| ??? 09/10/07 16:12 Read: times |
#144323 - hummm Responding to: ???'s previous message |
Jez Smith said:
If you use an lfsr and a hashing function such as md4 then the result will pass diehard Well, but if you use MD4 then most probably you won't need the LFSR at all, a single counter will suffice (of course from a viewpoint of a FPGA fan(atic) a LFSR is less costly than a counter :-) ) Jez Smith said:
if you use an lfsr on its own it will produce a sequence of pseudo random numbers and then repeat If you feed the result of lfsr through md4, it will certainly repeat, too... :-) But I might have misunderstood you - maybe you had something else in mind, such as chaining the previous digest back into md4 or similar. Nevertheless, the simplicity of LFSR which makes it suitable for implementation in an 8-bitter will be completely spoiled by the order-of-magnitude-bigger complexity (and, more importantly, required data space) of md4. I don't think it is really necessary to have some very strong pRNG for the library. The point is, those, who will need stronger pRNGs, shall do their research anyway. And, a LFSR or a LCRNG is certainly more adequate enough for a nice christmas light chain... :-) JW |
