Posts by Aurum

1) Message boards : News : Presensation of our work (Message 1757)
Posted 26 Jun 2022 by Aurum
I really enjoyed this presentation and now understand much better what their research is and the contribution we make to their work. This will also show how this de novo discovery is very different than any other BOINC project I know of.
He said they do not look for drug-like molecules at 2:20:

At 7:07 the chart says "de novo generation is not screening" I did not hear him explain the difference, maybe later. I think he means many of the molecules they come up with cannot be synthesized and are unstable (2/3s) and unusable. By the end I understand that they are discovering new molecules and not just screening libraries, e.g. in docking studies.

"...looking for few good examples that maybe have a ??? (sounds like shina) that is very very local." I think that missing word "shina" would be important to know. Anyone know what it is? CC said shyness.

Introduces BOINC collaboration:
2) Message boards : News : Credits and Gridcoin (Message 1753)
Posted 12 Jun 2022 by Aurum
The last point concerns the requests for Gridcoin. I've been asked by the developers and by some of you. I am not against this possibility, but three points do not allow for the moment to be whitelisted. First, I can't guarantee to always have tasks waiting to be calculated. Secondly, the incentive to cheat will increase and I find that increasing the quorum is a waste of resources. Thirdly, I'm struggling with the server to keep it up. The upcoming arrival of larger molecules should settle the first point. For the second point, we are thinking about a validation by analyzing the result. I have already made many optimizations for the third point, at the moment it's much better.
Benoit Da Mota

I hope whitelisting for GRC will be revisited. As I recall the reasons it failed before were twofold: 1, I didn't follow their rules when I proposed it regarding social media or something, and 2, that a QCP admin did not say yes.
QCP has proven for years that it has a wealth of work to do and so it should be easily accepted if one of the GRC insiders makes the proposal since they're expert at following their own rules.

Somehow quorum got folded into this discussion but has nothing to do with a Gridcoin requirement. As I recall that was just a GRC dev's opinion and hopefully it will not be brought up again to confuse the matter.
All that's needed from QCP is to just say "yes" and you're done with it. Should work stop for some reason then Gridcoin graylist's the project until it resumes, e.g. World Community Grid has been down for months and GRCgrid frequently has droughts.
3) Message boards : Number crunching : use hyper threading? (Message 1726)
Posted 17 Apr 2022 by Aurum
It's been a long time since I tried running with HT disabled. I had come to the conclusion that the problem with some BOINC projects is the amount of CPU cache they fill. Some projects like WCG's MIP had that problem and recommended a rule-of-thumb to run fewer WUs than your L3 cache divided by 5 MB/WU. I don't know if there's such a rule for QC but there probably should be. I set the maximum number of WUs to 8 in preferences with a single CPU and they ran much better. I leave HTing enabled since it seems to help normal projects more than it hurts. For CPUs with fewer threads I specify a different venue and use no more than half the threads for a project like QC. Run another project like Universe and HTing will fill in all the gaps with it and your overall throughput will be maximized.

I think it can also sense a difference between CPU cache designs, e.g. compare E5-2699 v4 to an i9-9980XE with different style caches. You didn't say what model CPU you're using just that it's 8c/16t. I suggest leaving HTing enabled and set your preferences for QC to a Max # jobs = 4 and Max # CPUs = 1. If that works good then up Max # jobs until you feel it slowing down and then back up one and your done.
4) Message boards : Number crunching : Stuck tasks (Message 1709)
Posted 11 Mar 2022 by Aurum
(Linux doesn't have the problem for some reason.)
But it does. I just aborted one that sat at 100% overnight. I don't get them very often.
I'm not running any python memory hogs or anything, just a few Gaia, TN-Grid & QC.
Cut back to 8 QC per computer and they run much better now.
5) Message boards : Science : NWChemEx (Message 1694)
Posted 5 Mar 2022 by Aurum
When I first read this article I thought here's a way to get more work done on BOINC. But then I realized it was intended for supercomputers:
“This change lets us develop methods to optimize the code over time and not force the code to run on a specific implementation. In terms of optimization, we are enabling different levels of parallelization within and across tensor contraction expressions. This allows us to tune the parallelization strategy to the specific execution,” states Sriram Krishnamoorthy, a laboratory fellow and computer scientist at Pacific Northwest National Laboratory. The change will also enable NWChemEx to run on the Intel Ponte Vecchio GPU units that will be incorporated in Aurora.
6) Message boards : Number crunching : High failure rate (Message 1679)
Posted 24 Feb 2022 by Aurum
My failure rate has shot up to 56% today!!!
Time to move on.
7) Questions and Answers : Getting started : Problem joining QuChem (Message 1667)
Posted 22 Feb 2022 by Aurum
Click on your account and then your Edit QuChemPedIA@home preferences.
Resource Share = 100
It uses a fair amount of memory so try a few and see how it runs.
8) Questions and Answers : Unix/Linux : BOINC 7.18.1 causes invalids (Message 1666)
Posted 22 Feb 2022 by Aurum
Why would one choose to run this version instead of the one in the Linux repository?
9) Questions and Answers : Getting started : Problem joining QuChem (Message 1663)
Posted 21 Feb 2022 by Aurum
Expired certificates sounds like a problem with a previous version of BOINC for Windows. I use the latest BOINC Linux version from the repository (7.16.6) and have not had that problem. If memory serves the easiest fix was to upgrade BOINC to the latest version.

This thread may help you:
10) Message boards : Number crunching : High failure rate (Message 1659)
Posted 21 Feb 2022 by Aurum
WUs are failing at the rate of over 11%. That seems high.
Are others getting similar failure rates?
Is anyone looking into reducing the failure rate?
11) Questions and Answers : Getting started : Problem joining QuChem (Message 1656)
Posted 19 Feb 2022 by Aurum
I had the same thing happen.
Click on your account and then click on Account Keys. It gives you an alternate way to attach a computer. I only had to do this once and then I added the others using BoincTasks.
12) Message boards : Science : NWChemEx (Message 1419)
Posted 12 May 2021 by Aurum
13) Message boards : Number crunching : Checkpoint? (Message 1324)
Posted 21 Jan 2021 by Aurum
Watching QChem running on BoincTasks 1.85 and it's strange. It says it checkpoints every 30 seconds but watching the atop dsk report writes to the drive are very minimal and nothing close to checkpointing every 30 seconds. My BOINC projects are set to Write to disk at most every 600 seconds. If I highlight all QC WUs and select Suspend at Checkpoint they start to suspend although some seem to take a long time to do so. Less than half suspend quickly but the rest still haven't suspended after 30 minutes.
What is the checkpoint interval?
14) Message boards : News : Credits and Gridcoin (Message 1323)
Posted 21 Jan 2021 by Aurum
Gridcoin wallet holders have voted overwhelmingly that they want QuChem whitelisted:
15) Message boards : News : Credits and Gridcoin (Message 1322)
Posted 21 Jan 2021 by Aurum
There are already people complaining on the forum. I'm not sure I can handle a massive flood of people, especially if a lot of people come to the forum to complain. For now, I prefer to take my time and be able to interact with the participants.
Then do what the staff of every other BOINC project do, ignore them.

Gridcoin is asking for a guarantee that there will be calculations to be made, I can't meet that requirement. One can think there are a lot of calculations available. At the moment there are, but it will go very quickly and each series requires a lot of work to be scientifically valorized.
Not one single BOINC project can guarantee a constant supply of WUs. Today GAIA, TN-GRID and SiDock are completely out of work for the second day and have made no statements as to when they expect WUs to be posted. That's the beauty of BOINC, while scientists take time to digest and process the results we can switch over to another project and keep on crunching.
16) Message boards : Science : NWChem vs NWCHem Long (Message 1163)
Posted 1 Nov 2020 by Aurum
...stability is also a concern.

Half of my NWChem Longs are Validation Inconclusive and a fifth are Invalid.
Any hope this will improve???
17) Message boards : Number crunching : Compile for AVX-512 VNNI (Message 1157)
Posted 22 Oct 2020 by Aurum
A free compiler may be available:
18) Message boards : Number crunching : Compile for AVX-512 VNNI (Message 1156)
Posted 21 Oct 2020 by Aurum
Thanks. I sure hope someone compiles this for Linux Mint 20 (Ubuntu) and I'll be glad to test it.
19) Message boards : Number crunching : Compile for AVX-512 VNNI (Message 1151)
Posted 19 Oct 2020 by Aurum
Are the sources available somewhere? Github?

Doctor Google suggests:
20) Message boards : Number crunching : Got any Betas??? (Message 1147)
Posted 16 Oct 2020 by Aurum
With only anecdotal data I'm not seeing anything close to Amdahl's Law. This is what I'm seeing: t(1) ~ t(2) ~ t(4)
Amdahl's Law might say 1.95*t(1) = t(2) or 3.67*t(1) = t(4).

So it's not that t(4) is working on a larger molecule than t(1) but that parallelization isn't working?

Next 20

©2023 Benoit DA MOTA - LERIA, University of Angers, France