List Commands Library for 50g
|
03-17-2018, 09:11 PM
(This post was last modified: 03-17-2018 09:16 PM by pier4r.)
Post: #285
|
|||
|
|||
RE: List Commands Library for 50g
For the speed, thanks for sharing. I imagined that LSHUF would have been costly on a large list, but I thought on small list (less than 100 elements) the difference would be negligible. It seems not from what you wrote.
Anyway before speed, come effectiveness. So I am testing the effectiveness now, whather it takes in time. For the "wouldn't it be better to test if the test works on a smaller input rather than waiting 8 hours for maybe nothing?" I agree. While I fall in this trap from time to time (damn me), I am slowly making progress so I already tested the test on one iteration with 4 elements building a list of size 6 (max 4096 possibilities). It took 25 minutes (I expected less, but I did not consider the LSHUF slow down as you wrote) and so from this I estimated 8 hours for 20 iterations. It worked as expected, or better, it worked but the results of LSHUF (1.1.0d) are not that encouraging. testing code Code:
21 iterations results (time is not the main focus) AttempsLSHUF: { 4096. 4096. 4096. 4096. 4096. 4096. 4096. 4096. 4096. 4096. 4096. 4096. 4096. 4096. 4096. 4096. 4096. 4096. 4096. 4096. } the one iteration done before hand is also 4096. So LSHUF 1.1.0d was never able to produce a solution before doing enough attempts to go through all the possibilities. AttempsRAND: { 2018. 1663. 1779. 4096. 2131. 370. 342. 2049. 2220. 4096. 2832. 822. 4096. 1305. 4096. 3202. 1997. 3936. 612. 3866. } The one iteration I had before returned 2574. Those made by rand are a bit more... reasonable? So if my test is not wrong, I would say that LSHUF 1.1.0d is not producing enough randomness (I am not sure if the sentence makes sense). Now I will install the last update of listExt and I will see if something change. The 50g can crunch for 8 hours more in the night. As I said it, test driven usage. More source code (although I think it is not needed), as usual in my assembla RPL repo: https://app.assembla.com/spaces/various-...cMathDir.s Wikis are great, Contribute :) |
|||
« Next Oldest | Next Newest »
|
User(s) browsing this thread: 6 Guest(s)