Posts by MarkJ

log in
1) Message boards : News : Running out of workunits (Message 1601399)
Posted 15 Nov 2014 by Profile MarkJProject donor
Thanks for the post Matt. Whatever happened to getting data from the Parkes radio telescope?

Also is the GBT a regular feed or will it suffer from the same funding/being off-line issues that Arecibo has?

2) Message boards : Number crunching : Your thoughts on the upcoming Haswell E CPUs (Message 1596393)
Posted 3 Nov 2014 by Profile MarkJProject donor
I've ordered a couple of rigs with the following specs. Memory ended up on back order so waiting for it before the shop can complete the build.

Asus X99-A motherboard
16Gb Kingston DDR4 @2133Mhz (4x4Gb)
Intel 5820k CPU
Corsair H80 water cooling
2 x GTX750Ti cards
Intel 520 series SSD (120Gb)
1Tb WD black HDD
Fractal designs ARC midi case
Seasonic modular 750w power supply
No-name DVD-ROM

The price for the next chip up would have added about $500 (AUD). The top of the line chip was about $1100.

Even Intel recommend water cooling, hence the H80. I replace the Corsair fans with Noctua ones as they don't make as much noise and last. Had a corsair fan seize up on a previous rig.
3) Message boards : Number crunching : gpu tasks - don't got any (Message 1590666)
Posted 23 Oct 2014 by Profile MarkJProject donor
Look in the BOINC event log and see what it's got to say. At the beginning it should say it's detected the cpu and GPU and driver version etc. later on there should be a message about requesting work, which hopefully will say it's asking for GPU work. It would help if you could post the startup messages about the cpu and GPU and the requesting work messages.

To get to the BOINC event log you can click on advanced (on the menu bar in BOINC manager) -> Event log
4) Message boards : Number crunching : Flash Drive for Boinic projects (Message 1588502)
Posted 17 Oct 2014 by Profile MarkJProject donor
I had to replace all the SD cards used on my Raspberry Pi's. They are supposedly rated for 100k writes. I got about a year out of them.

In the BOINC settings you can adjust the "write to disk" frequency. I currently have them to 600 seconds.
5) Message boards : Number crunching : all these different versions for AP cpu's? (Message 1586211)
Posted 13 Oct 2014 by Profile MarkJProject donor
No we're not testing them, that's already been done on Seti Beta.

As you noted they refer to the different CPU instructions. Some people may have older machines that don't support SSE2 for example, or it may be that the SSE2 app is slower on their particular machine.

As a general rule go for the highest instruction set app that your CPU can support, but it doesn't always mean its faster.
6) Message boards : Number crunching : headless computer? (Message 1575363)
Posted 21 Sep 2014 by Profile MarkJProject donor
I run a bunch of Raspberry Pi's headless. They run Debian Linux and I just ssh into them to do stuff and BOINCtasks running on windows laptop to keep an eye on them.

For my windows machines I use a KVM and have a single monitor/keyboard between a bunch of them. I use BOINCtasks to keep an eye on them as well. With the windows boxes you get the windows updates and driver updates to apply but that's typically once a month.
7) Message boards : Number crunching : crossfire or sli? (Message 1572631)
Posted 16 Sep 2014 by Profile MarkJProject donor
I believe CUDA 6 adds support for using cards in SLI mode (ie treating them as a single larger GPU). It may take some coding and I don't think anyone has produced a CUDA 6 Multi Beam app (yet). Jason would probably be the person to ask.

XT Libraries: Automatic Scaling to Multiple GPUs

CUDA 6 introduces XT Library interfaces which provide automatic scaling of cuBLAS level 3 and 2D/3D cuFFT routines to 2 or more GPUs. This means that if you have one or more dual-GPU accelerator cards in your workstation or cluster node, you can automatically take advantage of them for intensive FFTs and matrix-matrix multiplication. cuBLAS XT also enables multiplication of matrices that are too large to fit in the memory of a single GPU, because it operates directly on matrices allocated in CPU memory, tiling the matrix and overlapping computation with memory transfers. The result is linear scaling of very large GEMM operations to multiple GPUs, as the following figure shows.
8) Message boards : Number crunching : Fundraising time again...... (Message 1562981)
Posted 27 Aug 2014 by Profile MarkJProject donor
Thank you for your gift of $150.00 via Visa on 08/27/2014. Your gift will benefit the following area:

SETI@home - $150.00

Your confirmation number is: 136478
9) Message boards : Number crunching : ASIC computers (Message 1540667)
Posted 13 Jul 2014 by Profile MarkJProject donor
There was also this message thread.
10) Message boards : Number crunching : 120mm fans with ~ 59CFM / 100m3/h (installed) (Message 1496981)
Posted 29 Mar 2014 by Profile MarkJProject donor
I've replaced most of my failed ones with Noctua. They cost a bit more but they are quiet and more importantly reliable.

I had some Corsair ones that came with a H80 kit and they sound like a jet engine taking off.

Got some Fractal design ones (came with their case) and had them fail after about 12 months, replaced them with Noctua.

Also had some CoolerMaster ones that died. Totally seized up on a CPU cooler, ended up replacing them with Noctuas.

Have yet to have a Noctua fail and some of mine have been going for 3 years.
11) Message boards : Number crunching : app_config question (Message 1491778)
Posted 20 Mar 2014 by Profile MarkJProject donor
The lunatics apps all read command files, so no need to do it in app_config. You'd edit the txt files for each app, if you wanted to pass a value to the GPU app you'd put it into the txt file and leave the CPU version of the txt file blank.
12) Message boards : Number crunching : The (Preliminary and Premature) Green Bank SERENDIP Fundraiser - Help SETI@Home Build a New (non-Arecibo) Receiver! (Message 1491760)
Posted 20 Mar 2014 by Profile MarkJProject donor
Your order is complete! Your order number is 1330.

$490 so it should be down to $4000 left now, except isn't updating the amount
13) Message boards : Number crunching : CPU tasks run slow after upgrade to 7.2.42 (Message 1491084)
Posted 19 Mar 2014 by Profile MarkJProject donor
Are you sure it's not throttling? There are settings to do with suspending work when CPU usage exceeds a percentage. Have you checked these? Given BOINC itself doesn't do any crunching it sounds like it may be suspending and unsuspending. Messages should be in your BOINC event log if it is.
14) Message boards : Number crunching : Set of new CPU AstroPulse builds (Message 1477586)
Posted 15 Feb 2014 by Profile MarkJProject donor
Can someone explain the file names. The version number, SSE, etc I get, but what are the no_opt and the noarch? I am guessing no_opt means no optimisations.

Also was there an all_opt for x64? It looks like the only x64 was a vanilla version. Does avx yield anything?
15) Message boards : Number crunching : Will BOINC/ SETI run on a Beowulf Cluster ? (Message 1470199)
Posted 29 Jan 2014 by Profile MarkJProject donor
I haven't tried it myself but I believe there are hooks to use BOINC as a fill-in process under a condor cluster. BOINC itself doesn't support MPI that most clusters use to communicate with worker nodes in the cluster. Jason is/was looking to doing something in this area (superdreadnought).

You could install BOINC on each and use BOINCtasks to monitor them however for OS and other software updates etc you'd probably have to SSH into each one which is what I do with my bramble every so often.
16) Message boards : Number crunching : Stop network activity specific days? (Message 1459992)
Posted 2 Jan 2014 by Profile MarkJProject donor
You could set up a command script to use BOINCCMD (the command line tool for BOINC) to enable and disable network activity, so you'd have 2 scripts. Then use Task Scheduler if using Windows (or cron if Linux) to run it as desired.

Doco for BOINCCMD can be found here

Enable script
Cd "\Program Files\boinc"
BOINCCMD --set_network_mode always

Disable script
Cd "\Program Files\boinc"
BOINCCMD --set_network_mode never
17) Message boards : News : Our Winter 2013 fundraising drive is underway. (Message 1459452)
Posted 1 Jan 2014 by Profile MarkJProject donor
Thank you for your gift of $200.00 via Visa on 12/31/2013.
Your gift will benefit the following area:

SETI@home - $200.00

Your confirmation number is: 128828
18) Message boards : Number crunching : BONIC on RK3188 running PicUntu (Message 1446973)
Posted 26 Nov 2013 by Profile MarkJProject donor
They (the Pis) are classed as ARM hard float.

Actually its: arm-unknown-linux-gnueabihf
19) Message boards : Number crunching : BONIC on RK3188 running PicUntu (Message 1446957)
Posted 26 Nov 2013 by Profile MarkJProject donor
There is a Multibeam v7 app for the Raspberry Pi available at Burdeview by Daniel Carrion. They (the Pis) are classed as ARM hard float.

The bad news is you have to use anon platform until Eric can add it as a standard app. Daniel has an app_info in the zip file and instructions on his blog.
20) Message boards : Number crunching : New BOINC version (Message 1441369)
Posted 11 Nov 2013 by Profile MarkJProject donor
New BOINC version with Virtualbox.

What is this?


There are two flavours you can download. One with virtualbox and one without. You can also download virtualbox on its own from the oracle site. You only need virtualbox if you are going to run a project that uses it such as Test4Theory. If you don't want to run one of the virtualbox projects then it's best to get the version without.

The virtualbox version bundled with BOINC is already (slightly) out of date which is one reason you might want to get it directly from oracle.

Next 20

Copyright © 2014 University of California