Posts by dcdc

1) Message boards : Number crunching : WUs fail with "too many exit(0)s" message (Message 87160)
Posted 29 Aug 2017 by Profile dcdc
Post:
Hi Chilean

I have an i3 that's the same generation and found that running three threads was as efficient as running four. I think the cache is probably the reason for that- each task gets access to more of the cache if there are fewer threads competing for it, but it will also mean there's more RAM available for each thread. Also, the first two tasks get a full core each, whereas the next two only get a virtual CPU each (so they'll slow down the first two threads by competing for resources), so I'd suggest running two or three threads on that machine, by reducing the "Usage Limits" to 50% or 75% of the CPUs.

Also, under "Disk and memory", is "Leave non-GPU tasks in memory while suspended" checked? I also bump the "When computer is not in use, use at most" up to 95%.

D
2) Message boards : Number crunching : For the betterment of BOINC (Message 86858)
Posted 26 Jul 2017 by Profile dcdc
Post:
I might be a bit late here, but I'd like to see Efmer's Tthrottle incorporated (all parties willing!). Basically I'd like to BOINC being able to thermally throttle the projects, especially on a laptop.

I'd also like to see better cross-platform parity. I know there are difficulties in like-for-like comparisons, but basically I guess the thing it comes down to is how much are we saving the researchers if they were having to pay for similar time on a service like a supercomputer or AWS.

This one is not a priority, and more of a question: I'm not sure how the more modern processors deal with BOINC with regards to the various auto-overclocking features such as XFR on Ryzen, but there might be some work to optimise those(?)
3) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 86735)
Posted 25 Jun 2017 by Profile dcdc
Post:
On the "Computer information" page, could you add the machine name? I don't see it anywhere.


Are you logged in? It should show the computer names if you're logged in (works for me!).
4) Message boards : Number crunching : Constant Timeouts when Unpacking (Message 86719)
Posted 24 Jun 2017 by Profile dcdc
Post:
Could it be a read/write permissions issue for that location?
5) Message boards : Number crunching : Ryzen (Message 81505)
Posted 21 Apr 2017 by Profile dcdc
Post:
Intel\'s response - it can only be good for the consumer and DC projects like R@H.
http://wccftech.com/intel-moster-12-core-hedt-skylake-x-confirmed-launch-30-may-x299/
6) Message boards : Number crunching : Ryzen (Message 81343)
Posted 18 Mar 2017 by Profile dcdc
Post:
2 minutes on google led me to this R7 1700:

http://boinc.bakerlab.org/rosetta/results.php?hostid=2225133

and this R7 1800X:

https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=3181998

For the 1700, assuming it\'s running 16 threads, from the results returned the RAC is approx 15,210 and for the 1800X it works out to 12,268.

If they\'re not running 16 threads then the numbers above will be scaled too high. The granted credit-per-second is higher on the 1700 which suggests it\'s running a lower non-Rosetta load, or is overclocked quite a lot. Or maybe the 1800X gets powered down more frequently in that period. I\'d guess it\'s due to load though.

If those numbers are correct, then they\'re really good at Rosetta and I want one. Or five.
7) Message boards : Number crunching : DNS Problems and Late Work Units (Message 81287)
Posted 10 Mar 2017 by Profile dcdc
Post:
What happened? Some DNS issue, I know, but how come they took so long to fix it?
8) Message boards : Rosetta@home Science : Rosetta@home no longer mentioned in papers? (Message 81237)
Posted 26 Feb 2017 by Profile dcdc
Post:
I just had a quick look- rosetta@home is mentioned under \"Acknowledgements\" on the most recent paper:

...and Rosetta@Home volunteers for computing resources used in ab initio structure prediction calculations.
9) Message boards : Number crunching : Ryzen (Message 81227)
Posted 24 Feb 2017 by Profile dcdc
Post:
I\'m thinking at R7 1700...16 thread at 65W (300$)...not bad.
Waiting for single thread IPC benchmarks

I like the low power of that one too. We need some projects to test it on though. Rosetta is not ideal since the run times tend to be more constant anyway.


I find that RAC (recent average credit) is generally very consistent - you can take the granted credit from a few sample complete units and calculate the long-term RAC by finding the average RAC per second, and multiplying that by the number of cores and seconds-per-day.
10) Message boards : Number crunching : Ryzen (Message 81215)
Posted 23 Feb 2017 by Profile dcdc
Post:
Is anyone buying a Ryzen rig any time soon? If so, please post a link to the machine so we can see how well it crunches. Hopefully it\'ll be a Rosetta beast.

D
11) Message boards : Rosetta@home Science : Malware in downloaded Rosetta files (Message 81211)
Posted 22 Feb 2017 by Profile dcdc
Post:
Can you give more info? Why do you think they are infected? I.e. what program is saying they are?

D
12) Message boards : Number crunching : Unintended consequences of the new credit system? (Message 81141)
Posted 5 Feb 2017 by Profile dcdc
Post:
Overdue jobs do get credit - might take a day or so. E.g. https://boinc.bakerlab.org/rosetta/result.php?resultid=898173929
13) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 81108)
Posted 1 Feb 2017 by Profile dcdc
Post:
I think it\'s probably more likely that there\'s no funding for someone dedicated to the role, and everyone else has other priorities so it falls to no-one. There could obviously be lots more compute power available here but it might be that there is sufficient as-is so it works for getting the science done, regardless of how frustrating it is for users.

I just found my computer sat idle with the 24 hour back-off bug. I\'ll add a second project, but because the server code here is so old I don\'t believe I can add a project as a backup - only as a low % so I\'ll do that.

Maybe it would be useful if we maintained a sticky thread (Mod.Sense!) where we list the priorities from our point of view, so the team can see what we think needs fixing. I.e. under URGENT, we\'d have the 24hr bug, or make work, and then under the next heading (Less urgent?) we\'d have the server upgrade, maybe with a link to the discussion of it. If anything breaks, stick it at the top.

Might that help?

D
14) Message boards : Rosetta@home Science : MAJOR UPDATES to the Rosetta@home website (Message 80846)
Posted 14 Nov 2016 by Profile dcdc
Post:
Hi Ian, great to hear. I'll give some feedback once I've had a chance to look at it.

D
15) Message boards : Number crunching : Giving Zero Credit for Work Completed (Message 80541)
Posted 15 Aug 2016 by Profile dcdc
Post:
Validate errors also.

They are granted credit - it just doesn\'t show up until up to 24 hrs after submission, and only shows up in the task:

http://boinc.bakerlab.org/rosetta/result.php?resultid=858307319
16) Message boards : Number crunching : Stuck on Uploading (Message 80514)
Posted 8 Aug 2016 by Profile dcdc
Post:
This machine has received some tasks now too, but uploading is still stuck.
17) Message boards : Number crunching : Stuck on Uploading (Message 80511)
Posted 8 Aug 2016 by Profile dcdc
Post:
I\'ve got a stuck job on this machine and not picking up new work. BOINC log says:

08/08/2016 19:45:04 | | Project communication failed: attempting access to reference site
08/08/2016 19:45:06 | | Internet access OK - project servers may be temporarily down.
18) Message boards : Rosetta@home Science : The Protein folding revolution! (Message 80444)
Posted 27 Jul 2016 by Profile dcdc
Post:
Awesome :)

Haven\'t watched to vid yet but will in the morning...
19) Message boards : Number crunching : Raspberry Pi 3 (Message 80438)
Posted 27 Jul 2016 by Profile dcdc
Post:
I\'ve checked the power consumption now- it\'s averaging around 3.5W when running 4 threads of Rosetta.
20) Message boards : Number crunching : Raspberry Pi 3 (Message 80429)
Posted 24 Jul 2016 by Profile dcdc
Post:
I bought an Intel Compute Stick and fired up Rosetta on it. It cost £58 ex VAT from Ebuyer and has a Z3735F quad-core atom CPU, 2GB RAM and 32GB flash and Win10 (32-bit for some reason).

It\'s heading for a RAC of 760 which is really impressive. I haven\'t measured the power draw yet, but I don\'t expect much will beat it in terms of throughput to price or throughput to power consumption.


Next 20



©2017 University of Washington
http://www.bakerlab.org