Posts by HAL9000

21) Message boards : Number crunching : Merging Hosts (Message 1840231)
Posted 14 days ago by Profile HAL9000
Post:
I guess it's the differing processor count - How did that happen?


It's a VM

There is a cc_config.xml option to override the number of processors BOINC thinks the system has.
You could try setting that to 6, setting the host name to match the old system, updating, and then see if it will allow a merge.
If that doesn't do it then plan B would be to set the new host to the old host ID. Which is actually a bit of a PITA. Backing up the data folder and dumping it on the new host of the easiest way to keep the old host ID without merging or stabbing BOINC with a stick.
22) Message boards : Number crunching : Panic Mode On (104) Server Problems? (Message 1839815)
Posted 17 days ago by Profile HAL9000
Post:
Is anybody getting any work? I have not received any downloads since about 1:30pm Eastern Standard Time. Nothing is stuck or suspended, but I am just about of work units. Do we need to make any changes at our end because of the changes in the encryption? Just wondering if something wrong at my end.
Thanks.


I seen a problem where I had to tick off SETI@home v8 "YES" to get work on my Ubuntu box, even though it had Accept Other Work and was working fine before.

Thanks for the idea.

I changed:
Run only the selected applications AstroPulse v7: yes
SETI@home v8: no
If no work for selected applications is available, accept work from other applications? yes

- to:
Run only the selected applications AstroPulse v7: yes
SETI@home v8: yes
If no work for selected applications is available, accept work from other applications? yes

- and my PCs get again WUs.


Still the same...

Is this BUG reported to the admins?

http://setiathome.berkeley.edu/forum_thread.php?id=80573&postid=1838907#1838907
23) Message boards : Number crunching : Panic Mode On (104) Server Problems? (Message 1839796)
Posted 17 days ago by Profile HAL9000
Post:
If you checkout the Hosts in the Top 30 you'll find they have similar In Progress to Pending numbers. The difference becomes greater the Faster your Host is compared to the Average Wing-Person.
https://setiathome.berkeley.edu/top_hosts.php

If the User running all those Tesla K80s ever finds the Linux Special App it's all over. Or rather, he'll be all over the front page.

That person certainly has a lot of them, https://setiathome.berkeley.edu/hosts_user.php?userid=9299609. :-O

Cheers.


Yes and it seems the 8 K80's are really in groups of 3, how else do you explain the machine creation times.

8159697 Created 10 Dec 2016, 19:19:59 UTC
8159698 Created 10 Dec 2016, 19:19:59 UTC
8159699 Created 10 Dec 2016, 19:19:59 UTC
8159701 Created 10 Dec 2016, 19:20:02 UTC
8159702 Created 10 Dec 2016, 19:20:02 UTC
8159703 Created 10 Dec 2016, 19:20:02 UTC

Two batches of 3, 3 seconds apart, not sure I could do that!!

Also not all machines are running all the time

5 havent reported since the 10th, 11th and 12th of December .

Seems it could be another "test" using SETI@Home to test new clusters of machines.

Probably instances from AWS or a similar service. Depending on where you live the cost to buy, power, and maintain the hardware yourself can be more vs renting time for some cloud instances.
I have a few systems configured similarly in our development clouds at work and they seem to work pretty well.
24) Message boards : Number crunching : Panic Mode On (104) Server Problems? (Message 1839213)
Posted 19 days ago by Profile HAL9000
Post:
I noticed my pendings shot up a few weeks ago, but several of my machines received a lot of shorties that the wingmates haven't returned yet around that time.
25) Message boards : Number crunching : Open Beta test: SoG for NVidia, Lunatics v0.45 - Beta6 (RC again) (Message 1839210)
Posted 19 days ago by Profile HAL9000
Post:
With v0.45 installed, is there a way to download a newer app used on Main, then edit the app_info.xml from v0.45 accordingly?


You certainly can. I believe Ramster posted the files here somewhere.

Just put the new files in your folder, and carefully change the version numbers in your app_info
How about copying the requisite files from the Seti Beta directory to the Seti Main directory?  I think those files are

    ♦   mb_cmdline-8.22_windows_intel__opencl_nvidia_SoG.txt
    ♦   MultiBeam_Kernels_r3584.cl
    ♦   MultiBeam_Kernels_r3584.cl_GeForceGTX950.bin_V7_SoG_35900
    ♦   r3584_IntelRCoreTMi76700KCPU400GHz_x86.wisdom
    ♦   setiathome_8.22_windows_intelx86__opencl_nvidia_SoG.exe


Two of those files
MultiBeam_Kernels_r3584.cl_GeForceGTX950.bin_V7_SoG_35900
r3584_IntelRCoreTMi76700KCPU400GHz_x86.wisdom
Are created when you run the app.
26) Message boards : Number crunching : exclude GPU in cc_config (Message 1839152)
Posted 20 days ago by Profile HAL9000
Post:
That won't work because the card will attempt to pull the power it needs from the PCIe bus. Which leads to caterwauling GPU fans and quick system crash. Know from experience what happens when you forget to plug in those PCIe cables.


For AMD Fiji based cards, the card is disabled if the power cable is removed. I ran my triple produo block like this until fixed drivers were released. It worked perfectly fine, with only 1 power plug removed the card was completely disabled.

What they described also goes against the PCIe specification. Sounds like whichever manufacture they had that experience with should be avoided if they can't be bothered to fulls a spec.
27) Message boards : Number crunching : Windows 10 - Yea or Nay? (Message 1838285)
Posted 23 days ago by Profile HAL9000
Post:
I have a HP PC which started in Windows 8, then 8.1, then 10. I am getting Windows 10 updates every month, which happen by night, so I cannot suspend my BOINC projects before a reboot. This disturbs CERN projects using VirtualBox.
Tullio

You can use task scheduler and schedule BOINC to stop before the update starts.
28) Message boards : Number crunching : Panic Mode On (104) Server Problems? (Message 1838172)
Posted 23 days ago by Profile HAL9000
Post:
Well today's 9.5hr long outrage saw my 1060's out of work for 6 minutes this time before the servers came back online, another few more minutes and I would've grabbed some backup work for them (so much for that 3-4hr notice these days). :-(

Cheers.

When were those exactly? It seems like 5-6 has been more typical for the past few years at least.
29) Message boards : Number crunching : Portable BOINC Client (Message 1837693)
Posted 27 days ago by Profile HAL9000
Post:
I tried PE but failed in integrating OpenCL.

Which version of PE were you using? At one point they change the command for integrating drivers in the AIK.
For versions of Windows PE before 3.0
Peimg /inf="C:\driver\driver.inf" "D:\PE3_1_x64\sources\mount\Windows"
For versions of Windows PE after 3.0
Dism /image:D:\PE3_1_x64\sources\mount /Add-Driver /driver:C:\driver /recurse
30) Message boards : Number crunching : Panic Mode On (104) Server Problems? (Message 1837566)
Posted 27 days ago by Profile HAL9000
Post:
Some of the "biggest, baddest" supercomputers means to me most likely the top 10 or 20.

I don't know about you lot but I cannot see why seti@home shouldn't OFFICIALLY rank at No.218 in the top 500 super computers list, I think the project has deserved it don't you?

BOINC combined brings together about 311,742 active participants and 834,343 active computers (hosts) worldwide processing on average 11.747 PetaFLOPS as of 24 March 2016.
So what position should Boinc be in the top 500?

I suppose the strict interpretation of the rules for the top 500 Supercomputers would mean in practice that the computer in question was in it's entirety in one physical place. Seti and BOINC by their very nature are "distributed" therefore they wouldn't match that criteria. But it would be pretty mean not to include them would you not think?

The Top500 list is based on a specific benchmark https://www.top500.org/project/linpack/.
Given the value that BOINC generates for FLOPs can be rather sketchy I don't imagine any organization would use its output as a legitimate benchmark. It is not even useful to compare different BOINC projects.

Generally Supercomputers are in one location as the high speed interconnects they use tend to be limited in maximum length. I'm not sure that how far apart the computing nodes are is in the definition of a Supercomputer. More so that the nodes are interconnected.
Where in a system such as BOINC each node is independent and there is no inter-node communication.
31) Message boards : Number crunching : Panic Mode On (104) Server Problems? (Message 1837536)
Posted 28 days ago by Profile HAL9000
Post:
Well, as it stands right now, Seti is the most successful distributed computing project on the planet.
Amassing a power, as far as I know, greater than some of the biggest baddest supercomputers on the planet.

Not quite Mark :-)

With over 145,000 active computers in the system (1.4 million total) in 233 countries, as of 23 June 2013, SETI@home had the ability to compute over 668 teraFLOPS. For comparison, the Tianhe-2 computer, which as of 23 June 2013 was the world's fastest supercomputer, was able to compute 33.86 petaFLOPS (approximately 50 times greater).


. . But at those numbers it puts us in the ballpark :), only 50 times greater is a good achievement for us :).

Stephen

.

The references supercomputer was actually#1 as of 2015-11. As of 2016-06 it was overtaken by one that benchmarks nearly 3 times faster. https://www.top500.org/list/2016/11/
If the benchmark figures of the Top500 list were comparable to our computing numbers then we would be sitting at #218 on the Top500 list.
Even though the numbers are not directly comparable I would say the kitteh persons statement "Amassing a power, as far as I know, greater than some of the biggest baddest supercomputers on the planet." is most certainly true. SETI@home has a very respectable amount of computing power supporting it.
32) Message boards : Number crunching : Building a 32 thread xeon system doesn't need to cost a lot (Message 1837328)
Posted 29 days ago by Profile HAL9000
Post:
http://bfy.tw/94h7

Umm.. I already have SP1 installed, but thanks for the Google link! lol

I was looking for the one that I remember people referencing when they had massively long searching for updates when doing a WIndows Update in Win 7, taking a day or more. I know it was discussed here, I just couldn't find it.

Ah I had misread your post & thought you were having issues finding THE SP for Win 7. It makes more sense that you were looking for the rollup packages.
If I need a specific update of some kind to run an app I'll grab that KB by itself from the download center. Otherwise for my machines I go with 7 SP1 & don't bother with any updates.
33) Message boards : Number crunching : Building a 32 thread xeon system doesn't need to cost a lot (Message 1837264)
Posted 29 days ago by Profile HAL9000
Post:
http://bfy.tw/94h7
34) Message boards : Number crunching : Portable BOINC Client (Message 1837113)
Posted 21 Dec 2016 by Profile HAL9000
Post:
OK, I will see if I can make Windows PE images with BOINC installed, and perhaps the drivers required. I will also see if I can do both x86 and x64 versions

I find it is better to not make BOINC part of the PE image. Instead I place it on the flash drive that PE boots from.
Then if anything BOINC related needs to change I don't have to redo everything in the AIK.
35) Message boards : Number crunching : Portable BOINC Client (Message 1837106)
Posted 20 Dec 2016 by Profile HAL9000
Post:
OK I understand, for your portable BOINC you want the OS and everything. So you can just stick in a flash drive and boot the system from it?

I've made Windows PE images for that. Which really only consisted of installing the GPU drivers I needed and modifying the startup so it would launch BOINC from the flash drive.
36) Message boards : Number crunching : Portable BOINC Client (Message 1836983)
Posted 19 Dec 2016 by Profile HAL9000
Post:
Great ! Thx to HAL9000

It worked, not perfectly (no data loss) , but it worked so far.

If you want to have your GUI settings as well.
1. Set BOINC Manager like you want.
2. Close BOINC Manager (so we are sure it writes its settings to the reg)
3. Export HKEY_CURRENT_USER\Software\Space Sciences Laboratory, U.C. Berkeley
Then you can import your settings when you move to a new machine.

Besides clicking a .reg file to import it you can use the command line, reg add boinc_settings.reg, to import your saved settings.
37) Message boards : Number crunching : Portable BOINC Client (Message 1836932)
Posted 19 Dec 2016 by Profile HAL9000
Post:
Hey folks, a portable (useable,new) BOINC client can be found , where ?

Thanks for helping

Greetings

For Windows:
1. Install BOINC.
2. Copy application folder somewhere to be used later.
3. Uninstall BOINC.
4. Remove all registry entries for Space Sciences Laboratory, U.C. Berkeley.
5. Place BOINC app files in desired location. Such as F:\BOINC
6. Create data folder. Such as F:\BOINC\Data or F:\Data
Then when starting BOINC you can use a command line options such as boinc.exe --dir %~d0\BOINC\Data or boinc.exe --dir %~d0\Data to tell BOINC where the data folder is located.

If you don't specify a data folder BOINC will use the app folder as the data folder. Which is how I run it on most of my systems. As I don't actually Install BOINC. I just deploy the app folder, normally to D:\BOINC or E:\BOINC, and then run everything from there.
My standard .bat I use to launch BOINC
pushd %~dp0
boinc.exe --detach
start boincmgr.exe /s
38) Message boards : Number crunching : Building a 32 thread xeon system doesn't need to cost a lot (Message 1836877)
Posted 19 Dec 2016 by Profile HAL9000
Post:
I've never looked for CAD files. The most I've gone for is the specification diagram that shows where everything is located. Perhaps you could use that to make your CAD file?

For the Dual E5's in the Core X9 case I used a few of these stands-offs where the SSI-EEB didn't match up wit the E-ATX holes in the MB tray.


For my dual LGA771 system I ended up finding a backplate that let me secure the heatsinks. So I didn't need to find a MB try or case to mount the MB.
39) Message boards : Number crunching : BOINC 7.6.33 improved AMD GPU detection (Message 1836776)
Posted 18 Dec 2016 by Profile HAL9000
Post:
Jason wrote:
Though I'm pretty sure they don't have the development resources.
Yup.

Jeff wrote:
Not exactly promoting it as an important reason to upgrade, are they?
Who 'they'? Jord (Ageless) wrote those notes, and I've just added the line about the error code fix. BOINC will only improve if volunteers contribute to the process.

I believe Ageless had recently messaged D.A. asking that 7.6.33 be put up for download. Even if it were listed as a dev version.

Also the OpenCL name detection was introduced in build 7.6.23, which was released nearly a year ago.
40) Message boards : Number crunching : AMD processors? (Message 1836694)
Posted 18 Dec 2016 by Profile HAL9000
Post:
Take a look here.
Excerpt:
    ♦  The 8 core, 16 thread RYZEN CPU has a TDP of 95 watts, 45 watts less than its direct Intel competitor.
    ♦  The 8 core 16 thread RYZEN CPU competes with Intel’s fastest 8 core 16 thread offering, the $1199 Broadwell-E. i7 6900K, in terms of features and performance and exceeds it in power efficiency.


Many times in the past similar comparisons have been done, and not borne fruit on general release production hardware. Given the lack of hype over the current release, i'm very hopeful that this time they will be true comparisons.
But i'll wait to see what publicly purchasable hardware does before I get my hopes up too much.

More important TDP != power consumption it is Thermal Design Power, or how much heat the cooling system is expected to handle.
Depending on the designed thermal limits the TDP for a product can be adjusted up or down. Often the same product is marketed under multiple TDPs for different scenarios.

AMD has a long history of making claims and not actually delivering. So we will not really know the facts until reviewers get a chance look it over.


Previous 20 · Next 20


 
©2017 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.