Posts by Chilean

1) Message boards : Number crunching : Good bye Rosetta !!! (Message 91023)
Posted 13 Aug 2019 by Profile Chilean
Post:
Sorry, I leave this projet.
Admin publish information as links. Nice to be informed about . The latest is three years old !!!
The only recent (16july19 by admin) is about "audacious projects". Nice, but is not a part of Boinc.
And this link send us to donation requests !

And about projects who has nothing to do with sciencetific research, as educating people, sexual abuse,...
Of course there are also very important subjects, but
Does we think they not need the computing power of Boinc ?

But what they do to solve the problems of a lot of users ? Abnormal running, app crash,......
Nothing ! Or even good as nothing.Never answer about technical problems. The admin fully ignare the Boinc cruncher.
They just ignore us !


Bye bye.
2) Message boards : Number crunching : Rosetta v4.08 x86_64-pc-linux-gnu or Rosetta v4.07 i686-pc-linux-gnu (Message 90396)
Posted 19 Feb 2019 by Profile Chilean
Post:
Well, you can't execute the benchmark since that are all your completed WUs. Whatever you did, will sort itself out with time.
You can force 64-bit only with <no_alt_platform> in cc_config.xml, but I would wait with that and see if the 64-bit application is really faster than the 32-bit, that's not always the case. Just watch the GFLOPS values. If it really is faster, the server will send most WUs to that application anyway, so no really need to do anything.


I was surprised to see that Rosetta 4.08 is the only 64-bit binary for Linux. Both copies of Minirosetta are 32-bit.
The 32-bit binaries end up with a smaller code footprint and pass parameters on the stack. The parameter list quickly spills to the stack, but they will be in the L1 cache which has a 1-cycle access time.

A 64-bt version will always be faster if compiled properly, unless you aggressively inline functions. The larger code footprint is causing front-end icache miss stalls.
I want to see how hard it is to modify the source to redefine their 3-dimensional "vector" object into a 4-dimensional vector so they can use packed SSE or AVX. Right now all Rosetta computation uses scalar operations.

thanks again.


I have nothing to add, but I started reading your post and when it started going over my head I went "yo this dude knows his stuff...", then I saw the username and said "of course, it's rjs5" lol
3) Message boards : Number crunching : GPU calculation (Message 90270)
Posted 29 Jan 2019 by Profile Chilean
Post:
Rosetta "could" run on GPU if each GPU thread could handle a single model trajectory. But apparently each trajectory uses a lot of memory.
4) Message boards : Number crunching : AMD 1950x Threadripper performance (Message 90250)
Posted 23 Jan 2019 by Profile Chilean
Post:
I might throttle back to maybe 75% or 24 hyper-threads. See what happens. I went out and bought 2x16 GBs of DDR4 memory. So the system seems to recognize all 48 GBs of memory. I do not see much of a change in BIONC usage of memory. With 32 Active simulations, my memory foot print is around 13 GBs.


You only had 1 stick before?

You now have Dual Channel at least.
5) Message boards : Number crunching : AMD 1950x Threadripper performance (Message 90239)
Posted 21 Jan 2019 by Profile Chilean
Post:
I fiddled with that setting. It seems to really cut my real cpu utilization. I set in the manager under options, computing preference, use at most 50% of the CPUs. I thought that would have run close to 100% of the 16 cores, but it seems to cut the real cpu utilization in half. Makes my brain hurt. I maybe making a run to MIcrocenter soon. Will ponder more memory, but it seems to do reasonably well with 16 GBs of RAM.


This is expected. If I half the CPU core usage % in BOINC Manager, I'll see 50% total CPU usage and the workload is distributed evenly in the threads available. To Windows, you "do" have 32 cores... but in real life you don't really. You simply have 32 threads, but 16 cores. So 50% on thread 0 and thread 1 should mean 100% of core 0. Less heat, more cache hits, less RAM usage.

You can maybe run it at 75%. That's what I do. Makes running BOINC completely unnoticeable.
6) Message boards : Cafe Rosetta : Point a newbie at the right reading material (Message 90228)
Posted 19 Jan 2019 by Profile Chilean
Post:
Welcome aboard.

A few things:

  • BOINC is the "managing software" that allows projects such as Rosetta@Home or World Community grid from IBM to run their Workunits (WUs)
  • The developers of BOINC aren't necessarily involved in the projects that ruin on BOINC.
  • Credits are simply for "friendly competition" and keeping track of how much computation you've donated.
  • RAC is the "daily average" of credits you get. It takes around a week or two for this number to stabilize. It helps determine how "fast" you're going. Total credit is the total mileage.
  • It's good tip to not compare credits from one project to another due to the differences in how they calculate them. Rosetta gives you credit based on the number of models you create in each WU.


I don't know how updated this wiki is, but here ya go: https://en.wikipedia.org/wiki/BOINC_Credit_System

There are tons of teams. Just go to https://boinc.bakerlab.org/rosetta/team.php to search. Or you can join the one I manage! (click my signature icon)
Teams can be individual to a certain project or they can be BOINC-wide (as is mine)... meaning it is present in every BOINC project available.

Again, welcome aboard!

7) Message boards : Number crunching : AMD 1950x Threadripper performance (Message 90215)
Posted 16 Jan 2019 by Profile Chilean
Post:
It's a beast - looks like it's getting approx 845 credits per core, per day, so that'd give 27k if it's running 24/7 but it'll take a week or so to get there. That's insane throughput for a single CPU.

D


Seems I was incredibly off lol

The chip is the about the size of a business card and really requires a cooler built for it's form factor. I'm sure somebody is going to knock the power usage, but I'm 100+% solar so the sun is helping cure cancer....


One of my life goals is to build a crunching farm using nothing but solar or wind.
8) Message boards : Number crunching : AMD 1950x Threadripper performance (Message 90211)
Posted 15 Jan 2019 by Profile Chilean
Post:
It take time for the RAC to "reach it's top". Seems you've only been crunching for a day or two. Give it a week. I estimate around 6-7k depending if it runs 24/7 or not.
9) Message boards : Number crunching : tasks "aborted by project" with "result is no longer usable" (Message 90207)
Posted 15 Jan 2019 by Profile Chilean
Post:
To me these seem like WUs that got cancelled by the server that were running on your machine.

I too have a fair amount of these errors between the 14th and 15th of this month.

It's the project's fault. I'd wager someone realized an unnecessary batch was released and "recalled" them to avoid any further wasted computation from the volunteers.
10) Message boards : Rosetta@home Science : New article in "Nature" co-authored by Dr. Baker (Message 90188)
Posted 11 Jan 2019 by Profile Chilean
Post:
There is a new article in Nature co-authored by Dr. Baker:

De novo design of potent and selective mimics of IL-2 and IL-15


This is actually very cool and potent. De novo created, human-made protein that's selective to certain receptors.

We usually screen nature made structures. Sometimes we synthesize them, but now we might be able to easily create them following a template. We need XYZ 3D conformation to activate Q receptor that stops the uncontrolled growth of cancer cells in Prostate cancer.

Run XYZ 3D conformation in Rosetta "backwards" -> Output is amino acid sequence -> Synthesize protein following Rosetta's output -> Inject into patient -> Hope this protein doesn't activate other receptors in the body and wish for the best.

I mean it's never easy, but it's miles ahead of what we have now.

Very interesting and exciting.
11) Message boards : Number crunching : Validator problem? (Message 90167)
Posted 8 Jan 2019 by Profile Chilean
Post:
TN-Grid is getting to be my go-to project when nobody else has work. It is not yet listed by BOINC Manager, as it is still relatively new.
http://gene.disi.unitn.it/test/


Looks good. It even has that "poorly" designed website common in cutting edge scientific project sites (to me, the worser the website is, the more legit the software/project lol).

What about WCG? I use that as my back-up.
12) Message boards : Cafe Rosetta : CASP 13 (Message 90089)
Posted 27 Dec 2018 by Profile Chilean
Post:
We're crunching these, right?
13) Message boards : Number crunching : Ryzen vs Intel Performance (Message 90083)
Posted 26 Dec 2018 by Profile Chilean
Post:
There haven't been many posts in here. Sorry for the dead thread post and all...

The Ryzen 2600X is producing about 10,500 credits per day.

What seems to matter a lot are RAM speed and timings. I'm testing those now on the AMD systems. Going from 2133Mhz CL 16 to 2666Mhz CL16 produced almost a 5% gain in average credits over a 9-day period. I'm testing 3000Mhz at CL 15 now.

The Intel 6700k with 3000Mhz 15CL RAM is doing around 8,750 per day.


You are increasing speed AND tightening the timings at the same time. Ugh, I really want to build my own setup with premium RAM and play like this :(

I want to wait until the 3rd generation Ryzen tho.
14) Message boards : Cafe Rosetta : Live up to 20 years longer, healthy, drug-free, watch these (Message 89639)
Posted 26 Sep 2018 by Profile Chilean
Post:
Eating high amounts of processed red meat "might" increase the risk of cancer (apparently because of nitrate intake?), but being vegan = cures cancer definitely does not align with the current scientific evidence and literature.

The body doesn't fight fat molecules. It creates them. Now, excess amounts of fat "molecules" does disrupt the metabolic/endocrine system leading to diabetes among other diseases. And this could happen even if you are vegan, make no mistake... it's all a matter of calorie intake vs burning.

Eating vegetables is healthy. But healthy does not mean a cure for cancer. The body most of the time can't fight cancer not because it's "too much for it because it's fighting [insert youtube trending toxic molecule]" but because "it" doesn't know it's there because the cancer is YOU YOURSELF growing uncontrollably. It isn't a foreign body. Which is why antibody therapy is being researched, to let your body know there's a malignant "you" and to attack it. Which is why chemotherapy, the current gold standard against cancer is so toxic, because it attacks everything (specifically DNA reproduction to kill fast-dividing cells).

So folks, eat your veggies and get screened every so often if you have a family history.
15) Message boards : Number crunching : For the betterment of BOINC (Message 89634)
Posted 25 Sep 2018 by Profile Chilean
Post:
Since many of you seem interested, and mystified by the credit system at R@h, I'll try to clarify a few points. Some other projects simply use CPU seconds, or FLOPS of a task to issue credit. The problem with such a system is that it doesn't reward machines that achieve more work per second due to CPU cache, and memory available. It also depends upon the BOINC Manager to track and report the FLOPS expended on a task, and so some users began to falsify the FLOPS benchmarks and thus the FLOPS of their reported results.

To avoid these problems and design a credit system that would work going forward through new generations of hardware and capabilities, a system was adopted whereby credit granted is based on what OTHER user's machines have claimed for how difficult it is to complete each decoy (or model) of a task. The average of the results reported before you is used, rather than an arbitrary benchmark. So the credit granted does not favor a specific CPU type or operating system. It truly is based on completed work. Some tasks use different algorithms to compute their results. Assessing the effectiveness of a variety of algorithms is how the science progresses. This results in variations on computing resources required to complete them. Each type of task self-adjusts the credit it awards.

This makes a credit system that is great at reflecting the value of work to the Project Team, but makes it more difficult to benchmark your own machines. To do that well, you would have to run the same task, from the same starting seed for each benchmark run, and then compare CPU and wall-clock time, as well as the number of decoys produced. The BOINC Manager doesn't have a means of easily supporting this type of usage. Running tasks for other projects or a different class of proteins only tells you about those things. One machine might be faster than another at one type of work, and not with another type. The R@h credit system reflects the value of the outcome, without bias.


If that's the case than you should not run this project using a 16 core, 32 thread Threadripper as it is a waste of money and energy because it is being outperformed by a factor of 2 by a single 2 core Intel(R) Celeron(R) CPU G1620 @ 2.70GHz [Family 6 Model 58 Stepping 9].

As you can see here this host https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=3391974 is ranked at 45 and is producing slightly more in Rosetta than a 22 core, 44 thread Intel(R) Xeon(R) CPU E5-2696 v4 @ 2.20GHz [Family 6 Model 79 Stepping 1]


Thank you for the clarification.


There's definitely multiple computers merged into that host. There's no way a celeron could output that much RAC. Maybe a network (school?) composed of multiple computers with the same configuration (same CPU, RAM, HDD, OS, etc). It happens to me regularly when I spin up VMs with similar OS and hardware... BOINC thinks it's the same machine.

Not sure why people merge multiple hosts into one... (?)

2k RAC is what you get with a decently fast i5. A fast i7 could maybe pull 5k depending on the cores. That's my "thumb benchmark" I have for Rosetta. My 24/48 core/thread EPYC from AMD pulls almost 20k when sharing resources with World Community Grid... so it should pull about 30-40k running Rosetta exclusively.

Edit: I just realized my EPYC VM made it into the top 30 hosts lol.

Edit2: If you check the WUs of the celeron host you linked, you'll notice it's impossible for a 2 thread machine to report so many WUs a day consistently given that it takes 30k seconds each WU. I counted about 80 WUs returned just on the 25th of September... you'd need about 13 "2 threaded" CPUs to get so many WUs crunched per day @ 8 hours each WU.
16) Message boards : Number crunching : Exporting result data from website? (Message 89253)
Posted 10 Jul 2018 by Profile Chilean
Post:
Managed to do it with Excel Queries using VBA.

I have to do about 50 queries in total for a single host because each table shows only 20 results (hardcoded in the results.php according to the GitHub code...), but... that's what for loops are for I suppose lol
17) Message boards : Number crunching : Exporting result data from website? (Message 89247)
Posted 9 Jul 2018 by Profile Chilean
Post:
I'm more interested in all of the results of a particular host.

I looked at the wiki, thinking there might be a similar thing to what you suggested but instead of result ID, a host ID. But didn't find anything :(
18) Message boards : Number crunching : Exporting result data from website? (Message 89240)
Posted 8 Jul 2018 by Profile Chilean
Post:
Is there a way to easily export the results from a host? The table only shows 20 results at a time... it'd be extremely tedious to copy and paste 20 results at a time to Excel.

Is it possible to force the site to show 100 results at a time or so? Or maybe download everything in a CSV format?
19) Message boards : Cafe Rosetta : CANCER DATA INITIATIVE (Message 87758)
Posted 27 Nov 2017 by Profile Chilean
Post:
Do they take data outside the US?

Do you have to have cancer in order for your data to be helpful?
20) Message boards : Rosetta@home Science : Clean Energy research (Message 87542)
Posted 19 Oct 2017 by Profile Chilean
Post:
Apparently, the Rosetta software can help you build nano-structures (protein-based structures?) for more functions than just drugs... such as the fentanyl sensor recently published on the homepage. This "material" side of Rosetta's potential could potentially (ha!) be used in Clean Energy technology and what not. At least that's what I understand about the subject.


Next 20



©2019 University of Washington
http://www.bakerlab.org