4

Vertcoin comes with Adaptive-N-factor, which means the scrypt algorithm PBKDF parameter N will be changed according to the announced time table. Litecoin has N=10, while Vertcoin starts with N=11. And they claim the adaptive-N will resist the ASIC miner.

To produce an ASIC produce, it will need a huge amount of initial cost because of the prototyping (FPGA), HDL design, emulation, synthesis and layout, wafer fabrication, packing, and testing.

However, scrypt ASIC is now a real product (no matter how much it costs now). So re-target it to another new parameter should be not too difficult. I mean the whole architecture does not change (PBKDF is the same; only some parameters change, such as memory size and N), so the new ASIC development won't need to be started from the scratch up, such that cost is relative smaller than the initial product started from ground zero. Moreover, at the time of N adjustment at Jan 2016, the transistor size will be smaller and memory speed will be much higher, new technology will also enable the realization of the new ASIC miners for the new N factors.

If CPU/GPU can do the scrypt adaptive-N-factor algorithm at that time, I can say ASIC can also do that if there is a market and lots of demands. My idea is that, if vertcoin is very profitable (like or surpass BTC), Those ASIC company would also love to produce their ASIC miners with new N-factor. They could produce scrypt N=12 miner ready for the N-factor changes at Jan 2016, and announce for the pre-sale 3 or 6 months ago.

To me, adaptive-N-factor is just like one of BTC difficulty things. No one is mining BTC with miners whose hashrate is MHs or 10GHs or below, although all BTC miners are using the same proof-of-work algorithm SHA.

  • There is something wrong with your explanation, the cost savings described in Tony's comment is purely related to the difference in price between the GPU and ASIC ($100/$400). It has nothing to do with the additional memory cost which is the same for both platforms. Using this explanation memory requirements could add an additional million dollars to the total cost and the difference would still be 25%. – user15213 Apr 12 '14 at 18:03
2

The primary limiting factor is memory. As N increases, memory starts to become the most expensive component, whether's it's in a GPU or an ASIC. For example, let's say the memory needed to achieve a certain level of parallelization costs $800. Now a GPU chip costs $400, while an ASIC chip costs $100. So the GPU costs $1,200 overall while the ASIC costs $900. This represents only 25% cost savings, making people hesitant to buy a device which is useful for only one task and thus may be worthless in a few years (if GPU mining becomes unprofitable, they could always offload the GPU at half price or more to gamers).

To me, adaptive-N-factor is just like one of BTC difficulty things. No one is mining BTC with miners whose hashrate is MHs or 10GHs or below, although all BTC miners are using the same proof-of-work algorithm SHA.

Not true. Difficulty is a very simple parameter that does not affect the computations done; it merely is a threshold for determining which hashes are in and which are out. The N-factor actually affects the running of the Scrypt-N algorithm.

1

Flowertechnology is releasing 3 asic miners which can mine the new "scrypt-n" algorythm 1. july... It's been on pre-order for some time

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.