From sandy valleys, up to sand dune peaks. Kaktoos rises to the challenge!

Message boards : News : From sandy valleys, up to sand dune peaks. Kaktoos rises to the challenge!
Message board moderation

To post messages, you must log in.

Project administrator
Project developer

Send message
Joined: 15 Jun 20
Posts: 5
Credit: 45,268,055
RAC: 0
Message 479 - Posted: 6 Dec 2020, 20:05:33 UTC

After a somewhat long hiatus, Kaktoos is coming back for another round in the ring, and this time at higher and higher heights. Many of us who contribute to the project (Minecraft@Home) got a little busy with school/work after the Pack project concluded, so I apologize on everyone’s behalf for the silence and uncertainty as of late.

As for where we left off, the y=62 search wasn’t as fruitful as we would have liked. No taller cacti had been found at this height than were found at the slightly higher height, y=63.

With the way things usually go, we would need to create an entirely new version of the program in order to run kaktoos for different heights. We started on y=63 for the original version of the program, and then changed this to y=62 for the second run of the program.

This time around, we’ve flipped the script, and different heights can now easily be run at the same time or in series, with 0 program changes. Effectively we’ll be able to get the final answers we’re looking for from the kaktoos project much easier and without as much “red tape” as there usually is with program changes due to security reasons.

What does this mean for you?

This means more kaktoos, more often, and with less breaks between runs for the foreseeable future.

There’s some limits to how many heights we can attempt since there comes a point where you simply will not find any cactus, however, we have a fair range to work with now with these new changes in play.

What now?

We’ll be launching a run of kaktoos very shortly. We’ve been testing this in beta with a height value of 64 for a short period of time to make sure that things are running as smoothly as we would like, but we’re happy to announce that we’re ready to go!

Some have offered far more support than we ever expected during this beta, which definitely surprised us! We’re grateful for everyone’s enthusiasm, even if it’s to our own disadvantage to have this run done so quickly.

For the curious, this new run will encapsulate heights of 65, 66, and 67.

That said, happy crunching! Kaktoos runs will begin soon outside of beta. Thank you all for the support and understanding over the last couple months as we work through things.
ID: 479 · Report as offensive     Reply Quote
Profile neilcaffrei
Project administrator
Project developer
Help desk expert

Send message
Joined: 17 Nov 20
Posts: 17
Credit: 6,666
RAC: 0
Message 480 - Posted: 7 Dec 2020, 13:39:13 UTC

As you might have noticed, the 3 heights we chose has already been done. But fear not ! We already generated new tasks for y 68-(nice)-70, we will in a few hours pregenerate all the other heights till at least y100 so everyone can have tasks for the next 2 to 3 weeks (according to the current rate).
ID: 480 · Report as offensive     Reply Quote
zombie67 [MM]

Send message
Joined: 24 Jun 20
Posts: 21
Credit: 181,142,041
RAC: 0
Message 483 - Posted: 14 Dec 2020, 2:12:59 UTC

I see the queue of unsent tasks is dwindling. Is that all of this batch? Or does the queue just need to be re-filled?
Reno, NV
ID: 483 · Report as offensive     Reply Quote

Send message
Joined: 24 Jun 20
Posts: 1
Credit: 13,604,911
RAC: 0
Message 485 - Posted: 15 Jan 2021, 2:39:22 UTC

No new tasks since a month now. Is this project still alive?
ID: 485 · Report as offensive     Reply Quote
Profile Hy
Project administrator
Project developer

Send message
Joined: 15 Jun 20
Posts: 74
Credit: 19,537,761
RAC: 0
Message 489 - Posted: 24 Jan 2021, 18:09:34 UTC - in response to Message 485.  
Last modified: 28 Jan 2021, 3:58:15 UTC

Hi, in terms of the BOINC side of Minecraft@Home, we haven't used it as much as we expected to. But, we have more cactus-searching (Kaktwoos) tasks coming up as we have gone through enough of the previous data and last task's to do another run. Others are on break from finding various world seeds with code that was optimized so well that we didn't need BOINC.

What ran in December was Kaktoos, which provides all the potential cacti required for our multi-gpu and much more intensive Kaktwoos application. Kaktwoos Tall-Cactus searching has always been our long-term project, and we project to do runs for layers 64-75 after final adjustments of code on our Github page. Github made some changes to their compile environment, and we spent time to do debugging of the OpenCL code so the BOINC server could do much easier changes of where/which layers we are searching. Running for layers above 75 is possible, but the likelyhood of any tall, special cacti (except for in Superflat worlds) is 'essentially impossible'.

Each layer produces around 120,000 tasks that take around an hour on a mid-range gaming GPU (GTX 1060/RX 580), not including confirmations and re-tries. So, the long-term 'final runs' are being prepared and I hope you shall see them in your client, starting with a y=64 'test' run. There are many people behind the scenes involved that have the tools and knowledge (that I don't) to make these runs possible (interpreting the data and converting it into a BOINC-friendly format which we create jobs from) that can be another bottleneck at times. I have been asking and organizing ourselves over the past week or two as the other seedfinding projects yet again caused some distractions.

Y=63 was our best, as we found 20-22 tall naturally generating cactus, but no 23/24 aside from in Superflat
Y=62 found additional Superflat 23/24 tall, putting our 'cactus limit' at 22:24.

With any luck we can smash either of those records, but we must have two chunks with tall cactus at the *same block level/layer* to stack successfully.
ID: 489 · Report as offensive     Reply Quote

Message boards : News : From sandy valleys, up to sand dune peaks. Kaktoos rises to the challenge!