Quote from: cryptohunter on April 04, 2014, 11:07:24 PM

This person has a great understanding of these things, his skills in this area are unquestioned. So i personally would listed to what he is saying on this particular part.

If he tells me x11 asic implementation is easier that QRK i will take that as the case. Until greater arguement is heard.

So it would seem x11 is less asic resistant that perhaps scryptN and qrk.... the question is were the 11 algos in x11 blindly chained and could they theoretically lead to increased collisions?

Thank you for kind words, but just for record - I worked with very talented hardware engineers, learned a lot in the process, but newer did any HW project myself. Anything I wrote here is just an educated guess.
X11 is a bit simpler to implement in silicon, but I expect that seasoned engineer will not notice the difference. On the other hand, Quark will require less FPGA chips if someone will go crazy route with interconnected FPGA implementation

Regarding chaining - that's very theoretical speculation, even if there is 2^32 preimages for every hash value in Scrypt (or X11, or X1000), I know no way to exploit it...

Quote from: Spoetnik on April 04, 2014, 11:21:46 PM

show me a guy out there that woudln't sit on a gpu miner for 3 months privately while everyone else is stuck cpu mining.
see what i am saying ? i REALLY hope so everyone Wink

..AKA: Smelters gpu miner ring a bell guys Huh

?

Believe or not, I didn't mine on my GPU this year and only typed 2 (two) characters of crypto-related code. The core of Smelter - optimized groestl implementation - is published in sph thread. The other possible way of speeding up Quark (not implemented in Smelter) is finally made public by cbuchner1. So I expect sph-miner to become soon almost ideal Quark miner, without much room for optimization. And X11 contains 5 more hash functions to play with Wink

Quote from: cryptohunter on April 04, 2014, 11:07:24 PM

and one guy said earlier SHA256 is 100% optimized and then he said Scypt was not.. pure bullshit..
if there was some major optimization that could be done to scrypt it would been done already so don't feed me that bs to push an angle lol
i know i looked into it specifically actually and aside from some hashing checking optimization which did little i didn't see anything major..

There still is a possibility to speedup Scrypt on GPU (as used in Litecoin and implemented in cgminer, there are too many scrypts today

) - something like 20%, may be 50% with some luck. No one noticed misplaced 'if' operator Wink