Differences

This shows you the differences between two versions of the page.

--- mining [2020/08/19 10:53] – boot2thrill
+++ mining [2021/01/05 22:20] – [Custom PoW Algorithm - Cryptonight Conceal] boot2thrill
@@ Line 18: / Line 18: @@
 To fairly reward the mining effort we will start with **5 CCX per block** and increase it by **0.25 CCX every month** until we reach **15 CCX per block** (max reward in 3 1/2 years time).
-==== Custom PoW Algorithm - Cryptonight Conceal ====
+==== PoW Algorithm - Cryptonight-GPU ====
-Miners are so important to us that we invested a considerable amount of time and resources developing the best mining algo possible — Cryptonight Conceal (CN Conceal).
+Miners are very important to us.
-CN Conceal is a variant of the original Cryptonight mining algorithm and is designed to achieve maximum PoW hash function for egalitarian CPU & GPU mining and ASIC/FPGA resistance. Conceal is the most energy efficient of the FPGA-Resistant algos. It uses a scratchpad of 2MB with Floating Point Math (FPM) that makes programmed FPGAs inefficient. The FPGAs don’t have any advantage over high-end GPUs. Another main advantage of CN Conceal is the performance efficiency across multiple generations of GPUs. It is extremely efficient on all GCN architectures from multiple cards across the last 8 years. All these qualities allow CN Conceal to be one of the most decentralized mining algos in the entire crypto space.
+Conceal Network implemented the Cryptonight-GPU mining algorithm, developed by Psychocrypt and Fireice_uk.
+Cryptonight-GPU (CN_GPU) is an ASIC, Botnet, and FPGA resistant mining algorithm that assures a fair GPU based supported network.
+CN_GPU relies primarily on GPU core compute performance and is the first mining algorithm that is specifically designed to use floating point math at a single-precision (FP32).
+Floating-point operations per second (or Flops), represent just how many computations (additions and multiplications) can be handled per second by a GPU. FP32 refers to the level of accuracy of those numbers (and is the number format).
+A CPU generally consists of 4 to 8 CPU cores, while a GPU generally consists of hundreds of smaller cores. Since ASIC, Botnet, and FPGA builds rely on specialized low memory CPU computation, a mining algorithm based around GPU core compute performance renders them unprofitable.
+GPU performance with the CN_GPU mining algorithm relies on physical core count and its clock in Mhz, making AMD and Nvidia GPUs very similar in performance.
+The goal of the CN_GPU mining algorithm is to assure fair GPU mining:
+AMD and NVIDIA GPU lineups are equalized.
+Low-end and high-end GPUs are measured based on core compute performance (Flops performance).
+To eliminate any CPU and Botnet mining advantage.
+To force ASIC or FPGA design to be cost-prohibitive.