Posts by sgaboinc

1) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 103577) Posted 28 Nov 2021 by sgaboinc Post: no more work ? really? As of 28 Nov 2021, 12:00:15 UTC [ Scheduler running ] Total queued jobs: 0 In progress: 66,178 Successes last 24h: 48,304 Users (last day ): 1,376,949 (+11) Hosts (last day ): 4,479,172 (+48) Credits last 24h : 10,490,148 Total credits : 140,755,449,638 TeraFLOPS estimate: 104.901
2) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 103576) Posted 28 Nov 2021 by sgaboinc Post: I wish they'd use KVM/QEMU instead of Virtualbox for Linux. It's the much more efficient method of virtualization on Linux that doesn't require installing external DKMS modules since it's supported directly by the Linux kernel. That said, I don't see why we're even using virtualization when a sandboxed namespace does the job just as well. Anyway, call me when there's interest in seeking open source contributors to transition from Python to Rust. i've been wondering if some could use things like docker. that'd make do with not needing virtualization. after all python runs natively in linux. besides docker, there are things like lxc https://linuxcontainers.org/. but i'd guess setup is an issue. but i'd guess it isn't as 'cross platform' as virtualbox.
3) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 103068) Posted 31 Oct 2021 by sgaboinc Post: validator not running? I've a bunch of tasks that has not been validated for a few days https://boinc.bakerlab.org/rosetta/server_status.php Workunits waiting for validation 396840 Seem to be increasing. even though the server status page seem to say the validator is running.
4) Message boards : Number crunching : Windows 11 features (Message 103054) Posted 30 Oct 2021 by sgaboinc Post: 4.20 is x86_64 Yeaph. I took a quick look at rosetta.exe (with IDA64) and seems to be 64 bit native, whitout wrapper. I think there are still quite a few 32 bits os out there, but I'd think they are a minority. Either way, old OS are likely unpatched and vulnerable to virus, malware, ransomware infection. i think that old intel compute stick ran a 32 bit os, because its bios is 32 bits, it is one of those things that causes limitations to upgrades.
5) Message boards : Number crunching : Server error: feeder not running (Message 103034) Posted 27 Oct 2021 by sgaboinc Post: it is ok now, i'm not sure when it is resolved :) lol
6) Message boards : Number crunching : Server error: feeder not running (Message 103028) Posted 27 Oct 2021 by sgaboinc Post: I'm not sure if this is the correct forum to report this but after the server is back up after maintenance, i got these in my event log Thu 28 Oct 2021 02:09:36 +08 \| Rosetta@home \| Reporting 3 completed tasks Thu 28 Oct 2021 02:09:36 +08 \| Rosetta@home \| Requesting new tasks for CPU Thu 28 Oct 2021 02:09:39 +08 \| Rosetta@home \| Scheduler request completed: got 0 new tasks Thu 28 Oct 2021 02:09:39 +08 \| Rosetta@home \| Server error: feeder not running the 3 completed tasks stays in the queue as 'ready to report' status is anyone else having the same issues? the server status shows a running feader https://boinc.bakerlab.org/rosetta/server_status.php but that the feeder not running message persist each time update is clicked. I've tried suspending the tasks and restarting boinc-client and restarting all tasks etc to no avail.
7) Message boards : Number crunching : Time estimates seem less than optimal (Message 97445) Posted 18 Jun 2020 by sgaboinc Post: I also have a problem with Rosetta having short deadlines. Rosetta consistently sends work units that cannot be finished in time. Currently I have ten(!) work units estimated at 8 hours each. I've set up my "target CPU time" to two hours, "store at least" is 0.1, "additional" is 0.5, My PC is on 25% of the time with BOINC available 99.9%. Resource share is set to 15% (WCC plays a lot nicer). This is on an Intel I7-77xx eight core laptop. Given the information Rosetta has available why does it continue to send WU's that will not be finished? I'd like to look at the scheduler code you're using because it is not doing a good job. try using 0.1/0 instead i.e. don't cache tasks, i think partly the long run times of several hours for each task would stretch out timelines in particular if boinc-client is multi-tasking between projects. each project gets less time to run and that the tasks that needs a long time to run gets further stretched out. hence, if you run multiple projects don't cache tasks
8) Message boards : Number crunching : Running on a 4GB Raspberry Pi 4 - How to? (Message 97351) Posted 13 Jun 2020 by sgaboinc Post: my Pi4 4gb broke above 50,000 points. Others probably did better as i intermittently stops between blocks of run. i'd probably try out the 64 bit os at some point. are there visible point differences between running on 64 bit os vs the 'old' 32 bit Raspbian? i'd guess it should be 'the same' since r@h bundle all its own libraries
9) Message boards : Number crunching : Large numbers of tasks aborted by project killing CPU performance (Message 97249) Posted 5 Jun 2020 by sgaboinc Post: try 0.1/0 instead, i.e. don't cache work, i did that works well (on Pi4) so i'd guess bigger PCs probably managed with that. all that out-of-wack work downloading probably boil down to boinc's statistics, e.g. that the estimates of time for the work to complete is too short, so if you place 0.3 days of (extra) work in the cache, that could be a lot if those statistics are incorrectly calculating that a task takes a small amount of time to complete. but accordingly over time as you crunch more wu, the numbers may fix itself, but for a start use as little cache as is possible and use a recent boinc client, old ones may have bugs
10) Message boards : Number crunching : Waiting for Memory (Message 97233) Posted 5 Jun 2020 by sgaboinc Post: to avoid waiting for memory, there are few things that you can do - use at most % of cpus, try to change that to a lower % e.g. 75% if you have 4 cores, 50% if you have 2 cores this should reduce the number of threads running at the same time - change the memory settings to use by boinc so that it is > 50% like 70% or higher
11) Message boards : Number crunching : Running on a 4GB Raspberry Pi 4 - How to? (Message 97158) Posted 2 Jun 2020 by sgaboinc Post: thanks i think i'd get a 8gb one when it becomes available at my local retailer. 1.5GB tends to be seen for the 'larger' wu, but i've at other times seen about 30% of memory used on a 4GB Pi4 with all 3 threads running. Perhaps it is because the wu just started running and like you've mentioned, they tend to increase in size over time. 8 GB would also means one can do away with all that zram stuff, there really isn't a point 'wasting' cpu cycles compressing and uncompressing memory. takes more power returns less points. swap and zram only make sense if it is used to keep 'unused' stuff.
12) Message boards : Number crunching : Running on a 4GB Raspberry Pi 4 - How to? (Message 97149) Posted 2 Jun 2020 by sgaboinc Post: do share the peak memory consumption that you observe, you can try running rpi-monitor which would capture that. i'm thinking of getting a piece or two but my local distributor don't have stock yet it seemed with 4GB and running 3 threads, i don't seem to see 'waiting for memory' situations frequently any more, but of course with 8GB it is more comfortable running all 4 threads i noted credits earned seem lower with the more recent wu, various from robetta as they are prefixed with rb i'm not sure why
13) Message boards : Number crunching : Peer certificate cannot be authenticated with given CA certificates (Message 97125) Posted 1 Jun 2020 by sgaboinc Post: it seemed a large number of boinc-clients are taken offline related to this, there has been a dip in wu done just last few days and it seem to be recovering now
14) Message boards : Rosetta@home Science : New wus "Deep learning folding energy" (Message 97124) Posted 1 Jun 2020 by sgaboinc Post: maybe it is a comparison to verify results
15) Message boards : Number crunching : Running on a 4GB Raspberry Pi 4 - How to? (Message 96830) Posted 28 May 2020 by sgaboinc Post: I'll be ordering one today! I don't expect dramatically different performance from the 4GB model, but every once in a while I need more than 4GB of ram for Rosetta. Does anyone know if there is planned long term support for ARM in Rosetta? I would hate to buy 3 8GB Pis and have support dropped in a few months. it seemed to be an instant sold out on launch, Pi4s are getting too hot lol https://shop.pimoroni.com/products/raspberry-pi-4?variant=29157087412307
16) Message boards : Number crunching : Running on a 4GB Raspberry Pi 4 - How to? (Message 96828) Posted 28 May 2020 by sgaboinc Post: hi, Pi4 has got a new 8GB version, and more importantly 64 bit OS ;) https://www.raspberrypi.org/blog/8gb-raspberry-pi-4-on-sale-now-at-75/ https://www.raspberrypi.org/forums/viewtopic.php?f=117&t=275370 https://www.raspberrypi.org/forums/viewtopic.php?f=63&t=275372 it seem to be something to try out, it is Beta though, this is really on the bleeding edge lol
17) Message boards : Number crunching : Raspberry Pi4 (Message 96827) Posted 28 May 2020 by sgaboinc Post: 8GB wow, maybe one day we'd have RPi that runs more like AMD's threadrippers lol
18) Message boards : Number crunching : Linux tasks (Message 96790) Posted 26 May 2020 by sgaboinc Post: r@h tends to run quite well in linux, but i'd guess AMD E-450 is a laptop? some of the lessons learned running things on Pi4 are: 1) use a small to zero task cache setting e.g. download 0.1 days of work and 0 additional days. 2) memory may be a problem so if the notebook is not going to be used for other than boinc, you may like to set memory use as 100% 3) cooling in additional to memory may be a problem, one way with that is to limit the number of cpu cores used e.g. like 75% or 50% that may help reduce the number of concurrently running threads and reduce the number of 'waiting for memory' incidences and run cooler as well
19) Message boards : Number crunching : Large numbers of tasks aborted by project killing CPU performance (Message 96789) Posted 26 May 2020 by sgaboinc Post: reducing the work queue and not pre-fetching work probably solves it. e.g. store 0.1 days of work and 0 additional days of work. and a good idea is really to run rosetta@home on a Pi4 if you do switch off your desktops often. using a Pi4 gives decent points per watt performance, and it resolves the dilemma if you want to switch off the PC and there is still a queue of work outstanding. another way though as i've been doing when i'm running on desktops, is to stop fetching work and let existing tasks run to completion and submit the results, but this would need to be done with the 0.1 / 0 work cache settings. unless you are running off one of those extreme high core counts processors e.g. ryzen3990x https://www.amd.com/en/products/cpu/amd-ryzen-threadripper-3990x there really isn't a point of downloading a large cache of wus. i prefer the low or no work cache settings as well as i tend to find that using a large cache one often download a list of very similar wus. my thoughts are that doesn't really help since pseudo random numbers are involved in the simulations. my own thoughts are that running very similar wu probably makes things less 'random' than if i only fetch what i can handle at one point in time so that others can crunch the other similar wus. that'd probably give the results a higher entropy / dispersion and explore the search space more completely. as for waiting for memory, those on the 'bleeding edge' running r@h on the Pi figured things out. 1 reduce the number of cpus (cores) used, e.g. use like 75% (3 out of 4 cores) or 50% (2 out of 4 cores). that would hopefully reduce the number of tasks downloaded and running concurrently. it seemed it reduce the occurrence of 'waiting for memory' incidences. 2 if you are running it on a PC that you don't use, you may like to set the boinc memory use to 100%. that would let the bigger or more memory intensive tasks run and also reduce the 'waiting for memory' incidences. for those playing with it on the Pi, some even resort to using zram and such to maximize memory availability on that one single dram chip.;
20) Message boards : Number crunching : Discussion of the merits and challenges of using GPUs (Message 96729) Posted 22 May 2020 by sgaboinc Post: i don't like crunching on gpus, i do play with some python tensorflow stuff on the side, and watch how it works. the simple ones like if it is a pre-trained convolution neural network (CNN), it would run for a fraction of a second and one would not feel any different. but if it is they other way round say if you are training a complex CNN network with lots of data (say images) the gpu can run at full speeds (loud fans) maximum loads for hours consuming more than a hundred watts (the top tier ones probably consume many hundreds of watts ) . if electricity costs isn't after all cheap, doing such computation can be expensive in electricity bills. gpus are used where their use are relevant and appropriate, e.g. those CNN stuff, and a lot of those CNN models are rather huge, and the training / update process are so data intensive it would generate terabytes of network data if traiing distributed across the network even for a rather modest / small CNN model. so for those it would be more appropriate to just have it run in the GPU rather than spill terabytes of data in conventional inter-networks in minutes, flooding and choking the whole networks.

Next 20