Zcash Testnet Beta1?


#1

I see a run from someone on Testnet Beta1 here :
https://benchmark.minezcash.com/index.php?title=Main_Page

git clone https://github.com/zcash/zcash.git
cd zcash/
git checkout v???


#2

Beta 1 to be released soon, likely this evening or early AM Those values are just from Zcash speed center https://speed.z.cash

And because you have to fill in some values into the table else it collapses in Wikimedia.

EDIT- Holy cow, the last test they ran is down to 15 seconds? From 45?


#3

Timesolveequihash -66.5%


#4

15 seconds is supposed to be the base minimum - at least for a certain generation of hardware. Was that an average or did they just get lucky on one test with some outlier-case input data?


#5

I'm guessing it's just a lucky run they are doing while testing the new wallet. It may go back up once finalized into Beta1, who knows what all they are tweaking on the back end.


#6

The benchmark site shows median data, and uses random nonces each time, so all it means is that over half of the runs had zero results (which results in the minimum solver runtime). If you look at the benchmark graph, you can see that even for the latest run there were some iterations taking longer:

https://speed.z.cash/timeline/?exe=1&base=1%2B9&ben=time+solveequihash&env=1&revs=50&equid=off&quarts=on&extr=on

EDIT: Here's the log output showing the individual runs; five of ten had zero results:

https://ci.z.cash/builders/zcashperformance/builds/152/steps/time%20solveequihash/logs/stdio


#7

If you look at https://speed.z.cash/timeline/#/?exe=1&base=1+70&ben=time solveequihash&env=1&revs=50&equid=off&quarts=on&extr=on , it appears that particular benchmark run just got lucky (or unlucky, if it found zero solutions, which is likely). There are things we can do to substantially reduce the variability of the Equihash solve time; I'm not sure whether or not we'll get to that before the launch.

Edit: the link is broken, but I intended to point to the graph showing the history of Equihash solution time including the quartiles and extrema.


#8

I saw the more recent results, must have been a lucky run :four_leaf_clover:. I adjusted the time on the wiki accordingly.


#9

As @str4d says, that's likely to have been an unlucky run that found zero solutions. The graph is somewhat misleading because the higher the solution time, the more solutions are found and therefore the greater chance that one of them will meet the difficulty target. (This will be mostly resolved, at least when multiple mining threads are used, by https://github.com/zcash/zcash/issues/1239 .)