Your point is that there are 11 possible attack vectors instead of just one;
IF a collision attack can be applied to even one of the algos
IF collision attack compromises the full stack, as you say it does.
Demonstrate the part in bold. Everything else is proven and makes sense. Show us.
You're right, that is my point.
I just did... using crc32 as an example. You can create your examples using another collision attackable hash like say, md5.
phzi is right you plebs. It is statistically speaking less secure but it does have a benefit - lower power usage.
This is such a useless arguement. Has any "collision" attacks been discovered for sha256? If not, then it's 99.9999999% likely that there won't be any found out for Any other algorithms. Useless convo here...
lol.
phzi, don't even bother.
Heh, I resisted the urge to respond to this guy...
The thing about X11's lower power consumption, is that it's kinda a crock of shit. It only uses less power because the OpenCL code is horribly optimized. Properly optimized CL implementation will likely have similar power consumption to scrypt (or at least, the margins will be much decreased). We've already seen 40%+ improvements just by calling functions differently and not even modifying the cl code.