Throw CPU cycles at Rosetta@Home. They're doing an awful lot of COVID-19 workunits lately.
Distributed Computing takes on Coronavirus COVID-19
- Thread starter JimboPalmer
- Start date
Folding@Home went over an ExaFLOPS and still can't keep up with the volunteer's demand.
Thanks for the heads up. There's lots of Covid-19 work at Rosetta for us BOINC users. Have it running on 6 machines so far.
Thanks. The F@H is so bad sometimes it's not even uploading/accepting the completed WUs, that's how backed up their infrastructure is.
Heh, F@H just keeps getting more people even as they increase the number of servers.
https://foldingforum.org/viewtopic.php?f=24&t=33193
https://foldingforum.org/viewtopic.php?f=24&t=33193
Wow, Team Ars has really ramped up the the compute power. Look at the slope of that projected line! Now if only the WU allocate/receive situation would stabilize and we could get more consistent work. Today seems my GPU has been getting pretty consistent work, so running F@H on my GPU, and Rosetta@Home on my CPU.
https://folding.extremeoverclocking.com/team_summary.php?s=&t=14
https://folding.extremeoverclocking.com/team_summary.php?s=&t=14
Yeah, after seeing just how many problems F@H has right now, I've been dedicating all my resources towards BOINC and Rosetta@Home. Too bad they don't support GPU compute also, but that's what I have Einstein@Home for after all!
There is plenty of work with Rosetta@Home right now.
That's a cloud instance I'm playing with - preemptible CPUs, super cheap, about $0.12/hr to run. 16 Intel cores, about 20GB of RAM (the Covid units are massive - 1.2-1.3GB RAM per task.
Yea, even with 16GB of memory on my desktop (which I also do work, TeamViewer, browsers, etc) occasionally the Rosetta processes get marked as "waiting for memory".
Unfortunately there's no native Rosetta for FreeBSD, so there goes my NAS unless I want to spin up a VM.
Unfortunately there's no native Rosetta for FreeBSD, so there goes my NAS unless I want to spin up a VM.
These are just dedicated compute instances, so I have to override the default limits on RAM/disk/etc.
Code:
/var/lib/boinc/global_prefs_override.xml
<global_preferences>
<disk_max_used_gb>20.0</disk_max_used_gb>
<disk_min_free_gb>0.5</disk_min_free_gb>
<disk_max_used_pct>90.0</disk_max_used_pct>
<ram_max_used_busy_pct>90.0</ram_max_used_busy_pct>
<ram_max_used_idle_pct>90.0</ram_max_used_idle_pct>
</global_preferences>
I'm pleased to see that there are three of us managing more than a million PPD now. Keep it up!
Good job, y'all.
Ran Rosetta@Home clear out of workunits.
https://boinc.bakerlab.org/rosetta/server_status.php shows, currently:
And I noticed a box was a bit starving for WUs this morning.
Ran Rosetta@Home clear out of workunits.
https://boinc.bakerlab.org/rosetta/server_status.php shows, currently:
And I noticed a box was a bit starving for WUs this morning.
By now I get WUs more consistently, both CPU and GPU. Little wait in between. In the end it'll depend though how much my electricity bill will go up. I'm running 9900K and 2080Ti plus a core i7 laptop. The desktop itself uses about 450W when both CPU and GPU are loaded:/
Running most of my machines on Rosetta with TN-Grid as a backup for when Rosetta WUs aren't available. TN-Grid is also doing some SARS-CoV-2 related work. For new users: to do this set your Rosetta project priority to some number greater than 0 and set your TN-Grid project priority to 0 on your project preferences page. This will run Rosetta WUs unless there's no work available, at which point it will DL and run a TN-Grid WU on each unused CPU core.
BTW whatever project you choose to run, don't forget to join the Ars Technica team.
And don't forget to have fun.
BTW whatever project you choose to run, don't forget to join the Ars Technica team.
And don't forget to have fun.
R@H has an Ars team? How do I configure it?
F@H still seems to have issues consistently giving me WU, while R@H I always have a queue available.
F@H still seems to have issues consistently giving me WU, while R@H I always have a queue available.
Trying to give this a more generic title. Thank you Continuum!
(it was that or try to chase you off mylawn Folding@Home thread, and that would be stupid)
(it was that or try to chase you off my
This is a blast from the past. I used to run rc5 with TBR back in the late 90s-2000 when G4s were the best in the world at it. Was a heavy participant in Team Vodka Martini. I guess it's time to get back into this.
F@H seems to not have much GPU work for me anymore. And CPU work comes and goes, can't seem to get any steady work. Ah well. R@H for the times when F@H has nothing. Wish I'd get more GPU work since R@H doesn't do GPU.
I haven't seen a GPU WU in a week, but my CPUs have been running nonstop for a couple days now. Looks like (from an update on that extremeoverclocking link you posted) F@H is running a large CPU-only job at the moment.
On the upper right side of your account page under the "Community" banner it allows you to search for a team. Just enter Ars and choose Ars Technica.
Here's the Ars team: https://boinc.bakerlab.org/rosetta/team_display.php?teamid=473
My yesterday was very near a normal amount of work, if this is the new 'normal' I am all for it. 1.4 million WUs served!
Hey all,
Jumping in here with an old Thinkpad T420. Is there a way to use the Intel HD 3000 in this laptop? Or only AMD/nVidia? Just curious, no biggie if it turns out to be no.
My client says I am folding for project, 14308 but I am not sure if it's active? According to this grouping it's in one of the COVID-19 sets?
Jumping in here with an old Thinkpad T420. Is there a way to use the Intel HD 3000 in this laptop? Or only AMD/nVidia? Just curious, no biggie if it turns out to be no.
My client says I am folding for project, 14308 but I am not sure if it's active? According to this grouping it's in one of the COVID-19 sets?
I can only speak to Folding@Home, it never uses any Intel GPU. The HD 3000 does not support OpenCL, so even if it did, that is too old.
https://www.techpowerup.com/gpu-specs/h ... 3000.c1257
It also does not support Double Precision floating point math, a prerequisite to F@H.
My Ryzen 5 2600 and Vega 56 have joined the fray.
I joined some random team to start but am on Ars Technica's now.
I joined some random team to start but am on Ars Technica's now.
Ah, makes sense then.
I am now throwing my (previously mostly idle) GTX960 and i5-4690K at it!
Apologies in advance if this is not the correct venue for this question.
Being stuck at home with the kids, and seeing the details of the Folding/Rosetta COVID computing tasks has given me the idea of buying a small, cheap ... Something to throw at the task. The goal would be something cheap (maybe $250 Canadian) to give the kids and I a little project, and to help in some meager way.
I'm comfortable with Linux, Windows, Mac and reasonably fluent on the command line if that matters. I've done minor PC upgrades (HD, ram, video cards) and played around with Raspberry Pis and such.
I'm looking at refurbished Windows PCs, thinking that's probably be best bang for the buck, but at similar price points there's a lot of choices like "old i5 with a video card, or old i7 with integrated graphics", "big ram and spinning hard drive or small RAM and SSD", etc.. etc..
If I'm looking for best "bang for the buck" with a specific focus on doing "some COVID-19 related computing", where should I focus my funds?
Being stuck at home with the kids, and seeing the details of the Folding/Rosetta COVID computing tasks has given me the idea of buying a small, cheap ... Something to throw at the task. The goal would be something cheap (maybe $250 Canadian) to give the kids and I a little project, and to help in some meager way.
I'm comfortable with Linux, Windows, Mac and reasonably fluent on the command line if that matters. I've done minor PC upgrades (HD, ram, video cards) and played around with Raspberry Pis and such.
I'm looking at refurbished Windows PCs, thinking that's probably be best bang for the buck, but at similar price points there's a lot of choices like "old i5 with a video card, or old i7 with integrated graphics", "big ram and spinning hard drive or small RAM and SSD", etc.. etc..
If I'm looking for best "bang for the buck" with a specific focus on doing "some COVID-19 related computing", where should I focus my funds?
The biggest impact to folding performance, by far, is a good GPU. If you're also folding on a CPU, lots of threads is helpful, so a Ryzen is ideal, but you're less likely to pick up a second-hand one of those for cheap. The jobs don't take up lots of RAM, and you only need to load them from disc once at the start of the work unit, so drive speed doesn't matter. It seems to be easier to get folding running on an Nvidia GPU than an AMD one. Whatever you get running would be interesting and helpful, though.
There are plenty of older HP, I think 8100-8300 desktops out there that can fit a decent GPU in them.
I think Phil's Computer Lab on Youtube did a few builds with similar desktops, threw a GPU in there and it was a decent gaming rig. I don't know if you'll be able to grab a couple GPUs (I think he put a 1060 in one) cheaply about now, but I believe the 75W PCIe slot power limit is the real kicker on that one.
I think Phil's Computer Lab on Youtube did a few builds with similar desktops, threw a GPU in there and it was a decent gaming rig. I don't know if you'll be able to grab a couple GPUs (I think he put a 1060 in one) cheaply about now, but I believe the 75W PCIe slot power limit is the real kicker on that one.
These are examples:
https://www.ebay.com/itm/NVIDIA-Gef...sh=item3b44e27f15:g:MrcAAOSwcMNehfM0&LH_BIN=1
https://www.ebay.com/itm/HP-xw4600-...sh=item594291fc17:g:V-gAAOSwCOJeHHXO&LH_BIN=1
That may make 350k Points Per Day on F@H
https://www.ebay.com/itm/NVIDIA-Gef...sh=item3b44e27f15:g:MrcAAOSwcMNehfM0&LH_BIN=1
https://www.ebay.com/itm/HP-xw4600-...sh=item594291fc17:g:V-gAAOSwCOJeHHXO&LH_BIN=1
That may make 350k Points Per Day on F@H
At first I was confused about my GPU not producing as much as I could swear I went for a 1080ti, but now I remember that when I bought it they were all being snatched up for bitcoin mining and I went with a budget card. Since I stopped gaming on the PC I just never bothered upgrading as it played Fallout 4 at upper settings with decent speeds!
Now I am curious what GPU might fit in there since I have some extra bonus pay!
Now I am curious what GPU might fit in there since I have some extra bonus pay!
Run what 'ya brung. No point in buying a new GPU right now, they can't keep the flock fed.
I've got a pair of crusty old 980s in my office heater and I rarely see both of them working at any given point in time anymore. They used to be reliable as clockwork at sucking down and chewing on units, but now it's insanely rare for both to be running.
======
My 8700K is out, that PC is being put to better use doing video editing now vs an office heater.
On the other hand, I've got a dual socket decade old Xeon system showing up soon, with 24 threads and enough RAM to run even the 2GB workunits on all cores. So once I get that powered (somehow - it's apparently cranky about PSUs), it should be a respectable contribution to office heat. Unless I just mount it outside...
I'm seriously thinking about building a little server cabinet in the corner of my office that has a couple 4" PVC pipes to vent air outside in the summer.
I've got a pair of crusty old 980s in my office heater and I rarely see both of them working at any given point in time anymore. They used to be reliable as clockwork at sucking down and chewing on units, but now it's insanely rare for both to be running.
======
My 8700K is out, that PC is being put to better use doing video editing now vs an office heater.
On the other hand, I've got a dual socket decade old Xeon system showing up soon, with 24 threads and enough RAM to run even the 2GB workunits on all cores. So once I get that powered (somehow - it's apparently cranky about PSUs), it should be a respectable contribution to office heat. Unless I just mount it outside...
I'm seriously thinking about building a little server cabinet in the corner of my office that has a couple 4" PVC pipes to vent air outside in the summer.
Respectable contribution to heat? The only way it could be more like a space heater is if put the vents blowing on your feet to keep them warm.
I didn't have any problems with it on a 650W. I'm pretty sure that's what I had. No GPU other than onboard, so if you were going to throw your old GPUs into it for Compute, it'll of course use more power.
I didn't have any problems with it on a 650W. I'm pretty sure that's what I had. No GPU other than onboard, so if you were going to throw your old GPUs into it for Compute, it'll of course use more power.
Nah, I'll keep the old i7 box for the GPUs. It works fine. This new one will just be raw CPU compute, during good sun hours.
Is it normal for CPU-only F@H clients to not get work? I have one quad-core sitting at my works' office, but it hasn't gotten any WU sent to it.
Just stuck at the 'Downloading' part.
Just stuck at the 'Downloading' part.
F@H has gotten better, but can still be hit or miss if there's any WU available for clients. I've had CPU work with no GPU work, I've had GPU work with no CPU work. I've had both, I've had none.
I've found that in some cases where my Windows F@H client has stalled out waiting for work, closing it from the System Tray and restarting the processes has caused it to get some new work units. For now, though, I am splitting time between Folding@Home overnight (if it has work) and Rosetta@Home under Linux.
Interestingly enough, running six threads of Rosetta on my Ryzen 2600X results in temperatures a good 3 to 4 degrees C cooler than the same CPU-only workload under Folding@Home (under either Windows or Linux). I'm guessing that the F@H client may be making more use of AVX instructions that are more energy intensive, perhaps?
Interestingly enough, running six threads of Rosetta on my Ryzen 2600X results in temperatures a good 3 to 4 degrees C cooler than the same CPU-only workload under Folding@Home (under either Windows or Linux). I'm guessing that the F@H client may be making more use of AVX instructions that are more energy intensive, perhaps?