|
I only do it when it stays cold, but I'll be contributing my 5800x3d and 3080FE shortly. (Again)
|
# ? Dec 1, 2022 21:56 |
|
|
# ? Jun 10, 2024 13:32 |
|
mdxi posted:Also, be sure to check your global compute prefs and the project-specific preferences, both on the GPUGrid site. I think that compute prefs pulled as part of an update (i.e. runtime) override prefs set though a manager app (which happen at startup). Thank you! I think that this was the root of my problem, in a way. Under the preferences in GPUGrid, I discovered the following: code:
|
# ? Dec 5, 2022 18:35 |
|
Current SAGoons FAH status: 4,990,103 work units. Less than 10k to 5M. Current WCG status: mostly broken again. Their planned storage array upgrade hasn't happened yet, and they tried turning on ARP1 WUs again, with completely predictable results. It's getting depressing just checking on that project. Also, as a sidenote, the organization that they keep describing as their datacenter operator -- which my brain mapped to "colo space provider" -- is actually a Canadian high-performance computing research group. Learning this has just made all of Krembil's problems that much more confusing. Truly, all I can say at this point is "smdh".
|
# ? Dec 15, 2022 08:12 |
|
Less than 5k WUs to 5 million for FAH goons. Also less than a week until we overtake another team and move into 61st place in the all-time rankings. The rapidity of this seems to mostly be due to MaxxBot, who is now Edit: We're also up to 24th in monthly rankings mdxi fucked around with this message at 05:48 on Dec 27, 2022 |
# ? Dec 27, 2022 05:43 |
|
mdxi posted:The rapidity of this seems to mostly be due to MaxxBot, who is now This December is also the first month the team has surpassed two billion points.
|
# ? Dec 27, 2022 22:02 |
|
35th place, all-time ranking, by WUs
|
# ? Jan 11, 2023 05:12 |
|
I really need to upgrade the fans in my wife's case so I can get her machine crunching WUs under my ID . . . .
|
# ? Jan 11, 2023 06:00 |
|
DENIS has posted in their forums that they will resume crunching by Wednesday. No front-page announcement of work restart yet.
|
# ? Jan 16, 2023 23:06 |
|
DENIS has posted, as an official update this time, that crunching will restart soon; they are updating their model to account for stuff that came up in the analysis of data gathered before their break. But if I remember correctly, that was also the reason for the break? It could be both -- that they realized some changes needed to be made, leading to the break, and then realized further changes needed to be made due to analysis during the break. Still seems like an odd thing to say, two days after saying you were going to restart. Feels like "oops I forgot to do my homework" energy to me. Oh well. Edit: This kinda thing is why I have turned down half of my crunchboxes, and will be rebuilding the remainder into converged storage/compute machines with decent GPUs. CPU-based grid compute projects have not been doing great lately Edit2: Okay, at least the DENIS people are good at follow-through. Also it turned out that both of their statements were true in a way: the main project is still paused, but beta WUs are flowing. I've got 38 of them right now. mdxi fucked around with this message at 17:17 on Jan 18, 2023 |
# ? Jan 18, 2023 05:07 |
|
A WCG update appears!quote:Mapping Cancer Markers: We have increased the number of threads assigned to creation of MCM workunits by about 60% on each workunit management server. As a result, the 7 day average of completed batches sent back to MCM1 research servers (10,000 workunits to a batch) has risen to 57 from 45, roughly a 25% increase. As we increased the number of threads in a two-step process and conducted maintenance on the storage server twice during the 7 day period over which the average is calculated, we do expect this number to increase further even without additional adjustment. We will continue to assess whether we can optimize it more (without interfering with the work currently underway at SHARCNET to resolve our network congestion ticket and move to the new, faster storage server). So the good news is that all extant projects are working on new work. And the downside is that Sharcnet still hasn't fixed WCG's network issues or moved them to solid state storage arrays. Also, DENIS beta WUs continue to flow. My machines have crunched over a thousand of them in the past 48h. mdxi fucked around with this message at 21:16 on Jan 21, 2023 |
# ? Jan 20, 2023 08:15 |
|
Aaaaaaaand DENIS is going back into hiatus for "a few more weeks" after this round of beta WUs.DENIS posted:After analyzing our results, we believe that one of the limiting factors could be in the number of simulations we do to fit the behavior of the model in the experiment. Therefore, before continuing with the model fit, we are going to do a new test to study this in detail. We just put in new simulations to help us analyze this experiment to see if we can improve the model and its fit.
|
# ? Jan 21, 2023 09:38 |
|
mdxi posted:A WCG update appears! Looks like they're starting to send out ARP WUs again. Unrelatedly, my WCG downloads have stalled and hit a project backoff.
|
# ? Jan 23, 2023 23:39 |
|
unpronounceable posted:Looks like they're starting to send out ARP WUs again. Unrelatedly, my WCG downloads have stalled and hit a project backoff. Lots of people were reporting the return of network issues with the return of ARP WUs, but I've actually gotten a few of those and OPNs during the past day: code:
|
# ? Jan 24, 2023 21:20 |
|
WCG The OpenZika team published a new article on their research. Key item: OpenZika posted:From almost 404 million results generated by WCG, amounting to almost 93 thousand years of computation, 61 hits were prioritized for testing. Using enzymatic and phenotypic assays, five compounds were ultimately selected as they inhibited the function of or destabilized the three viral proteins of interest, NS2B-NS3 protease, NS3 helicase and NS5-polymerase proteins. Further tests identified 8 compounds as being able to protect the cells from death caused by the virus, while showing low cellular toxicity in liver and kidney-derived cells. The two sets of 5 and 8 molecules overlap for two compounds, named by the authors LabMol-301 and LabMol-212. WGC also announced that all OPN WUs are now exhausted until that team finalizes their work on designing and generating new batches of WUs. ARP is also still on hold pending new WUs from that team. FAH No update on anything specific from FAH themselves, but I recently started folding on my desktop's CPU -- the only machine I have doing FAH on CPU. I wasn't expecting massive numbers out of it compared to a GPU, but I was shocked at how close it was between a relatively-new CPU and the several-generations-old GPU in my desktop:
Team-wise, the push to 50B points continues. Currently at 47.3B
|
# ? Feb 14, 2023 00:22 |
|
The DENIS team has announced a good news/bad news situation:quote:First, the good news: thanks to the results you've sent in, we have been able to identify the problem with the marker and gain a better understanding of what was happening. Additionally, it turns out that we won't need to increase the number of simulations per iteration, which is good news (for us). quote:We know now that one of the characteristics will not be included in this version, so we will need you for the next improvement of the model. This, in addition to the other subprojects that we have not yet started. In short, we will continue needing you. So: they're having to make more drastic changes to their software model, but there is more work planned. It's just going to be even longer before that work gets into our grubby little mitts. In personal news, I hit 1B points for FAH yesterday I'm also experimenting with under-watting Nvidia GPUs and characterizing their performance. It's gonna take some weeks to get a good, well-averaged dataset though. My control data is 10 weeks of running a 3060 Ti stock (200W); the current experiment is running the same GPU limited to 150W. I'm both abstractly interested in this, and view it as important for gathering performance metrics ahead of summer crunching since my nodes are still in an un-airconditioned garage and there are no plans to move them.
|
# ? Feb 20, 2023 16:59 |
|
DENIS is serving WUs again. Their most recent post makes it sound like they're in a testing period, in order to refine their model, before going back to full-time research work. But if you're down with supporting a beta phase of a project, that's happening now.
|
# ? Feb 26, 2023 21:56 |
|
And they'd been doing so well for the past week. I mean, not compared to the old WCG, but...
|
# ? Mar 3, 2023 18:26 |
|
World Community Grid WCG has been down for 10 days now. Summarizing a week of twitter updates:
DENIS Testing continues and the new model continues to tighten up: quote:Dear volunteers, Folding@Home We're almost to 50B points, and are up to 59th overall.
|
# ? Mar 11, 2023 19:20 |
|
mdxi posted:Folding@Home Ugh I really need to hurry up and fix the fan situation on my partner's computer so I can have that running full-tilt for the next 3 months until we need to start using A/C. Meant to do that all this winter but I got busy with . . . everything. RIP.
|
# ? Mar 11, 2023 20:27 |
|
Zarin posted:Ugh I really need to hurry up and fix the fan situation on my partner's computer so I can have that running full-tilt for the next 3 months until we need to start using A/C. On the flipside, I'm already experimenting with power-limiting GPUs, in order to revise my year-round crunching strategy. Last year I gave up on keeping the machines in the house with me and banished them to the (not climate controlled) garage and played how-low-can-you-go with Ryzen's PBO + PPT. I feel like everything I learned on the CPU side last year can be summed up as "with a good cooler and airflow, just setting Eco Mode is good enough". I was still running old, low-power GPUs (GTX 1650 non-Super) all summer though, but in autumn I shut four of those down in favor of a single 3060 Ti, which I let run wide open all winter. I've since added a second 3060 Ti, and have been letting it crunch at 150W rather than the stock limit of 200. I expect that as the weather warms up I'll be turning them both down to 125W or possibly even 100W. In any case, I should have more numbers to share by the time things heat up in more northerly climes (I'm in the desert southwest of the US). I'm waiting for reviews on the 4070 (non-Ti) before getting a third new GPU though. Anyway, the nutshell version of my learnings so far are that (1) Nvidia GPU performance/watt graph is very linear, opposed to Ryzen's inverse hockey stick, and (2) it's really easy to set power limits under Linux; just two commands, with instant response, and you can play around with it at runtime.
|
# ? Mar 11, 2023 21:13 |
|
FAH science updates! First one is deep for me, but if I'm reading correctly then a specific gene (examined by FAH) has been targeted as a site for stopping some leukemias in actual human trials. https://www.nature.com/articles/s41586-023-05755-9 Second one is less dense: FAH work has led to a predictive AI system which gives researchers as a very fast tool to determine if a specific protein is worth deeper research looking for "cryptic pocket" drug targeting sites. This makes it feasible to do studies which could result in a doubling a treatable cancers. https://www.pennmedicine.org/news/n...iously-believed
|
# ? Mar 15, 2023 23:04 |
|
WCG Update I'm rapidly running out of fucks for their systems problems. The current status seems to be (warning: I've been skimming rather than reading carefully) that all data is verified as intact, but it's living on temporary hardware so they don't want to restart everything just to shut it down again in a week to transition to their new storage array. At the least, they have refactored and old design decision which meant that the website was down whenever any of WCG's filesystems were not available, so now the site can be up when BOINC is offline. Meanwhile, there are actual updates from three subprojects.
|
# ? Mar 23, 2023 18:40 |
|
FAH Either yesterday or the day before, team SAGoons passed the 50 billion point mark We are currently 59th overall (up from 64th roughly a year ago) and 32nd for the month (monthly rankings are much more volatile, but fun).
|
# ? Mar 28, 2023 20:14 |
|
WCG has finally fixed their poo poo, moved their data to SSDs, and I'm currently running OPNG WUs. https://www.worldcommunitygrid.org/about_us/article.s?articleId=784
|
# ? Apr 4, 2023 02:11 |
|
Took another day, but one of my nodes has finally picked up WCG WUs again (MCM and OPN so far; really want new SCC WUs)
|
# ? Apr 5, 2023 18:02 |
|
DENIS I've been seeing large batches of WUs every few days, lately. From a recent update, it sounds like they're trying to tune their model further: quote:In recent weeks the search algorithm has been improving the model very little. This indicates us that the optimization is close to be finished. Once this is complete we will pass to the validation phase which we will be done outside of DENIS...
|
# ? Apr 20, 2023 03:53 |
|
WCG SCC has restarted: quote:The SCC team began research into a new target named FLI-1, a gene that controls cell proliferation, differentiation and survival. When fused with EWSR1, the combined protein of EWS-FLI-1 is an important factor in the growth of Ewing’s sarcoma cells. Creating countermeasures against FLI1 in the form of inhibitors may lead to curing Ewing’s sarcoma in the future. I'm seeing very, very small numbers of SCC WUs, but I am happy to see them for the first time in about two years.
|
# ? Apr 23, 2023 20:42 |
|
Just installed a 4070, and the first FAH WU that it pulled was w e e d s c i e n c equote:Endocannabinoids are produced in postsynaptic neurons and bind to Cannabinoid Receptors to maintain homeostasis in neuron signaling and a plethora of physiological processes. These receptors are targeted for pain, obesity, and other central nervous system disorders. Endocannabinoids show more than 10-fold selectivity towards CB1 compared to CB2. Therefore, a mechanistic understanding of endocannabinoid selectivity will help us design selective therapeutics. In this project, we are running binding simulation of anandmide to CB1 and CB2 to decipher ligand selectivity. Seriously though, I didn't know that FAH was beginning to do basic research on cannabinoids. I suppose the weed research goldrush is beginning (or has begun).
|
# ? Apr 26, 2023 19:40 |
|
I got a 4070 and did some testing with FAH: http://firepear.net/grid/updates-2022/
|
# ? May 3, 2023 18:38 |
|
Recently picked up a used EVGA 3080TI FTW3 and started to let it crunch Einstein units. Seems to do really well. There is a Somethingawful team for Einstein but its pretty thin with only 4 members in its ranks. Two that are active including myself. I like the project tho so I'll let it keep crunching for a while. https://www.boincstats.com/stats/5/user/list/0/0/780/#2
|
# ? May 15, 2023 21:42 |
|
yummycheese posted:Recently picked up a used EVGA 3080TI FTW3 and started to let it crunch Einstein units. Seems to do really well. There is a Somethingawful team for Einstein but its pretty thin with only 4 members in its ranks. Two that are active including myself. I like the project tho so I'll let it keep crunching for a while. Thanks for the reminder about Einstein. With WCG's interminable and ongoing problems, I decided to reattach my CPUs to the project and switched to the SA team. In other news (or news-adjacent things), DENIS hasn't published a new update but they did push out a new batch of some tens of thousands of WUs yesterday. I assume they're continuing work to refine their model. I'm out of words for WCG at this point. They have become the partner that you love dearly, but who is having problems and refuses to talk about them in any meaningful way. Many people on their forums have been posting loudly about how they're gonna take their cores and go home -- which I think is a very boomer approach to the problem, though I certainly also feel the frustration. Personally I'm pondering just switching entirely over to FAH, since their focus is also biomedical and that's very important to me... but that's not a decision I'm going to make quickly or out of pique.
|
# ? May 17, 2023 18:36 |
|
mdxi posted:I'm out of words for WCG at this point. They have become the partner that you love dearly, but who is having problems and refuses to talk about them in any meaningful way. Many people on their forums have been posting loudly about how they're gonna take their cores and go home -- which I think is a very boomer approach to the problem, though I certainly also feel the frustration. Personally I'm pondering just switching entirely over to FAH, since their focus is also biomedical and that's very important to me... but that's not a decision I'm going to make quickly or out of pique.
|
# ? May 17, 2023 22:39 |
|
A first for me: having DENIS, Einstein, Milkyway, and WCG WUs all in flight on one machine. Also, I finally passed 390y of CPU time for WCG. I used to be ticking over a decade of CPU time every 20 calendar days or so, but it has taken from the IBM hand-off until now for me to go from 378 CPU-years to 390. I feel like that says more about the problems they've been having than any words i can put together.
|
# ? May 29, 2023 19:07 |
|
Looking good.
|
# ? May 30, 2023 03:53 |
|
yummycheese posted:Recently picked up a used EVGA 3080TI FTW3 and started to let it crunch Einstein units. Seems to do really well. There is a Somethingawful team for Einstein but its pretty thin with only 4 members in its ranks. Two that are active including myself. I like the project tho so I'll let it keep crunching for a while. I've been blasting einstein on my main pc for a while and I'm on the SA team on this: https://einsteinathome.org/community/teams/173122/members , but not on the BOINC stats. Thats a separate team thing huh?
|
# ? Jul 2, 2023 17:48 |
|
Lobsterboy posted:I've been blasting einstein on my main pc for a while and I'm on the SA team on this: https://einsteinathome.org/community/teams/173122/members , but not on the BOINC stats. Thats a separate team thing huh? Most projects implemented a GDPR checkbox that controls data exfiltration, and it's off by default. Check that.
|
# ? Jul 3, 2023 04:44 |
|
mdxi posted:Most projects implemented a GDPR checkbox that controls data exfiltration, and it's off by default. Check that. ah beans, okay updated that and it now has my host on BOINCstats but nothing connected. I'll give it a day or two and see if that synchs. edit: ah yeah I have my old pc on one user: https://www.boincstats.com/stats/-1/user/detail/1550037/projectList and the new pc on another user: https://www.boincstats.com/stats/-1/user/detail/171916629747/projectList ah well. at least it updated! Lobsterboy fucked around with this message at 04:58 on Jul 4, 2023 |
# ? Jul 3, 2023 18:52 |
|
A friend of mine has a synology nas and a lot of extra storage laying around that she doesn't know what to do with. Is there a storage heavy grid computing workload she could dedicate it to? Run a waffleimages node?
|
# ? Jul 12, 2023 18:30 |
|
I've never heard of one. The problem is going to be bandwidth. If she wants to do a community service she could sign up as a Bittorrent mirror for some obscure scientific content, open source distributions, etc. - in other words legal content. She could even earn "torrent-coins" for hosting, with some clients. For example, some of the content on The Internet Archive has torrent downloads, and you might want to seed those files to help other people download them faster. Another thing she might do is digitize movies and other cultural artefacts which aren't in the public domain for preservation for the future, but she can't legally open that up for other people.
|
# ? Jul 17, 2023 14:15 |
|
|
# ? Jun 10, 2024 13:32 |
|
Updates, for the first time in a long time! WCG Mapping Cancer Markers quote:We continue our analysis of the most common biomarkers amongst lung cancer patients. Following discussion of the genes VAMP1 and FARP1 that were described in our March and April updates, here we focus on GSDMB. ... Similarly to VAMP1 and FARP1 genes, we investigated the role of GSDMB in lung cancer, and found that its presence has a protective role in lung cancer ... Extending our observation beyond lung cancer, as it is an aim of our WCG project, we find that GSDMB behaves in a similar way in the majority of tested cancers. This suggests that GSDMB has an important role in carcinogenesis Africa Rainfall Project quote:The ARP project running at the WCG produces a very large dataset. At this moment, there is about 300TB in more than four million files. Storing and moving such enormous amounts of data is not trivial. Until now, all data was stored in the computing centre of SURF, the cooperative association of Dutch educational and research institutions. After about four years of support, the ARP at SURF is about to end. So what to do with the data? DENIS quote:June Update
|
# ? Aug 5, 2023 15:42 |