Posts by Hy

1) Questions and Answers : Windows : no new work? (Message 760)
Posted 8 Jun 2022 by Profile Hy
Post:
Hi Plejaden,
Sadly, there are no current plans to revive our full BOINC system due to lack of "reasonable" work to have it carry out now, and for reasons mentioned in my last News (and forum) post:

https://minecraftathome.com/minecrafthome/forum_thread.php?id=136

As Minecraft 1.19 has been released, if there is increased interest, time and knowledge about the changes to Minecraft (and how to quickly work around or with them with code) there may be discussions about what potential projects may be created, or revived.

The worlds in Minecraft are much different now both in code, "infiniteness" and complexity with the latest two updates, though within our [email protected] Discord there are still smaller projects carried out by our coders, Minecraft fans and those with the resources to run the checks and calculations that bring us cool things such as the OneChunk Speedrun Seeds, and totally broken 12-eye End Portals as you may have seen over the past few months and year
2) Message boards : News : All GPU & CPU (Onechunk SSG + TrailerCrack) tasks completed + current research projects & future of [email protected] (Message 738)
Posted 7 Dec 2021 by Profile Hy
Post:
Hello all BOINCers and [email protected] fans,

As a preface, would like to subtitle this post the 'Beginning of the End (...of the Year)' update. You'll see why soon. This is a long post, and I hope all those BOINCing with us at [email protected] will read and see this.

First, I shall give everyone an update on the events of the past two months since the launch of Trailercrack and the OneChunk Speedrun seed-finding project. All tasks have been completed as of Mid-November 2021, and all results have been collected as of December 2021! We still must go over the final results, or any missed results for both projects, but safe to say all your efforts and data are now collected on our servers, and all BOINC computing is on 'break' for the month of December, and possibly into mid-January.

Now, we have tried to stay to our promises of a 'Minecraft Trailer' 10th anniversary recreation video, which you can view our announcement of below. As said by Tomlacko, one of our major Discord administrators and contributors today, "The full remake, including the reveal of all seeds, will come out later when we're finished! (This might still take us a month or so, we'll see)".

Trailer Remake Preview: https://www.youtube.com/watch?v=kox1ZR8xJ5E

Additionally, we have an announcement regarding the results of the SSG Project (OneChunk). For a limited time, [email protected] held the Enter-End World World Record of 17.5 seconds! All thanks to your computing power allowing us to find an insanely rare, easy to access End Portal seed. After loading the fresh seed, we literally were able to dig down and jump into the End Portal (one of the biggest challenges of Minecraft speedrunning). I would would call that a major time save and success. You can view this Speedy Steve in these videos:

Official Speedrun announcement: https://www.youtube.com/watch?v=HtPZIHqGYXo
Speedrun detailed by 'TheMisterEpic': https://www.youtube.com/watch?v=1Gvgf5cAAGY

Or, retrace our steps in the seed: -5362871956303579298 with Minecraft Java, Version 1.12!

I hope you all understand that there shall be some downtime while our team investigates the major changes within the Minecraft 1.18 'Caves and Cliffs Part 2' update. Some of which are critical to the now more difficult future of Minecraft seedfinding, but also brings us some hopes of wonderful new worlds with the drastically different and expanded world generation that it brings. These changes were revealed to us with the expansion of the seed-space (how many unique seeds that exist) from 44 bits (48-4 bits) to 60 (64 full bits, minus 4 again). In simple terms, this means we now have 65,536 times more work to do to find a single 'lost' seed. That's many thousands of times more possibilities than before.

Those familiar with our previous BOINC projects, or current ones may know the complex relationship between computational time, computational workload, possible acceleration of work and the seedspaces we must work with, or constrain for to complete our projects. The Minecraft 1.18+ RNG (Random Number Generator) algorithm itself has changed to one called 'Xorshiro' from my quick review of the latest discussion chats, and a method to 'skip' over full-RNG computation for our future projects has already been theorized and possibly in development.

It also appears that this new RNG is fairly easy to calculate beforehand, with results small enough that it is possible to create into a few-kilobyte sized lookup table that our computing devices can quickly compare the input and output results from. One may scheme that with newer, larger GPU or CPU caches and cores (See, RTX 4000/AMD RX 7000), these LUTs can be further massively distributed across the device to help provide some of the speedups we shall need to reasonably compute more results within this 60-bit seedspace. Our projects often find, or have weird, super fast optimizations that can turn multi-GPU projects into more solo, or CPU based tasks depending on the dedication, time and luck our [email protected] geniuses have during our projects' lifecycles.

This leaves us to the remaining tasks on hand for us to work on over the next month or two, which are:
1. Post new SSG results (One new possible candidate from Mid-November 2021 found)
2. Trailercrack trailer results (for the video thumbnail) final seed-collection from your BOINC results
3. Minecraft 1.18 SSG (ie, OneChunk2 & 12-eye portals) project creation and coding, for our next Long-Term BOINC tasks
4. Celebrate the New Year, and to 1.5 Years of [email protected]!
5. Give all participants in the last two major projects badges

In 'Super Short': Minecraft Seedfinding will become potentially a lot slower as of 1.18, require more computers, time and coding until new tricks are found. We may have more CUDA and OpenCL (long-term) projects as well along the way. Thank you to everyone who has been with us, whether from the beginning, or the past few months to dedicate your resources to finding lost, new, or the impossible Minecraft seeds.


Now, for 'The End':
If you know anyone who may have some free time, and are good Java/C/C++ coders and/or interested in Minecraft, please let us know, or join us on the Discord at: [url] https://discord.gg/FVM4SPp [/url].

Many of us have spend hundreds or thousands of hours (and dollars) into this passion project of ours since June 2020, with occasional external assistance from both content creators, and even from Minecraft/Microsoft themselves. As we continue into the New Year, I'd like for you to keep in mind that some of us may not be able to dedicate as much time as we have over the first, or any at all. That includes myself, as one of the original [email protected] members and contributors. My own professional career has kicked off over the past few months and life has changed quite a bit, meaning I have not been as attentive and able to assist as I have before. I feel and read that the same is certainly been true with other members' lives, whether it due to changes in their own jobs, family, other hobbies or ongoing educational aspirations. As [email protected] grew, so did we.

You may already know we are quite different than other BOINC projects, as we're almost fully open-source on Github, and both self or community funded by a bunch of mostly anonymous, worldwide and passionate Minecraft fans. A handful of our notable and contributing members are experienced professionals, but many of us (including myself) are much younger compared to other BOINC project teams and not affiliated with major academic or research institution. Especially as some of us do not have extensive educational and work experience outside a Bachelors degree (yet), if any degree at all! It's amazing what we've been able to accomplish over the past 18 months, especially the scale of our computing infrastructure and the increases of non-developing but active members of our Discord server, who all share the same passion and excitement for the interesting worlds of Minecraft, regardless of if they are lost or (re)found. Occasionally we have trouble with distributing our findings and announcements, but Youtuber experts and a few historical Minecraft figures have pitched in over the past year and a half to help us solve these problems- and to our relief.

With almost all of the historical seeds of interest found, and all/many of the recent Panorama (menu) seeds discovered by our Discord members' own quick computational projects, some questions may arise to what can we find going forwards? This situation, along with the reduced ability to dedicate our free-time to projects from external and internal factors means that we might have a slower period of announcements and minimal administration of projects in the New Year, until or unless new blood can be found or major, exciting discoveries are presented. There are so many techniques, tasks and knowledge one must keep in their minds when working on and with [email protected], making it difficult for some of us to both train and retain possible new members of our development, testing and project teams. Many are too inexperienced, young or busy to become accustomed to our project's (chaotic at times) development processes and cycles, and so it leads me to wonder personally what we can do to improve this, or how long we can continue running the projects. We're not having a computational resource problem, but a human (and hyena!) resource related one. Not enough meat in some places, we got the bones still in these big seedfinding steaks, but they're not as juicy or fresh as they once were.

Regardless, I feel I can say that we're not going away yet, or within the next half year. You and I may need to keep our expectations smaller and more realistic, and there may be chances that a lot of work done on an unknown, future project is for 'nothing'... aside from knowing that either some code is wrong, or that quite literally, something is truly impossible due to time, resources or lack of possibility of occurring.

Goodbye for now BOINCers, and enjoy your December 2021, whether freezing, festive or boiling hot-
Hyenadae/Hy & [email protected] team
3) Message boards : News : Trailer Thumbnail 2 Crack (CUDA/Nvidia GPU) Released! (Message 708)
Posted 7 Oct 2021 by Profile Hy
Post:
Hi, our minimum BOINC driver version for nvidia unironically NV-418 and newer because we chose that version last year as an "outdated driver" for reference. We had a lot of devices running ancient drivers (380 and older) and failing tasks, and once we set this restriction and got them to update, things worked better and it was easy to see who could run OpenCL 1.2 or not.
4) Message boards : News : Trailer Thumbnail 2 Crack (CUDA/Nvidia GPU) Released! (Message 695)
Posted 3 Oct 2021 by Profile Hy
Post:
If anyone is wondering, here is the image of the trailer terrain / hole we are trying to crack :)
By estimates from the past full day of work, it will take around 42 days to complete all tasks!

5) Message boards : News : OneChunk launch and onwards (Message 693)
Posted 3 Oct 2021 by Profile Hy
Post:
Also, some fun statistics from our BOINC server.

BCM2835 [Impl 0x41 Arch 8 Variant 0x0 Part 0xd08 Rev 3] 38 4.11 2.31 9.47
[Impl 0x41 Arch 8 Variant 0x0 Part 0xd03 Rev 4] 19 4.21 1.38 5.79

We have around 60 64bit Raspberry Pis connected to our BOINC network. Each Pi 4 ('0xd08') can provide 9.5 GFLOPS of compute on average, and each Pi 3 can provide 5.8GFLOPs. There are as many people using Raspberry Pis to compute OneChunk tasks as people with Ryzen 3700Xs, though their performance isn't obviously comparable.
If all of the Pis had their CPUs fully maxed out by our applications, they could provide our network with 450 GFLOPs of compute power. Our total network capability is around 39 TFLOPs, so sadly these Pis are only worth '1%' of our BOINC estimated compute for CPUs. Our GPU network, at peak is between 4-6 PetaFLOPS of compute, which is an insane amount to have if you put into perspective that most consumer GPUs are under 15TFLOPs.

P.S.
We are still discussing 12-Eye tasks for BOINC and you all to run later, it should be a relatively quick run though as we have mentioned, a lot of data has been pre-filtered by a more 'centralized' but still powerful server due to various requirements and input data sizes.
6) Message boards : News : Trailer Thumbnail 2 Crack (CUDA/Nvidia GPU) Released! (Message 692)
Posted 3 Oct 2021 by Profile Hy
Post:
Almost time for production / full BOINC launch! Good luck to whoever calculates and finds the seed out of the billions of possibilities.
May that weird pit be filled one day!
7) Message boards : News : Trailer Thumbnail 2 Crack (CUDA/Nvidia GPU) Released! (Message 669)
Posted 30 Sep 2021 by Profile Hy
Post:
I'm curious how long this will actually take, but enjoy having your GPUs loaded up (though, not as heavily power-wise compared to Kaktwoos-cl!). I think I estimated around a month with 500 'GTX 1650 (1060)' level GPUs. Though having a couple of faster ones in the mix shouldn't change that too much.

Will be great after all these months to have the final seed for the trailer too.
8) Message boards : News : OneChunk CPU In Testing (X86_64 & ARM64) (Message 614)
Posted 10 Aug 2021 by Profile Hy
Post:
Also, code is being uploaded to https://github.com/minecrafthome/minecrafthome/pull/77 for final approvals before being put on our server.
9) Message boards : News : OneChunk CPU In Testing (X86_64 & ARM64) (Message 612)
Posted 10 Aug 2021 by Profile Hy
Post:
This week marks the period where we finally begin creating tasks and testing the OneChunk CPU application in an actual BOINC environment. Once tonight and tomorrow's testing shows that we can validate these tasks, and that our work-gen / task data creation system is working as expected we shall be able to roll this out to our main [email protected] server as a 'Beta App'. Over the past month since my last progress post we've had yet another set of delays. Some of this is because 'Sia' has not implemented the features we were hoping to use, and also some of our developers have been travelling in the past few weeks.

(See; https://minecraftathome.com/minecrafthome/forum_thread.php?id=102 )

On the bright side, we shall now be checking seeds for 11-Eye dungeons in version 1.7 and 1.8 at the same time, and our code has had some significant changes for the order of 'filtering' and how it works to find these special seeds and spawns required for the Super Speedruns / OneChunk challenges.

Shown below is a screenshot of my Raspberry Pi 4 (2.025GHz) running 4 tasks concurrently. We will be tweaking the time estimate (known config issue), but each individual task will take 4 hours (per core) on an RPi, and for example on a 5800X it takes about 1.5 hours per task per core.



For our BOINCers and those interested in running on a Raspberry Pi 4 (64bit), please note that the latest Debian Repos / Raspbian 64bit OS images have an issue with BOINC. This may appear as the BOINC client / BOINC manager saying you do not have a "GUI-RPC" permissions or password, and make the process of having the BOINC GUI working more frustrating than it should be. There's a few guides and other forum posts on this issue, but I shall probably make a quick step by step guide to ensure BOINC is operating properly *if* your 64bit ARM distribution wants to not play nice
10) Message boards : News : OneChunk CPU Pre-Announcement (X86_64 & ARM64) (Message 608)
Posted 2 Jul 2021 by Profile Hy
Post:
Hello all!

So, remember when I said we had some issues with the work generation, validation, and the immense amount of tasks we expect to store? Well, we've made progress towards making OneChunk more manageable and actually practical on our infrastructure. Some of our admins have recently collaborated with 'Sia', a platform for distributed cloud storage to test out alternative workgen data-input methods. We thought, our BOINC' network is distributed, so why not use another distributed service to solve some problems we kept worrying about?

Boysanic (Boincboi), Matt, I and others on our Discord are discussing some work that has been done the past few days. Boysanic is testing out a basic, but good method of converting our pure-text 9GB seed input file into a 3.5GB binary input file, which we plan to be put on Sia. They have also made adjustments to a validator to easily read our changed data and ensure it compares to the original between conversions. This compression ratio could also apply to the seeds your clients output (or for outputs of future projects with similar data created), but we expect our 1.5 million tasks should not take up much space on your devices, or our servers once this project is officially launched.

Essentially, while generating work (gradually) we'll stream in the data for seeds directly using Sia based file with an offset, bypassing the need to thrash our I/O and RAM storage. It's basically free cost-wise as our file isn't too large, and the many arguments over the method of 'dynamic workgen' can be put to rest I figure.

This may be a week or two, but I am happy to help test and assist with the last touches coding-wise on our BOINC project. Mainly to enable CPU-tasks and add the new platforms (ARM64, ie Raspberry Pi) I mentioned in the original post.
11) Message boards : News : Kaktwoos version 2.13 released + reminders for Nvidia Linux OpenCL packages (Message 606)
Posted 6 Jun 2021 by Profile Hy
Post:
We've released the latest version of Kaktwoos-cl, which is now 2.13!

In this update, we've added further checks for GPU model detection and printout for what GPU and its 'name' a given Kaktwoos-cl task is on. Please note that we highly suggest you run sudo apt-get install nvidia-opencl-dev if you are on Ubuntu or Debian (or a corresponding package + command on your distro) in order to install the missing OpenCL headers and allow Kaktwoos-cl to run. We would appreciate that all BOINC users thinking or currently running Nvidia GPUs on Linux to verify that BOINC tasks are not reporting computation errors accidentally, and leaving them running 'carelessly'. We have not truly resolved the very rare 'stuck at 100% / Infinite Task' post-crash issue, and we recommend for those tasks to be aborted manually if you encounter one at >99% for more than 10 minutes on a new GPU.

Changes:
Nvidia GPUs will use one of two kernels depending on their age and model. For example, any RTX or 16xx (Turing) GPU shall continue to use optimizations introduced in Kaktwoos v2.10. Older or weaker GPU models shall use the previous kernel, which means we should see a 5% boost to any GPUs reported to have a regression, as some of you may reported on our BOINC threads.

AMD RDNA1/2 GPUs are now detected and matched to use another set of optimizations, increasing their speed to their original (pre 2.10) or further increasing it. Due to architectural differences, changes made that improved GCN (ie RX 480, Vega 56) performance by 25-40% reduced RDNA 1/2 performance by up to 2X, which is now mitigated.

All task outputs will include the seed range searched as well (s: and e: parameters) for general interest and debugging potential. Otherwise, there is not much left I (Hy) feel there is to code for Kaktwoos-cl, and CPU projects are now more of a focus for [email protected] than before, as you may have seen with our OneChunk pre-announcement.
12) Message boards : Number crunching : Kaktwoos out of work . . . (Message 604)
Posted 6 Jun 2021 by Profile Hy
Post:
So, I was actually about to make an announcement that Kaktwoos v2.13 has been pushed to the server for beta-testing and some of us are running tests on it, but I myself have not had issues receiving new work. We might need to give the server a reboot and switch everyone over to the new versions if for some reason it isn't providing old applications new tasks or it is 'stuck', as that sometimes happens
13) Message boards : News : OneChunk CPU Pre-Announcement (X86_64 & ARM64) (Message 601)
Posted 5 Jun 2021 by Profile Hy
Post:
Small update, for now. Sadly things are still in progress as our team's work is divided between multiple projects and IRL (school/work) in-between us and them. I personally have been studying for my next IT certification over the past few weeks, and occasionally have been speaking to the [email protected] team to check in and provide reminders on a few things, and also to post here for updates

Code hasn't been changed for OneChunk/Filter9000 since my previous announcements, and other members are providing input and confirmation on the exact plan on how to generate work and manage the input+output files to keep track of things. The main delay is the work towards our 1 year anniversary of BOINC and [email protected] as its own group as we know it currently. We are preparing scripts, videos and other material for a celebration in the upcoming weeks. A good amount of that is reviewing previous work and projects for any new insights we can make since our discovery of Pack.png last summer. Please check the main page of our site to join us on Discord for any news on it, and to participate with us and the hundreds of users during the celebration!

For updates to Kaktwoos (v2.13), we are waiting for final approvals and merging before sending it to the 'infamous' testbed, which we suggest that you do not get too attached to if you happen to stumble upon it like other BOINCers have.

Additionally, another long-term CPU project (terrain checking) is being theorized, involving a very frustrating "Shot T" which contains nothing but a few grass and dirt blocks, in a pit of some sort. Supposedly, it may take a year or more to find Shot T with 400+ "Core i5" level processors, assuming the code is functional and ported to BOINC after our developers can do as many optimizations as they can. If it indeed takes a long time after we eventually get it onto BOINC, then we may begin cracking "Custom World" example seeds which are another major interest to us.

Various shots for the Trailer seedfinding and eventual recreation project are here:
https://imgur.com/a/ozHEde3

We also recently found the Youtuber 'Seannaners' original Let's Play seed, and I am happy that we did so. Especially after our work on both the recreation and the dungeons, which lead to a relatively quick find after we temporarily paused the project in December.
14) Questions and Answers : Web site : [email protected] still down (Message 600)
Posted 5 Jun 2021 by Profile Hy
Post:
[email protected] I think that's for one of our members' 'Primes' search application.

Honestly it isn't an official [email protected] project and I have no idea, but please contact us (admins) on the [email protected] discord if you want to see what they're doing with it.

https://discord.com/invite/9MBZSmws?utm_source=Discord%20Widget&utm_medium=Connect
15) Message boards : Number crunching : Kaktwoos out of work . . . (Message 599)
Posted 5 Jun 2021 by Profile Hy
Post:
Hi again Keith,

Yep, we gave Kaktwoos another few rounds of work. We're not (sadly) finding any new seeds since the past few weeks, and the amounts of seeds being returned for taller-cacti is shrinking with each new milestone.

In other news, we may have additional two projects in the pipeline inbetween us preparing for our 1 Year "BOINCiversity"
16) Message boards : News : OneChunk CPU Pre-Announcement (X86_64 & ARM64) (Message 585)
Posted 17 May 2021 by Profile Hy
Post:
One of our coders, Matt, is working on some optimizations for OneChunk while we wait for our BOINC-server side code to be finished. So far, he has given my Ryzen 5800X a 12% performance improvement (110 seconds for 10 input seeds, versus 124-125 before). My Pi 4 (doing two seeds with dungeons and lots of output) went from taking 95 seconds to 88 seconds, or an 8% speed improvement.
17) Message boards : News : OneChunk CPU Pre-Announcement (X86_64 & ARM64) (Message 582)
Posted 10 May 2021 by Profile Hy
Post:
Thanks for the signs and support and ARM64 hardware being prepared to contribute to this project! We've currently got 4 projects (Kaktwoos-cl update, OneChunk, "new seed video" and Trailer-seed) in progress, but I will keep you updated as we work to close out some of them, and finalize the OneChunk CPU tasks with later testing.
18) Message boards : Number crunching : Kaktwoos 2.13 announcement / testing (Message 573)
Posted 6 May 2021 by Profile Hy
Post:
For the actual release of kaktwoos-2.13, there are about two or so github pull requests we need to verify and close (I submitted them, other devs need to make sure it's alright) and then we shall send that to the testbed again for a BOINC-sanity check. I've already done local, BOINC-free runs of the kernels on my Vega 56, GTX 1650Mobile and i5-5600U (Linux-OpenCL) GPUs without any issues. Benchmark seeds came back fine, no crashes in the usual environments so things should be alright.

I'd say give it another week/past the weekend before we are done, as we have had a surprise on a previous shelved project and some more Minecraft Alpha seed-announcements are likely within the next week!

Progress is also slowed on the OneChunk project from the above event, but once Kaktwoos' code changes are done, then focus can be put on to that
19) Message boards : Number crunching : Kaktwoos 2.13 announcement / testing (Message 572)
Posted 6 May 2021 by Profile Hy
Post:
Uh, that's not a server anyone should really be using for anything, outside of [email protected] devs logging in to run tasks for like 2hrs to make sure nothing's broken and us then forgetting to kill it, so people don't have... issues or complaints

Also the version of 2.13 on the server has a few coding issues that are resolved, but literally my only advice is to please get any wuprop people off the testbed and onto either our real [email protected] server or another project if they'd like to run tests or get hours racked in.

In general the Nvidia driver's interaction/usage of OpenCL is a nightmare as it happily locks a whole thread and does nothing inside the locked thread aside from syncing stuff. AMD and Intel GPUs don't have this problem. At some point in 2.13 also I swapped the kernel out, so your GTX 1060 should be utilized like 5-10% more than before, because reasons. RTX GPUs and GTX 1600 series cards will use the same Nvidia kernel found in 2.11/2.12 so everyone can get boosted speed, or higher avg speed of seed calculations without weaker or older cards missing out. Note that Kaktwoos is only designed to run one instance per GPU, so if you mean "8X" to be "8 kaktwoos-cl tasks on 1 GTX 1060", then there's no guarantee of any good outcomes / usable systems / efficient calculations from that. It's a resource issue and more isn't better, unlike some other GPU projects.

Aside from that, there's no support for the boincboi_testbed, pretend it doesn't exist (unless you want to heat your house up, I suppose)
20) Message boards : News : OneChunk CPU Pre-Announcement (X86_64 & ARM64) (Message 568)
Posted 5 May 2021 by Profile Hy
Post:
Small update, we've decided roughly on the input work and are slowly making pieces of the BOINC template files, and task estimates. Keep in mind, here we are using "BOINC Benchmark Whetstone MIPS = OPS"
These break down to:

5,800,000,000 OP/sec per Ryzen 5800X Core Whetstone (what credit is based off of)
12.5 seconds per seed/input line (ie, 10 seeds computed with our current program takes 120-125 seconds on a single 5800X 4.85ghz core)

So, 72,500,000,000 "OPS per seed input line per core" on Zen 3

Thus:
21,750,000,000,000 OPS (2.175e^13) (300 lines of input seeds) of computation done per core, per BOINC task.

If everything works as expected, then CPU task time estimates should be close to reality, being 60-65mins on a Ryzen 5000 CPU, and say, 90-120mins on a Ryzen 1000 CPU due to IPC and frequency differences. Also for fun stats, a 3.2GHz Broadwell laptop core is 2.2X slower (271 seconds for 10 seeds) versus my 5800X


Next 20