• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Info 12th BOINC Pentathlon 2021

StefanR5R

Elite Member
Dec 10, 2016
4,181
4,684
136

https://www.seti-germany.de/boinc_pentathlon/
from 05 May 2021 00:00 UTC to 19 May 2021 00:00 UTC — 5 disciplines, 1 winner

Code:
                      We Th Fr Sa Su Mo Tu We Th Fr Sa Su Mo Tu
                03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18
  SiDock@home   __ __ ▄▄ ▄▄ ▄▄ ▄▄ ▄▄ ▄▄ ▄▄ ▄▄ ▄▄ ▄▄ ▄▄ ▄▄ ▄▄ ▄▄   14 days Marathon
    PrimeGrid          _ __ __ ▄▄ ▄▄ ▄▄ ▄▄ ▄▄                      5 days City Run
Einstein@Home             _ __ __ __ __ ▄▄ ▄▄ ▄▄ ▄▄ ▄▄             5 days Cross Country
     NFS@Home                               _ __ __ ▄▄ ▄▄ ▄▄       3 days Sprint
      WCG MIP   __ __ __ ▄▄                                        1 day  Javelin Throw
                      __ __ __ ▄▄                                  1 day
                            __ __ __ ▄▄                            1 day
                                        __ __ __ ▄▄                1 day
                                                    __ __ __ ▄▄    1 day
                03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18
                      We Th Fr Sa Su Mo Tu We Th Fr Sa Su Mo Tu             (UTC dates)


We won Silver overall! Thank you to all who helped!
Big thanks to the organizers, and to the project administrators who hosted this event.

Marathon at SiDock@homestatsWe won Silver! :-)
Start: Wednesday May 5, 00:00 UTC (Tuesday May 4, 20:00 EDT / 17:00 PDT)​
End: Wednesday May 19, 00:00 UTC (Tuesday May 18, 20:00 EDT / 17:00 PDT)​
  • To create a new account, use the invitation code Crunch_4Science
    Join TeAm AnandTech on our team page.
  • CPU only, for Windows x86-64, Linux x86-64, Mac x86-64
    The Linux application requires glibc 2.27 or later (at least the libm portion of it).
  • Minimum quorum: 2, CreditNew
  • Current average run time: 8+ hours
  • No checkpoints. Therefore, don't suspend a task to disk, leave it in RAM.
  • Reporting deadline: 2 days (cmdock application), 3 days ("cmdoc + zipped input" application) [edit: it's 2 days again for both]
    Hence, work at this contest can begin after Monday May 3, 00:00 UTC (after Sunday May 2, 20:00 EDT / 17:00 PDT) – i.e. download tasks, start computation but suspend network transfers.
    After the stats table at the contest site was initialized, which should be shortly after May 5, 00:00 UTC, enable networking again to let the boinc client upload and report the results, as well as fetch more work.
  • Limit on tasks in progress: 2 per active logical CPU, but no more than 128 per host [edit: the latter limit was removed on May 6]
  • Scheduler server: www.sidock.si
    Upload server: www.sidock.si
SiDock@home is aimed at drug discovery, currently for COVID-19. The application simulates molecular docking.
SiDock@home is looking for ligands – small molecules that can successfully bind to protein targets and modulate a specific process that is crucial for the virus biochemistry. Based on molecular docking, the ideal ligand should be complementary in shape and properties to the binding site of the target biomolecule. (source)

The current target is the envelope protein of the SARS-CoV-2 virus. It is the smallest of the four structural proteins that make up the SARS-CoV-2 virus particle, and it is essential for the virus to infect cells. (source)

City Run at PrimeGridstatsWe won Gold! :-D
Start: Saturday May 8, 00:00 UTC (Friday May 7, 20:00 EDT / 17:00 PDT)​
End: Thursday May 13, 00:00 UTC (Wednesday May 12, 20:00 EDT / 17:00 PDT)​
Best performance is to be had with GPUs, notably NVidia GPUs. Best performing/ most credited application is PPS-Sieve for GPUs on most hardware. Some Turing and Ampere GPUs may perform similarly or possibly better with the application WW for GPUs (Wieferich and Wall-Sun-Sun Prime Search).
  • Minimum quorum: 2, fixed credit: 3,371 (PPS-Sieve), 12,000 (WW) per result
  • Run time: depends a lot on the hardware, but is very consistent on a given GPU.
    E.g. on a GTX 1080Ti, ≈190 s (PPS-Sieve), ≈1240 s (WW).
  • Possible optimizations for PPS-Sieve via app_config: Set 'cmdline' to -m64 on most Nvidia GPUs. Set 'ngpus' to 0.5 if you have slow CPU cores. See pschoefer's post for the syntax; don't hesitate to ask here in our thread for more explanation.
  • Reporting deadline: 7 days (PPS-Sieve), 8 days (WW)
  • Limit on tasks in progress: practically none
  • Scheduler server: www.primegrid.com
    Upload server: www.primegrid.com
Acknowledgment: The better part of this information is a straight copy of pschoefer's info.
The PrimeGrid project is, of course, dedicated to the search for prime numbers, preferably really BIG prime numbers.
Proth Prime Search Sieve (PPS-Sieve)
The Proth Prime Search looks for primes in the form of k*2^n+1. With the condition 2^n > k, these are often called Proth primes. This project also has the added bonus of possibly finding factors of "classical" Fermat numbers or Generalized Fermat numbers. However, the PPS-Sieve subproject does not find primes itself, but is a stage before actual primality tests begin. The sieving application serves to narrow down the search space. While you can't become finder of a prime when you run the sieve application, you are getting a credit bonus for these workunits since you are facilitating and accelerating the subsequent primality tests.
(more info on Proth Prime Search)

Wieferich and Wall-Sun-Sun Prime Search
A Wall–Sun–Sun (or Fibonacci–Wieferich) prime is a prime p > 5 in which p^2 divides a corresponding Fibonacci number. They are named after Donald Dines Wall and twin brothers Zhi-Hong Sun and Zhi-Wei Sun. Drawing on Wall's work, in 1992 the brothers proved that if the first case of Fermat's last theorem was false for a certain prime p, then that p would have to be a Wall–Sun–Sun prime. Although it has been conjectured that infinitely many exist, there are no Wall–Sun–Sun primes known as of yet.
A prime p is a Wieferich prime if p^2 divides 2^(p-1) - 1. They are named after Arthur Wieferich who in 1909 proved that if the first case of Fermat’s last theorem is false for the exponent p, then p satisfies the criteria a^(p-1) = 1 (mod p^2) for a=2. Despite a number of extensive searches, the only known Wieferich primes to date are 1093 and 3511.
(more info on Wieferich and Wall-Sun-Sun Prime Search)

Cross Country at Einstein@HomestatsWe won Silver! :-)
Start: Tuesday May 11, 00:00 UTC (Monday May 10, 20:00 EDT / 17:00 PDT)​
End: Sunday May 16, 00:00 UTC (Saturday May 15, 20:00 EDT / 17:00 PDT)​
Best performance is to be had with GPUs, notably AMD GPUs. Best performing/ most credited application is "Gamma-ray pulsar binary search #1 (GPU)" on most hardware. The following data are applicable to this application but not necessarily to the other Einstein applications.
  • Minimum quorum: 2, fixed credit: 3,465 per result
  • Run time: depends a lot on the hardware, but is very consistent on a given GPU.
    E.g. on a GTX 1080Ti, ≈315 s.
  • Possible optimizations via app_config: Run more than one task at once on a GPU for better utilization. See pschoefer's post for the syntax; don't hesitate to ask here in our thread for more explanation.
  • Reporting deadline: 14 days
  • Limit on tasks in progress: A per-GPU quota may apply, depending on whether or not a host returned results recently. I saw 384 per GPU on a new host.
  • Scheduler server: scheduler.einsteinathome.org
    Upload server: einstein4.aei.uni-hannover.de as far as I have seen, YMMV
Acknowledgment: The better part of this information is a straight copy of pschoefer's info.
Einstein@Home is searching for weak astrophysical signals from spinning neutron stars (often called pulsars) using data from the LIGO gravitational-wave detectors, the Arecibo radio telescope, and the Fermi gamma-ray satellite.
Einstein@Home volunteers have already discovered about fifty new neutron stars and hopefully will find many more.

Einstein@Home's long-term goal is to make the first direct detections of gravitational-wave emission from spinning neutron stars. Gravitational waves were predicted by Albert Einstein a century ago, and were directly seen for the first time on September 14, 2015. This observation of gravitational waves from a pair of merging black holes opens up a new window on the universe, and ushers in a new era in astronomy. To learn more about Einstein@Home, explore the linked publications at the project site.
(source)

Also, check out this video: Searching for Continuous Gravitational Waves

Sprint at NFS@HomestatsWe won Gold! :-D
Start: Saturday May 15, 00:00 UTC (Friday May 14, 20:00 EDT / 17:00 PDT)​
End: Tuesday May 18, 00:00 UTC (Monday May 17, 20:00 EDT / 17:00 PDT)​
Most points-per-day are given to the lasieve5f application. Mind the RAM requirements.
  • Minimum quorum: 1
    Fixed credit:
    130 per result for lasieve5f
    50 per result for lasievef_small
    44 per result for lasievee and lasievee_small
    36 per result for lasieved​
  • Run time: at the order of an hour for lasieve5f. At the order of half an hour…an hour for the other applications – which doesn't make up for their lower points per result.
  • RAM requirements, per thread: 1.25 GB peak but 0.7…0.9 GB typical (lasieve5f), 1…0.8 GB peak (other apps)
  • Reporting deadline: 7 days (lasieve5f), 3.5 days (other apps)
  • Limit on tasks in progress: 100 per active logical CPU
  • Scheduler server: escatter11.fullerton.edu
    Upload server: escatter11.fullerton.edu
Acknowledgment: Quota looked up at pschoefer's info.
Number-theoretical project: Factorization of large integer numbers, lattice sieving step.
As a young school student, you gained your first experience at breaking an integer into prime factors, such as 15 = 3 * 5 or 35 = 5 * 7. NFS@Home is a continuation of that experience, only with integers that are hundreds of digits long.

Integer factorization is interesting from both mathematical and practical perspectives. Mathematically, for instance, the calculation of multiplicative functions in number theory for a particular number require the factors of the number. Likewise, the integer factorization of particular numbers can aid in the proof that an associated number is prime. Practically, many public key algorithms, including the RSA algorithm, rely on the fact that the publicly available modulus cannot be factored.

The numbers which NFS@Home is factoring are chosen from the Cunningham project. Started in 1925, it is one of the oldest continuously ongoing projects in computational number theory. The third edition of the book, published by the American Mathematical Society in 2002, is available as a free download. All results obtained since, including those of NFS@Home, are available on the Cunningham project website.
(source)

Javelin Throw at WCG MIPstatsWe finished in 4th. :-)
WCG MIP = Microbiome Immunity Project at World Community Grid​
Day 1: starts Thursday May 6, 00:00 UTC (Wednesday May 5, 20:00 EDT / 17:00 PDT)​
Day 2: starts Saturday May 8, 00:00 UTC (Friday May 7, 20:00 EDT / 17:00 PDT)​
Day 3: starts Monday May 10, 00:00 UTC (Sunday May 9, 20:00 EDT / 17:00 PDT)​
Day 4: starts Friday May 14, 00:00 UTC (Thursday May 13, 20:00 EDT / 17:00 PDT)​
Day 5: starts Tuesday May 18, 00:00 UTC (Monday May 17, 20:00 EDT / 17:00 PDT)​
The result of the 3rd best day will count as the result of this sub-contest.
The optimum use of our resources would be to have three consistently good throws.​
  • CPU only, for Windows x86-32, Linux x86-32 and x86-64, Mac x86-64
  • Minimum quorum: adaptive (2 initially and later at random, 1 otherwise),
    CreditNew or equivalent
  • Current average CPU time: 2 hours, tends to vary a lot.
    I see 0.5…3 hours CPU time on Xeons.
    The application uses a lot of processor cache, or a lot of memory bandwidth if there is not enough cache. This means, MIP performance depends on how many of these tasks run simultaneously on a given CPU.
  • The initial download is >100 MB large, subsequent downloads are ~14 MB…>50 MB per task. But result files are small.
  • No checkpoints, or scarce checkpoints? WCG's FAQ recommends to enable the option 'Leave applications in memory while suspended' in the device profile, or respectively in boincmgr's Options -> Computing preferences -> Disk and memory -> [x] Leave non-GPU tasks in memory while suspended.
  • Reporting deadline: 10 days generally.
    Occasional double-check tasks and other replica have 5 days deadline.
  • Limit on tasks in progress: low, until the computer is deemed reliable.
    According to SG, >15 validated results turn a computer reliable.
  • Scheduler server: scheduler.worldcommunitygrid.org
    Upload server: upload.worldcommunitygrid.org
Microbiome Immunity Project studies Type 1 diabetes, Crohn’s disease, and ulcerative colitis. The application simulates protein folding for protein structure prediction.
The human microbiome is a collection of up to 30 trillion (million million) cells that coexist with the human cells in our bodies, including bacterial cells. Early findings show that most of the bacteria in the human microbiome are beneficial. However, some are linked to diseases. For example, the microbiome in the human gut has been linked to autoimmune diseases including Type I diabetes (T1D), Crohn’s disease, and ulcerative colitis. These are complex diseases which are affected by both host genetics and gut microbial composition.

The first steps to understanding the microbiome's collective genome and its role is to determine the structure of the protein molecule coded by each gene. Knowing the structure of the proteins will permit further work in discovering protein function and how the proteins interact with other molecules. The Microbiome Immunity Project tackles this problem by using computational protein folding, a process through which computers simulate how a protein 1-dimensional sequence folds into its final 3-dimensional structure. The simulations are using Rosetta, which is developed in David Baker’s lab at the University of Washington and collaborators. (source)

Timeline
May 15, 00:00 UTC: Javelin Throw day 4 announced
May 12, 18:00 UTC: Sprint announced
May 11, 00:00 UTC: Javelin Throw day 4 announced
May 7, 00:00 UTC: Javelin Throw day 3 announced
May 6, 06:00 UTC: Cross Country announced
May 5, 12:00 UTC: City Run announced
May 5, 00:00 UTC: Javelin Throw day 2 announced
May 3, 00:00 UTC: Javelin Throw project, and Javelin Throw day 1 announced
May 1, 13:00 UTC: SiDock@home reduced its deadline for reporting of results from 3 to 2 days
April 30, 00:00 UTC: Marathon project announced


Prior info
Important dates:
April 12 — team registration is open, TBD by team captains (update: the TeAm is registered now)
April 30, 00:00 UTC ( = Thursday, April 29, 20:00 EDT / 17:00 PDT) — announcement of the Marathon project
May 2 — registration closes
May 5, 00:00 UTC ( = Tuesday, May 4, 20:00 EDT / 17:00 PDT) — start of the Marathon
Until then, let's heed this advice from the organizers:
Dedust the computers. Upgrade the home wiring. Stock up on food reserves — we will be stuck in front of our computers for two weeks straight. :-)

TeAm AnandTech's previous Pentathlon discussions: forum link


I shall update this post as new information becomes available.
 
Last edited:

voodoo5_6k

Member
Jan 14, 2021
191
180
86
Good news everyone :D 23 days left... Time to block the calendar...

I'm hoping they'll include a good share of medical projects :)
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
21,560
9,623
136
Well, I am out of electricity ability, but I will upgrade the 7551@1 ghz to a 7601 retail@2.7 ghz. And maybe replace a 3900x with another 5950x....
 

Icecold

Senior member
Nov 15, 2004
878
720
136
I got my 3 raspberry pi's setup a couple weeks ago, so hopefully there are some ARM projects. Oh, and I guess I also have a 5950x sitting here I'll have setup by then :D
 

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
21,560
9,623
136
Time to sell some of the older but newer stuff to us :D
I have tried... I have THREE 14 core Xeons, motherboard and cpu for $250, but no takers. You are welcome to PM me if interested.
 

Skillz

Senior member
Feb 14, 2014
310
282
136
Which X99 boards?

Send me a PM on exactly what ya got. I need to upgrade my Plex server. Currently rocking an ES 10-core Xeon. I'll have to look at what model number it is later. Can't remember.
 

StefanR5R

Elite Member
Dec 10, 2016
4,181
4,684
136
Upgrade the home wiring.
I still need to fix several wall outlets with corroded contacts. They are getting uncomfortably warm when subjected to the load of several computers per circuit.

Stock up on food reserves — we will be stuck in front of our computers for two weeks straight.
I will be stuck in front of the cable modem, to reset it whenever the connection breaks down and does not recover. There has been a wave of such incidents again lately. Well, maybe I can come up with a scripted approach; just need to figure out if wget or the likes can feed the authenticated web interface of the modem/router.

An internet connection which actually works, or if it doesn't is actually serviced by the provider, would probably cost a magnitude more than my lowly home link.
 

voodoo5_6k

Member
Jan 14, 2021
191
180
86
I will be stuck in front of the cable modem, to reset it whenever the connection breaks down and does not recover. There has been a wave of such incidents again lately. Well, maybe I can come up with a scripted approach; just need to figure out if wget or the likes can feed the authenticated web interface of the modem/router.

An internet connection which actually works, or if it doesn't is actually serviced by the provider, would probably cost a magnitude more than my lowly home link.
I'm not so knowledgeable in the cable-modem area, but is this an ISP modem or 3rd party? My situation is not directly comparable, as I have a VDSL2 connection (35b/Annex Q). But when I switched from the ISP router to a bridge modem (VMG3006-D70A) more than a year ago, I had zero connection issues due to my side of the infrastructure. Before that, there were several instances of the ISP router failing and requiring a restart (or being actually defective...), in addition to the issues on other side of the infrastructure. However, some time this year, I'm finally getting FTTH. I'm already curious to see how that'll impact connection stability (no more 3rd party modem option)...
 

Icecold

Senior member
Nov 15, 2004
878
720
136
This may sound absurd, but it could be worth setting up a 4G failover for just the pentathlon if it's really that unreliable. In the US at least, I could get a prepaid phone with a small amount of data for relatively inexpensively to use as a mobile hotspot and then you could use something like an inexpensive router running DDWRT to use for load balancing/failover.
 
  • Like
Reactions: voodoo5_6k

voodoo5_6k

Member
Jan 14, 2021
191
180
86
This may sound absurd, but it could be worth setting up a 4G failover for just the pentathlon if it's really that unreliable. In the US at least, I could get a prepaid phone with a small amount of data for relatively inexpensively to use as a mobile hotspot and then you could use something like an inexpensive router running DDWRT to use for load balancing/failover.
Good idea. Depending on the ISP router model, you could also connect a USB SIM device/module and use that as an alternative gateway.
 

StefanR5R

Elite Member
Dec 10, 2016
4,181
4,684
136
is this an ISP modem or 3rd party?
It's from the provider. In the past, I thought of using an own router to have a purpose of security and privacy, plus maybe features. As yet I haven't thought of it being a way to improve availability. I'll need to look into that.

Edit, before I moved 2 years ago, I had a cable modem from the other ISP back then, plus my own router. Now it is a combined modem/router, and my old router is stashed away. I might as well go searching in which storage box I put it, and try it plugged between the combo modem/router and my LAN.
 
Last edited:

voodoo5_6k

Member
Jan 14, 2021
191
180
86
@StefanR5R, looking into an own router is definitely worth the time, if you're interested into networking stuff. I've used pfSense (that's just my choice, there are of course alternatives) for many years now, and can't think of any reason to go back.

Initially, I had the pfSense router behind the ISP router (drawback: dual NAT), what already provided two benefits. First, security & privacy, obviously. Second, ease of use. Changing the ISP no longer has any impact on my own network. Each ISP router only sees one single client, the pfSense box. It'll never see the network itself. Internally, my network is 10.0.0.0/8, with further subnets for switches, VLANs, access points, PBX, server, WSUS etc. The ISP router config is irrelevant in such a case. Just connect the pfSense box with your entire network to a new ISP's router, and you're done already ;) Well, apart from e.g. switching off the ISP WiFi etc.

Later, I removed the ISP router completely, and used a bridge modem instead. Just reconfigure WAN from DHCP to PPPoE, enter the ISP account & password, done. No more dual NAT. And with strong enough hardware (I have a Supermicro E200-9A with 16GB ECC and 256GB SSD), you can do a lot of stuff. My WAN consists of 30 OpenVPN clients in a gateway group (for clients that should have privacy, also with a separate DNS resolver through the VPN- no leaks...). Other stuff, that needs non-VPN WAN, can have that too. Full control over all firewall rules, with ACLs. DHCP, DNS resolver (no more DNS censoring by the ISP). And so much more.

I won't lie, it is a lot of work and will cost you a lot of time (depending of how deep you want to dive into that). But in my opinion it is definitely worth it.
 
  • Like
Reactions: Icecold

Markfw

CPU Moderator, VC&G Moderator, Elite Member
Super Moderator
May 16, 2002
21,560
9,623
136
It's from the provider. In the past, I thought of using an own router to have a purpose of security and privacy, plus maybe features. As yet I haven't thought of it being a way to improve availability. I'll need to look into that.

Edit, before I moved 2 years ago, I had a cable modem from the other ISP back then, plus my own router. Now it is a combined modem/router, and my old router is stashed away. I might as well go searching in which storage box I put it, and try it plugged between the combo modem/router and my LAN.
My setup is a comcast cable modem/router that supplies the input to my personal router.
 

Icecold

Senior member
Nov 15, 2004
878
720
136
@StefanR5R, looking into an own router is definitely worth the time, if you're interested into networking stuff. I've used pfSense (that's just my choice, there are of course alternatives) for many years now, and can't think of any reason to go back.
I've been using PFSense for the last 10 years or so. I initially switched to it because even a high quality router(such as a WRT54GL running DDWRT at the time.. so maybe it's been more than 10 years) would need rebooted a couple times a year and I wanted as close to 100% uptime as possible. All of the extra features of PFSense have been a bonus, but the main benefit has been never needing to reboot it.
 
  • Like
Reactions: voodoo5_6k

StefanR5R

Elite Member
Dec 10, 2016
4,181
4,684
136
Initially, I had the pfSense router behind the ISP router (drawback: dual NAT),
In my case, there would effectively be no regression: I am already behind dual NAT; it would just become triple NAT.
(There is NAT in the modem/router, and then NAT within the ISP's network.)
 
Last edited:

voodoo5_6k

Member
Jan 14, 2021
191
180
86
[...] but the main benefit has been never needing to reboot it.
So true. My server is also running on a FreeBSD based OS. Rock stable. Plus ZFS. If they were just a little faster with new drivers and other stuff.

In my case, there would effectively be no regression: I am already behind dual NAT; it would just become triple NAT.
(There is NAT in the modem/router, and then NAT within the ISP's network.)
Let us know how/if it works for you, in case you start looking into this.
 

Icecold

Senior member
Nov 15, 2004
878
720
136
Any guesses on how many teams will have to combine together this year to have a shot at beating us or P3D? 2? 3?

I'm guessing 3.5 (3 full teams, some stragglers from other teams to make up the .5)
 
  • Haha
Reactions: Ken g6

ASK THE COMMUNITY