One thing I've realised. All of the AES candidates are based around encryption from a key. If we go for a quark-style hash system, then if we are being secure, we need 14 different keys.
Something for donators to suggest?
Matthew:out
Should I start a donation fund?
If you want to. I'm not forcing you to.
One thing I've realised. All of the AES candidates are based around encryption from a key. If we go for a quark-style hash system, then if we are being secure, we need 14 different keys.
Something for donators to suggest?
Matthew:out
Are we doing the random hashing algos too? (like first sha, then whirlpool, then md5, then random, then random again)
As far as I can work out, each pass of Quark's "random hashing" simply checks whether the least significant 3 bits in the second most significant 32-bit word are zero. If they are, then it runs one hashing algo, else, it runs a different one. It screws PGO massively.
I like your idea, but I think we can make a different scheme which makes optimisation even more awkward.
Some pseudocode:
void Hash(int* input, int* output)
{
imt mask = 8;
int i = 0;
for (; i < 8; i++) {
switch (input[5] & 8) {
case 0:
Rijndael(input, input);
case 1:
Square(input, input);
case 2:
Serpent(input, input);
break;
case 3:
RC6(input, input);
case 4:
MARS(input, input);
case 5:
Twofish(input, input);
break;
case 6:
Twofish(input, input);
case 7:
Rotate256(input, input, 3);
}
memcpy(...);
}
That ought to make GPU mining difficult. With up to 24 rounds of pseudorandom hashing algos, and a rotate in there for good measure, it may well make CPU mining difficult, too.
Matthew:out