RE: Just 3 weeks for Multi-Bagging News....9 Apr 2026 09:27
Let's be clear about headers:
The header is 8 bytes of data or a 640 sequence of bits, representing the current Block to be mined. Generally, only 2 sections of that bit stream change the NONCE and the extra-NONCE. ASICs take the header with all the bits fixed except the NONCE. The NONCE is 32 bits and represents, in decimal, a number which can change from zero to approx 4.3 Billion. The ASIC begins hashing the header beginning with the NONCE at zero and produces a hash value at its output. The NONCE is then incremented to one, all other bits remaining at a fixed value, another hash is ouput. Because the ASIC has multiple channels of identical SHA256 circuits, perhaps 2,000, each incrementing a part of the 4.3 Billion total range of the NONCE and being clocked at hundreds of MHz the ASIC produces hashes at a fantastic rate 1TH/s or easily more. The hash boards also contain many ASIC chips so the rate it can increment the NONCE is phenomenal. After just milliseconds or less the NONCE counter will flip over to zero and the same sequence of hash outputs will pour out once again, totally pointless. At this point the extra-NONCE, which along with all the other 608 bits of the header has remained unchanged is incremented. The NONCE is now incremented through its 4.3 Billion range again. The header having changed by the increment to the extra-NONCE.
So, during a block hash, the header remains fixed for 4.3 Billion increments to the NONCE and then the header changes slightly every 4.3 Billion hashes as the extra-NONCE changes. The remarkable thing is a slight change to the header produces hash output a totally different hash output which may be anywhere in the range from zero to two to the power 256 or as normally written 2^256.
That number is so impossibly large it is difficult to describe.
So now wr have a good grasp of what's happening, we can appreciate Gardin's story of using an ESP32 to assess each NONCE variant of the header was total fiction, it would have to work at least a THz/s rate. So he changed the story to the software now assesses an entire header and the variants in one go and somehow does this without looking through the 640 bit pattern.
The software having the time it takes to assesses a header with a new extra-NONCE increment in the time the NONCE increments through its 4.3 Billion possibilities I.e. milli-seconds. But with hashboards now containing tens of ASICs the assessment time will shrink to micro-seconds and once again the process is unbelievable
Sign In
Follow the stocks
