Posts by Greg_BE
|
1)
Message boards :
Number crunching :
Not getting any WU's
(Message 113143)
Posted 9 Oct 2025 by  Greg_BE
Post: The reason for the small batches is that we are only refining a specific protein that the AI generated and the team wants to examine. We don't do the old way of raw data crunching to find new proteins. The AI does that faster and more efficiently. -- Check this out: With recent credit 4754 Registered in past 24 hours 98 How many tasks? 1,358....you do the math why you don't get any work. You want to help protein folding....don't care about boinc credits...go sign up on Folding@Home. CPU and GPU. Always work never a shortage. |
|
2)
Message boards :
Number crunching :
Not getting any WU's
(Message 113140)
Posted 7 Oct 2025 by Greg_BE
Post: As I said in the other thread, you can write them, but don't expect an answer. And why you don't get much work is two fold. The machines attached to this project are who knows how many times greater than the number of tasks generated. The reason for the small batches is that we are only refining a specific protein that the AI generated and the team wants to examine. We don't do the old way of raw data crunching to find new proteins. The AI does that faster and more efficiently. The reason I found via the chatbots for slow amounts of work is that the team takes our results and analysis them and when they are done with that they put the next batch out. It's just the way Rosetta is now in the last few years. And if you look at the structure of how Dr. B has things now...its the Institute for protein design and within that the Baker Lab and within the Baker Lab Robetta (the AI) and Rosetta (BOINC) But for all these fancy names it really is just a handful of people. There is no dedicated IT guy anymore. Some will remember DEK. There are no mods (so we get spammed now and then) and there are no Grad Students that have time to post here. They write the academic papers (research always take publishing above average communications) and move on to the next project or next protein. We get what they want to give us. Thats the way it goes now. Writing will do nothing. I tried 3 years ago when things quieted down and died. I got no reply at all. I wish you the best in getting an answer, but I would be highly surprised if you do get one. If you do it will probably reflect the above points. BTW look at the stats: Users With recent credit 3414 Registered in past 24 hours 59 Computers With recent credit 5293 Registered in past 24 hours 143 I'll leave it at that...good luck |
|
3)
Message boards :
News :
Rosetta@home Update
(Message 113139)
Posted 6 Oct 2025 by Greg_BE
Post: One other thing from Grok: Recent & Upcoming Batch Details From the project's news and forums (no updates since March 2025): Past Batch Example (September 2025): A short "RosettaVS" batch for small-molecule virtual screening (drug candidate conformations). Focused on sampling novel molecules where AI like AlphaFold falls short. Completed quickly due to volunteer scale (~55,000 active hosts). Planned Focus Areas (Teased in March 2025 Update): RosettaVS Protocol: Large-scale docking for therapeutics (e.g., sampling 100k+ conformations per target). Tied to a recent Nature Communications paper on improved accuracy. Peptide Simulations: Non-canonical amino acids and mixed-chirality designs for drug-like peptides. Volunteers run extensive sampling to validate structures. Timeline: "Near future" relative to March—no ETA given, but forums speculate Q4 2025 based on Baker Lab grant cycles. No signs of launch yet.[/b] |
|
4)
Message boards :
News :
Rosetta@home Update
(Message 113138)
Posted 6 Oct 2025 by Greg_BE
Post: From Grok: Rosetta@home, like many BOINC projects focused on scientific research (in this case, protein structure prediction and design), doesn't generate a continuous stream of work units (WUs) the way some projects do. Instead, it operates on a **batch-based model** tied to specific research campaigns or computational experiments. Here's a breakdown of why this leads to the pattern of small batches followed by weeks (or even months) of dry spells: ### Key Reasons for Intermittent Work Availability 1. **Targeted Research Batches**: The team at the Baker Lab releases WUs in large, focused sets designed for particular goals, such as virtual screening for drug discovery, peptide simulations, or modeling specific proteins. Once a batch is exhausted by volunteers worldwide, the servers go dry until the next one is prepared. This isn't a steady "production line" but rather episodic, driven by scientific priorities like new hypotheses, data analysis, or grant-funded initiatives. 2. **Preparation and Analysis Periods**: Between batches, the project team spends time: - Analyzing results from the previous batch (e.g., identifying promising protein designs). - Generating and validating the next set of WUs, which can involve complex setup on their end. - Addressing technical updates or resource allocation. This downtime is explicitly acknowledged on the project's homepage: "there may be periods of down time when no work is available." A 2020 news post elaborated that these gaps occur "while preparing for future large-scale batches of jobs and analyzing results." 3. **Current Status (as of late September 2025)**: The scheduler recently showed **0 total queued jobs** available, with only ongoing progress on existing tasks. However, a March 2025 announcement indicated plans to "launch more such jobs in the near future," suggesting a new batch could drop soon—though no firm timeline was given. Recent forum chatter (e.g., July and September 2025 posts) echoes user frustration over the lack of predictability, with questions like "Is there ANY predictability about when more Rosetta tasks might surface?" but no detailed admin responses in those snippets. ### Platform-Specific Notes - Some dry spells may feel more acute on certain OSes. For instance, recent reports (September 2025) suggest Windows users are currently not receiving WUs, possibly due to a shift toward Linux-optimized or VirtualBox-based tasks. If you're on Windows, double-check your BOINC settings for VirtualBox support and ensure your work cache is low (e.g., 0.1 days) to avoid over-downloading during bursts. ### Tips to Manage the Droughts - **Diversify Projects**: Attach to complementary BOINC projects like Folding@home (also protein-related) or World Community Grid for steady work during gaps. - **Monitor Status**: Check the [Rosetta@home server status page](https://boinc.bakerlab.org/rosetta/server_status.php) daily for queued jobs spiking. - **Forum Alerts**: The [Rosetta@home forums](https://boinc.bakerlab.org/rosetta/forum_index.php) often light up when batches drop—subscribe to the "Cafe Rosetta" section for heads-up. This model maximizes scientific impact but can be frustrating for crunchers. The project's value comes from those high-impact batches, so hang in there—the next one could yield breakthroughs in drug design or beyond! If this doesn't match your setup (e.g., individual errors vs. project-wide), share more log details for troubleshooting. |
|
5)
Message boards :
News :
Rosetta@home Update
(Message 113137)
Posted 6 Oct 2025 by Greg_BE
Post: First of all your wasting your time writing them. They will not answer you. Second, i asked the chatbots about this and it came back roughly as the AI is doing all the work we used to do for raw protein finding. It can do that faster and better than we can. What we get is whatever the team sends down in small batches to further refine that target to its best models. They take that data and work on it and if they get around to sending out the next target great. If not..well your out of luck. Roestta is gong the way of a lot of other health projects, the AI can do the job better and at a lower cost than using us. Denis@home had a good first run. Then the other person on the team got busy with his masters and then it was summer break. Last I knew they were working on new models, but again communication is zero from the project. Rosetta is not a high priority for Baker Lab now. So welcome to the new version. Plus...what work is released to us is gone in a few hours as there are more machines than there is work. I gave up on Rosetta. I have been since the early years. But this is the new reality. You can leave your machine attached and hope you get some work and not do any other projects so you can grab the first tasks that come out or you just realize that the medical science aspect of BOINC outside of WCG and GPU grid (which are both having their issues is going away and that just leaves astronomy or physics or math as the main raw data crunching types that never run out of work. Sorry to be negative, but I have seen the decline of Rosetta happen over the 4 years or so and I also tried writing the Baker lab. When the students quit posting about the work and Keith the IT guy quit posting and then the moderator left and then as the work started slowing down I suspected something was up and now you see it. This is the harsh reality. Another person didn't like my analysis last year..but I have to say to him, I was right. |
|
6)
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
(Message 113080)
Posted 16 Sep 2025 by Greg_BE
Post: Probably need to "unhide" your computers before we have even a chance of helping. What admins? There hasn't been an admin on here years. We are a help your self with the help of other users type project now. |
|
7)
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
(Message 113016)
Posted 18 Aug 2025 by Greg_BE
Post: Rosetta has run out of work for the moment. Nothing new. It is hit an miss these days. Short runs gobbled up by the massive amount of people here. The AI took away our old work that was steady. Now we run refinements of its work if needed. If you get some work, be glad, if you don't oh well..better luck next time. |
|
8)
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
(Message 113002)
Posted 11 Aug 2025 by Greg_BE
Post: And yes, that does make me a cheapskate. I'll pay out for mboards, CPUs, coolers, power-supplies even cases, but I draw the line at sound and graphics. If I can get away with onboard versions of both, I will.For me, price is always a factor. I'll jump in with this. We are volunteers unless you want to fork out serious money and pay for a fancy system just to get credit and and a higher electric bill. It would cost me a 2,00 Euro or more to upgrade this system to the lower end of something to crunch the likes of a LLM task. I don't have that kind of money to burn. At the minimum I would upgrade my old 1050 to something better that still fits me MOBO. But I am not rebuilding my entire system to gain status on BOINC. By the time I recover that money working the tech will be outdated again. So my system will crunch whats out there. If GPU goes by the wayside due to not supporting my tech well so be it. Im not starving for work on my system. 8 other projects and folding at home. I have this system using its full potential. |
|
9)
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
(Message 113001)
Posted 11 Aug 2025 by Greg_BE
Post: I can't run GPU grid anymore for the time being. They are working LLM for Chemistry and my 1080 is too old to run them. Wrong architecture. I used to...but its not something I am interested in anymore. |
|
10)
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
(Message 112985)
Posted 7 Aug 2025 by Greg_BE
Post: To run small llm you 16gb vram, the regular llm's are being run be the 3,4 and 5,000 series cards which will cost you an arm and a leg and half you life savings. The ATMML tasks are as high as we can do with the 1,000 series I think. At least with my 1080. But right now it is summer break so there is next to nothing coming through on GPUgrid. Maybe in September when schools start again there will be something more. Denis@home is a hit and miss project. He's still kind of Beta level. You get a few weeks or a month of tasks to do and then it goes dormant while they process the data and refine the models. It is not a full on project just yet. The lead researcher just wrote a paper to defend his PhD and it was accepted. So now maybe they can start focusing on the project. The project scientist said after the break he would release more. And it is a very mind boggling thing they are after. Something to do with cardiac energy if I remember. I asked him to explain it once and got lost after a few minutes. |
|
11)
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
(Message 112980)
Posted 6 Aug 2025 by Greg_BE
Post: All I am seeing are Rosetta 4.20 tasks. I don't get any rosie tasks anymore. WCG I am taking all your work. lol. I was doing MCM but now I have ARP instead. I can't run GPU grid anymore for the time being. They are working LLM for Chemistry and my 1080 is to old to run them. Wrong architecture. It used to be 7000 tasks would last forever. Now they are gone in a heartbeat. Your running spinning neutron stars and I am running that and gravitational wave searches and gamma ray searches. And at the rate a certain somone is going on destroying things in the health world, there will be no need for our research here, Because its too advanced for that person to understand. He doesn't like new. |
|
12)
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
(Message 112971)
Posted 1 Aug 2025 by Greg_BE
Post: Is there ANY predictability about when more Rosetta tasks might surface? Another biomed project from Spain I think it is has also shut down for the break. So yeah probably september. I am from Seattle originally, I know where the con is being held in general. Don't know the resort. |
|
13)
Message boards :
News :
Rosetta@home Update
(Message 112970)
Posted 1 Aug 2025 by Greg_BE
Post: Server and funding constraints: But to take the human time to create the tasks and analyze the result outside of what they are already paid to do....i think that is where it is at. Why have so many biomed projects gone silent if it is so cheap? |
|
14)
Message boards :
News :
Rosetta@home Update
(Message 112968)
Posted 31 Jul 2025 by Greg_BE
Post: According to the bots this is what is going on with this project. 🔬 Who runs Rosetta@home? Rosetta@home is operated by the Baker Lab at the University of Washington, led by Dr. David Baker. The lab's focus is protein structure prediction and design, which is highly research-driven and project-based. A significant portion of their work now also happens through DeepMind collaborations, Foldit, and the Institute for Protein Design (IPD). ⚙️ Why does Rosetta@home run out of work? It's not a constant workload system: Unlike climate modeling or SETI@home, Rosetta doesn't have an infinite stream of data. Tasks are created for specific research experiments or papers. Once a batch of protein design or folding simulations is finished, there's a pause until new targets are ready. Limited human resources: Grad students and postdocs often prepare the work units. If they're busy writing papers, debugging models, or working on grant proposals, the BOINC pipeline slows down or pauses entirely. Shifting focus and tools: The Baker Lab now uses newer tools like AlphaFold or internal deep learning pipelines for many tasks. This means Rosetta@home is used only when BOINC-scale brute-force simulation is essential. Server and funding constraints: BOINC infrastructure needs maintenance. If funding is tight or there's no active grant that relies on BOINC computation, the system can sit idle. --- 1. Virtual Screening & Peptide Design The Baker Lab is now using Rosetta@home primarily for large-scale virtual screening—in particular, a physics-based protocol named RosettaVS was introduced in early 2025 for small-molecule drug discovery across massive compound libraries . Rosetta remains indispensable for modeling peptides with non-canonical amino acids and mixed chirality—a domain where AI tools like AlphaFold still struggle rosetta.n-helix.com . 2. Testing Novel Methods at Scale Rosetta@home continues to serve as a platform for testing new structure-prediction and docking methods before they go into tools like RosettaDock and other Rosetta-based applications. These include broader scientific community benchmarks like CASP and CAPRI Wikipedia ----- Less emphasis on protein folding workloads Since AlphaFold and RosettaFold now handle many structure-prediction tasks, R@h is reserved for projects where physics-based sampling is still required; this means fewer, more specialized WUs Feature Status Structure prediction Mostly AI now (AlphaFold) Physics-based design tasks High usage via Rosetta@home Virtual screening (RosettaVS) Main focus area Job availability Intermittent, batch‑driven Volunteer communication Low transparency ------------- Lastly Work Unit Creation Workflow Design & Protocol Development Postdocs and grad students formulate the targets—e.g., peptides with non‑canonical amino acids or ligand docking scenarios—for computational testing. Protocol Testing & Validation These team members test the RosettaVS screening or peptide sampling protocols on local clusters and internal servers (e.g. Robetta), refining scoring functions and parameters. Batch Packaging & Submission Once validated, batches of WUs are packaged and submitted to the Rosetta@home server pipeline for distribution to volunteers. Monitoring & Artifact Checking Lab members monitor job queues, validate results, catch anomalies, and decide on subsequent batches. The lag between batches often reflects bottlenecks in this human-driven process. The intermittent availability of new WUs often stems from the manual, experiment-specific preparation by lab personnel—and periods when postdocs or students shift focus to other work (papers, grant writing, lab experiments). . |
|
15)
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
(Message 112577)
Posted 20 Apr 2025 by Greg_BE
Post: We have been hashing on about this IPV6 thing since it started. You can still use IPV6 you just have to add these addresses to the hosts file. 128.95.160.156 boinc-files.bakerlab.org 128.95.160.156 bwsrv1.bakerlab.org Once I did that in hosts I did not have to disable IPV4 anymore and everything worked again. |
|
16)
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
(Message 112567)
Posted 18 Apr 2025 by Greg_BE
Post: Not getting any WU's on Android, so please remove the project from the Android BOINC app Project list, if it's not supported anymore, so people don't waste their time, trying to make it work. No one from the project monitors the forums. Best that you just take this project off your android device. As far as I know this project has never supported android. The foldit program may. But not Rosetta. |
|
17)
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
(Message 112560)
Posted 17 Apr 2025 by Greg_BE
Post: Folks over at Denis are complaining about no new tasks. Dunno exactly what the problem is, but I suspect it's CloudFlare creating roadblocks all over the place. I currently have...3983 tasks queued up so I'm not having any problems, at least there. Must be..he has 55,000+ in queue I am based in Europe and have no issues connecting to him. |
|
18)
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
(Message 112553)
Posted 16 Apr 2025 by Greg_BE
Post: Cosmology@Home has been unreachable since January 8, 2004. No reason given why. RNA is dead even though BOINC lists it. SiDock is working on getting a new target. They said this month or next they should be ready. Right now there is no work. You might want to explore DENIS@Home if you like medical stuff. He's trying out a new beta version. It's been pretty steady. |
|
19)
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
(Message 112551)
Posted 16 Apr 2025 by Greg_BE
Post: Tasks Ready to send is back to zero. We already scooped up over 150,000 tasks that are being processed. Tasks ready to send 15 Tasks in progress 157006 Workunits waiting for validation 0 Workunits waiting for assimilation 1436 And from the home page: Successes last 24h: 171,275 So I guess someone has to go kick the server again... |
|
20)
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
(Message 112544)
Posted 15 Apr 2025 by Greg_BE
Post: Ever since they got their act together things have been working just fine on my end. Finally can let the computer just do its thing and don't have to check if the project is working or not. |
Next 20
©2025 University of Washington
https://www.bakerlab.org