitcoind (with all the transactions that make up the merklehash)
NONE of which the ASIC does.

The ASIC does not do 3 rounds as I explained above, it only does 2 (or to be exact a little less than 2)

You complain about me not watching you silly video or reading your white paper on some other web site, yet I've written here above EXACTLY how mining works and you don't bother to read or understand it, and thus get your reply WRONG.
Your a joke. Full on.

One round of SHA256 is 64 cycles (steps as you write).
Bitcoin hashing requires 3 rounds of SHA256 with 64 cycles = 3 * 64 = 192 cycles (steps). If you do 2 cycles = 64 * 2 = 128 steps, you will not get a Bitcoin hash.


Criticism is when pointing out errors and shortcomings of the system. Here no one criticizes us. Here our work is simply called "shit" and "garbage."

Picking a fight with the community isn't going to earn you their trust, that's for sure... You should have expected criticism and doubt from the start. This IS a space often abused by scammers after all...

LOL

The mining software (cgminer) does the first half of the first round, since it doesn't need to be repeated for the 4 billion (2^32) different nonce hashes, since they all have the same first half of the round one result.
This is called generating work for a miner - the midstate.

The work/midstate is procesed by the mining hardware, doing the 2nd half of the first round - equivalent to a 64 step full SHA256 - completing the first half of the double SHA256 - since the input is 80 bytes and the SHA256 64 step hash only processes 64 bytes at a time - so is processing the remaining 80-64 bytes.

So that's, so far, ONE time though the 64 step hash loop.

It then does a partial final round - equivalent to less than a 64 step full SHA256 - almost completing the second half of the double SHA256 - only once through since the input is the 64 byte output of the first round, and the incomplete hashing explained quite well at those links of my posts in August 2011, why it doesn't need to complete it

So that's a total of slightly less than TWO times though the 64 step hash loop that are done inside the mining ASIC.

Now since I've written more mining and pool code than you'll ever do, and there are VERY FEW who have written more mining code than me, and I'm one of the secondary developers of the piece of software that is in almost every miner ever made, and if you did actually make a miner, that software would be in your miner also, I'm pretty sure I do know what I'm talking about

So since you don't know that this is how mining actually works, I wonder how you could even design a miner correctly

I looked at the white paper and the content is full of rubbish.

Mr. lover of "shit", you are so ignorant in mining issues that you do not even know that Bitcoin mining requires 3 rounds of SH256, and not 2 as you claim. What can I talk about with you?
You as a person are not worthy of me explaining something here. I explained everything in the official documents of our project and if you cannot understand what is written there, then your intellect is not suitable for this work.
I stay with my losses, and once again I ask you to leave our topic, or do you poorly understand English?

You clearly know nothing about Canaan when you make that statement.

As for having to waste my time watching some video someone made that has who know what in it, rather than supplying a few lines of text to explain the difference, well that's your loss

Edit: looking at the white paper and reading 15 pages of rubbish, in the middle says you claim doing serial implementation is faster than parallel implementation for "add"
I do clearly know that the one single function "add" is the slowest part of the whole implementation.
(I alas worked that out only back last November after spending three days trying to wrote a new optimisation of the SHA256)

You can't do it in a single clock, but you can do it in fewer clocks if you break it down to multiple 8 bit adds (i.e. 4 times the clocks) and use a lot of transistors.
I don't think the number of transistors is that much of an issue ...
(and I do wonder why the SHA3 removed the "add" - maybe to make it easier to hack?)

But again, you state a whole bunch of numbers that you'd need someone who has already fully implemented it to verify that is possible and you didn't just make it all up
Or you need to prove it by making a miner
Read sidehack's posts ... again ... if you did before

Edit2: In case what I said above wasn't obvious, your white paper starts with pages and pages of crap ... oddly most people would stop well before getting to the parallel vs serial discussion.

You do not have enough intelligence to watch an explanatory video and read the Whitepaper, where everything is explained as for children. Therefore, you are destined to ask stupid questions for a very long time, hoping that I will explain to you on your fingers what is what.

TO ALL
All the questions about why we are not running into Bitmain or Canaan, the answer is very simple - we wanted this technology to be available to the ALL crypto community. To reduce network centralization. But if the community insists that they want to be slaves and hostages of several monopolistic companies that the miners are building, then you can even be sure that sooner or later we will do that. YOU do not leave us no other choice.

No, no one here would bother to test it, or probably wouldn't even know how to test it.
But that's not necessary ... as explained further below ...

--

But firstly, that's not what I've been posting, that your other pseudonym has been arguing with me about.

He keeps telling me that GMO made miners in some way to think that will make his verilog code seem better?
GMO did not. So he keeps stating bullshit.
So of course that means I'll not trust anything he says.

--

'Your' verilog code says it does an SHA256 double hash.
But I don't know enough about verilog code to say if it really does, or if it does anything special that anyone else would bother to look into it.

Every miner manufacturer that has delivered miners has already done the full process of going from idea to miner production.
What is in your code that makes it better?

I can say a few very relevant things:
1) What is special about your SHA256 hash that means it's better than any others?
I do completely understand (more than most) what is involved in the actual hashing algorithm, since I have written code myself that analyses the (2nd) sha256 and optimises it and then generates the optimised code (in C)
Before that, in Aug 2011, I posted details about the first well known optimisation (that I independently came up with), that I found someone had already worked out and it was already being used in all GPU miners.
The following post and the 2nd post correcting it:
https://bitcointalksearch.org/topic/m.454564
https://bitcointalksearch.org/topic/m.455128
Fun fact, it would seem that Intel recently got a patent on that

2) There are many who have done the SHA256->Chip already
So are you saying you are able to do this process way better (i.e. nothing to do with SHA256 optimisations) than anyone else who has done it before?
i.e. if you showed this to Bitmain or Canaan they'd jump up and down and say "Yes we want you to do our verilog code"?
That's pretty easy for you to find out since you have already released it
(with a copyright - though Bitmain is known for ignoring copyrights, licenses and patents )
Go ask Canaan.

So either you are saying:
1) you have a great optimisation no one else is using
or
2) you are able to translate to verilog better than anyone else who makes miners
or
Both 1) and 2)
Which is it?

3) Now for the continuation of building a miner, I also posted this a number of years ago:
https://bitcointalksearch.org/topic/optimal-firmwarehardware-design-for-mining-with-cgminer-294499
While some of that may seem old with regards to how mining is done, it's actually still relevant since all new miners come with controllers that are just a simple computer running a modified version of cgminer.
Have you got that far yet? Have you implemented everything I mentioned there?

4) You say it does a double SHA256 - but no miner does a double SHA256
They only do a partial single SHA256, since that's all that's necessary, the first hash is done once by cgminer when it generates work for the mining hardware, that the hardware then hashes another 4 billion (2^32) times with the changing nonce.
i.e. it's WAY faster than doing the double SHA256 every time.
Have I missed something in your descriptions about this?

The code means nothing zip nada.   

Wafers of chips have meaning.

Here is why lets say guys like kano or ck check the good and it is good code.

The op has said it is open source. So the race is to build chips if it is good.

For a good coder to check it and say wow it is brilliant unreal  a new way to mine 10x the speed and 10x less power.

A large company would look at open source and build chips. Asap

Bitmain to have a chip in under 60 days.

Well? Can any of the noble bitcointalk knights test the code? Or will they continue to throw poop?

Well it happened I could look up threads under group buys and find it.