Posts by robertmiles

21) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 107459)
Posted 16 Oct 2022 by Profile robertmiles
Post:
I was able to open profile at cosmology@home, but then i logged out and can't log in.
Somebody needs to contact them. Anyone got a live contact?

All my computers are still connected with Boinc, and I can turn things off and on in Boinc, I just can't to the website to change settings any more. So I can probably still do work for them ok as I've got most stuff in that project set up perfectly here.

I THINK I was able to find a live contact and report it.

Let's watch over the next few days to see if this worked.

The error was reported there months ago.

http://www.cosmologyathome.org/forum_thread.php?id=7812
22) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 107260)
Posted 12 Oct 2022 by Profile robertmiles
Post:
For those of you interested in other BOINC projects:

collatz has shut down, and expects a few months before it is ready to restart.

SETI@home has shut down, and plans to restart only when they get a new source of data.

cosmology@home appears to have a server problem - they've been unreachable for about a day.

RNA World stopped creating workunits over a year ago, but is still trying to get their last 20 or so workunits finished. The remaining workunits are expected to run for months each.
23) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 107108)
Posted 7 Oct 2022 by Profile robertmiles
Post:
Any idea why completed work units are not being validated?

Sometimes it's because they are waiting for enough other copies to complete the quorum,

Sometimes it's because the validator program isn't running.
24) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106916)
Posted 13 Sep 2022 by Profile robertmiles
Post:
I wonder what the bottleneck is at WCG? I've noticed if I'm downloading tasks I've just been issued, they're downloaded much more easily than ones I'm retrying. This suggests to me a disk bottleneck, and the recent tasks are still in the cache. Universe benefitted immensely from going to SSD, perhaps WCG is still on the old rust spinners?

They appear to SLOWLY be making progress against their server problems.

Also, they are starting to offer GPU tasks again.
25) Message boards : Rosetta@home Science : More metal-binding proteins? (Message 106765)
Posted 12 Aug 2022 by Profile robertmiles
Post:
Scientists identify a plant molecule that sops up iron-rich heme

https://scienceblog.com/532839/scientists-identify-a-plant-molecule-that-sops-up-iron-rich-heme/

Unclear if this molecule is a protein.
26) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106753)
Posted 10 Aug 2022 by Profile robertmiles
Post:
enable ntfs compression on slots directory

If you don't mind rather slow access to that directory.
27) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106723)
Posted 7 Aug 2022 by Profile robertmiles
Post:
[snip]

Some projects give stupid estimates for time. For example Einstein CPU tasks take me 8 hours on a fast computer and a day and a half on a slow computer. Yet on the slow computer they start off by saying estimated 2 hours, so they download way more than they can do on time.

GPUGRID is worse - all task start out with an expected runtime of about 400 days, and a deadline more like a week.

This usually gives them the highest priority for GPU use at first. Once they get progress enough to have an expected remaining time of about half a day, they'll let other GPU tasks run instead. Actual run time is seldom more than five and a half days.

Another peculiarity of GPUGRID tasks - if they resume from a checkpoint, they will initially show a progress no more than 2%. This will be corrected when the next checkpoint is written.
28) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106720)
Posted 7 Aug 2022 by Profile robertmiles
Post:
[snip]

Thanks I run several projects at once. I haven't noticed any missed deadlines, unless I go on vacation or something. As fa as I know the project sends the number of WUs it thinks I can handle. Anyway, why are there no WUs? All I get are Einstein and Milkyway WUs.


A few possibilities:

1. All the remaining workunits are set up to run only under operating systems other than the one you're using.

2. You didn't get many from the Einstein and Milkyway projects for a while, and now BOINC is requesting more from them to make up.

3, The Rosetta server has decided that tasks sent to your computer are too likely to fail, and therefore it should avoid sending any more.
29) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106707)
Posted 4 Aug 2022 by Profile robertmiles
Post:
I haven't had any WUs for weeks. No explanation or notices. Is it time to disconnect from this project?

I've had quite a few. You may need to check a few things,

Is Vbox installed? If so, what version?

What CPU does your computer have?

Does your computer have virtual computing enabled?
30) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106704)
Posted 3 Aug 2022 by Profile robertmiles
Post:
[snip]
[quote]And that's been so for the dozen-plus years I was more involved in comparing them as well as the other day when I looked at the most recent results. Combine that with the integrated nature of Norton (Firewall, VPN, Anti-tracking, DarkWeb monitoring and separate utilities) contained in the package, AVG falls into the toytown bracket I'm afraid.
And yet I see it consistently cleaning up very nasty viruses Norton failed to.

One way to do that is to report things as very nasty viruses even if they aren't viruses at all.

Unfortunately, people who like to get everything free do like to claim that a failure to report problems it can't find is some kind of guarantee of comprehensive quality. That always gets a laugh from me.
I wouldn't worry about what I pay for. All my devices are covered by otherwise unused licences
The day I pay for software is the day programmers stop putting bugs in it.

Most bugs are put in accidentally, not deliberately.

Paying for software allows more testing to find the accidental bugs, so they can be removed before users see them.

Users using a variety of antivirus software makes it difficult for virus writers to produce a virus that can evade all of them.

However, antivirus software often has features that other antivirus software expects to find only in viruses, so it is often a bad idea to install more than one antivirus program on the same computer. For example, they often have strings of bytes telling them what to look for to recognize certain viruses, and other antivirus programs looking for the same strings of bytes will see them as viruses.
31) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106697)
Posted 2 Aug 2022 by Profile robertmiles
Post:
Have you guys seen DENIS project yet?
Still running some beta, but short work.
It's about the heart.
It's just CPU

I participated there years ago. I'll look again.
32) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106679)
Posted 30 Jul 2022 by Profile robertmiles
Post:
damn trash is showing up in the python stuff.
2 x I have had to abort stuff with 6 hours run time and 20 mins cpu time and cpu usage of .20% and a completion rate of .003%/ 2 secs.

aaae-mNHM_pp-mNMABU-mPHE-mACHA14C_pp_2867243_12_1

[snip]


Also, soon after BOINC starts, its event log file should include a line showing the processor features, such as:

7/30/2022 6:06:16 PM | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx smx tm2 pbe fsgsbase bmi1 hle smep bmi2

Show us this line from your machine so we can check for some missing features that are required for the Rosetta Python tasks.
33) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106678)
Posted 30 Jul 2022 by Profile robertmiles
Post:
damn trash is showing up in the python stuff.
2 x I have had to abort stuff with 6 hours run time and 20 mins cpu time and cpu usage of .20% and a completion rate of .003%/ 2 secs.

aaae-mNHM_pp-mNMABU-mPHE-mACHA14C_pp_2867243_12_1

[snip]

I haven't seen that error, but I've seen a different problem under another BOINC project using Python (Cosmology) and found something you might want to try.

If the task goes into a state where it says it will need to retry later, set BOINC's state to Suspend, then shut down BOINC, About a minute later, restart BOINC, then set its state to Resume.

The tasks that were trying to resume later either start or are now waiting only for their turns on the CPU, this fixes it, although at the cost of losing any progress on any task that has not gone to a backup yet.

Let us know if this is even relevant to your problem.
34) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106670)
Posted 28 Jul 2022 by Profile robertmiles
Post:
I'm a bit mystified why my RAC keeps diving despite 9 validated tasks in the last whatever time period it is allowed to show.

That means that you are not returning validated tasks as fast as you were earlier in the month the RAC is for.

If you can't download tasks, you can't complete and return them.

The tasks that are shown is usually for a shorter period, such as a week.
35) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106608)
Posted 20 Jul 2022 by Profile robertmiles
Post:
I have attached everything that was on boincstats. And set all projects that i don't want to crunch to no new tasks
They don't have everything. Private GFN server (an offshoot of Primegrid) for example wasn't on there until I requested it: http://boincvm.proxyma.ru:30080/test4vm/ but I can't find the bloody invitation code (I'm already joined, but I can't post it here as I can't find it now and didn't note it down).

The projects I've seen that use invitation codes put it on their home pages, but only left it there until they thought they had about as many users as their servers could handle.
36) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106606)
Posted 20 Jul 2022 by Profile robertmiles
Post:
[snip]
[quote]***Country of origin for science has nothing to do with politics. I don't care these guys are Russian. If they are doing something good, great.
Indeed. I started an argument about that on their forum when a few people said they were boycotting it, and for once I wasn't banned. The admins just deleted all the posts to do with politics. They seem to be getting through a lot of work, so I think the boycotters were a small minority.

SiDock in in Slovenia, not Russia. Slovenia used to be part of the Soviet Union, before it broke up.


https://www.gpugrid.net/join.php
But there was something I heard recently about them producing CPU and AMD GPU work.

They tried producing OpenCL work that could run on AMD GPUs, Then they decided that such work would run too slowly to be worthwhile, and stopped generating any more.

Some CPUs from Intel can also do OpenCL work. But, few BOINC projects mention whether their OpenCL work will run on those CPUs.

They have generated some CPU work, using programs that don't use GPUs. It seems to be unavailable for any computers that have ever run some of their GPU work, though.
But again, electricity costs are to much.
Get some solar panels.

If you want your computers to run only during daylight, and only if the clouds aren't too heavy.
37) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106599)
Posted 20 Jul 2022 by Profile robertmiles
Post:
They do. They use programs partially written in CUDA, which will run only on Nvidia GPUs.

They have had some CPU work, but there is none available there currently.
I thought this new python stuff they had was meant to be for CPU and AMD GPU?

Many GPU BOINC projects now use OpenCL instead of CUDA. OpenCL will run on Nvidia GPUs, AMD GPUs, the GPUs built into some Intel CPUs, and maybe a few more types I haven't read about.

Python is for the CPU portions of the tasks only, and does not affect which GPUs will run the GPU portions of the tasks.

You may need to install a rather new version of VirtualBox to get the vbox64-mt that some of the projects using Python require.
38) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106596)
Posted 20 Jul 2022 by Profile robertmiles
Post:
I think gpugrid only uses nvidia gpus,
but i don't have amd gpu to test it.

They do. They use programs partially written in CUDA, which will run only on Nvidia GPUs.

They have had some CPU work, but there is none available there currently.
39) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106595)
Posted 20 Jul 2022 by Profile robertmiles
Post:
Uh oh. In 2038 Boinc will cease to be.
https://en.wikipedia.org/wiki/Year_2038_problem

Only for projects and computers that use versions of Linux that store the time in 32-bit integers. Newer versions that store the time in 64-bit integers won't have this problem.
40) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 106576)
Posted 19 Jul 2022 by Profile robertmiles
Post:
[snip]

I'm getting work every other day en masse and a trickle inbetween, but then I'm pestering the server with all 10 machines every 5 minutes. Fuck them. If they can't produce the work, I'm gonna keep asking for it.

What if they start giving priority for getting work to machines that have gone the longest since their last request?


Previous 20 · Next 20



©2022 University of Washington
https://www.bakerlab.org