Posts by HAL9000


log in
1) Message boards : Number crunching : Intel Integrated GPU not working (Message 1658618)
Posted 2 days ago by Profile HAL9000
I just followed the rules and the intel drivers for IGPU are installed. The question is should some additional drivers in addition to the standard device drivers be installed?

You may find that a monitor needs to be connected to the iGPU output, and/or the Windows desktop extended to include the iGPU. Claggy is the expert on this.

Hal6000 did a How too:

http://www.hal6000.com/seti/manual_display_detect.htm

Claggy

I found that I only had to extended the desktop when starting BOINC. Once BOINC was running I could switch back to a normal desktop until I restarted the system or BOINC.
2) Message boards : Number crunching : Nvidia Titan X (Message 1658616)
Posted 2 days ago by Profile HAL9000
You meant slower in MHz when you said:
"Interesting, but according to that article it would be slower than the Titan X which is slower than the current GTX 980s"

GTX 980 > Titan X > GTX 980Ti

That's where I started to get confused, because I was talking about GFLOPS

I wasn't comparing clock speeds, you were.
What I was refering to was the amount of cuda cores in relation with clock speed to calculate sp performance. That is the same for different gpu cores like 980 or titan x.

That order is how I would expect them to be then the 980Ti comes out. It would also fit the pattern from the 700 series. With the x80Ti being faster than the lowest end Titan. Although the DP performance of the 900 series titan is rather... lacking compared to the previous generation. Not that we need DP for SETI@home.
3) Message boards : Number crunching : Nvidia Titan X (Message 1658230)
Posted 2 days ago by Profile HAL9000
Interesting, but according to that article it would be slower than the Titan X which is slower than the current GTX 980s

GTX 980 > Titan X > GTX 980Ti

why would you do that?

You would think it would at least make it as fast if not faster than the current model.

But how does it compare to the 780Ti which it is replacing?
4) Message boards : Number crunching : I need some help to find a software (Message 1658226)
Posted 2 days ago by Profile HAL9000
erm, well just some points of order. Firstly if you have XP service pack 2 or newer, then you have Windows firewall, and AFAIK won't uninstall, but you would have to manually disable (which I would never recommend except under tightly controlled laboratory conditions, in isolation, behind a NAT router). Assuming you uninstalled some third party firewall mentioned in the thread (which I didn't read, freely admitted) I would recommend not dropping any more firewalls, but figure out how to open the requisite port(s) instead.

[Edit:] sorry if that sounds tetchy, it's not meant to. Just seen enough chaos induced by imprecise firewall setups to last me a lifetime ;)

The Windows Firewall service can be removed. There is no menu for it, but there is a command line function. I wouldn't suggest removing it as things that apply firewall settings will still be applied, probably via the registry. That may cause some network things to not function & leave you without a way to change/disable the settings.


Configuring the correct ports is the proffered way to go. However if the machine is behind a hardware firewall. Then disabling the software one(s) for a quick test should be OK. However if the hardware firewall is setup like a sieve then it could end in tears.
5) Message boards : Number crunching : Need some help or advice (Message 1658221)
Posted 2 days ago by Profile HAL9000
Edit- I hit F8 went into the DOS screen where it gave a me a choice. I picked start normal windows and it booted up.

What is shown in Device Manager for the USB hardware, mouse?


I seem to allway get booted back to the log on page.

The only way that would occur is if you log off, switch user, or the system reboots.
The system rebooting randomly could easily be due to power supply issues, and they could also result in other random hardware issues.

I found a randomly rebooting system can also be from a intermittently faulty case power switch. Which I only figured out when it failed completely in the closed/pressed position.
6) Message boards : Number crunching : Multi GPU physical support (Message 1658022)
Posted 3 days ago by Profile HAL9000
Well, if one is picky about the appearance, model builder's spray paint comes in a variety of colors as well.

I hope it is sturdy enough. Hard to tell from the pictures how well made it is, although the price would hopefully indicate that it is not TOO flimsy.

While it may look like a small adjustable curtain rod I would hope it is must stronger for the $40.
7) Message boards : Number crunching : Panic Mode On (96) Server Problems? (Message 1657945)
Posted 3 days ago by Profile HAL9000
Well the AP splitters were turned off, I guess they were just teasing us.

The AP splitters are "Not Running" meaning they have nothing to do. When they are listed as "Disabled" they have been shut off.
Also the status coloumn is known to lie at times showing things as "Running" when they have been disabled or are not working.
8) Message boards : Number crunching : Intel Integrated GPU not working (Message 1657860)
Posted 3 days ago by Profile HAL9000
So the latest status is that Intel GPU is detected by Boinc but no crunching on it.

Actually BOINC has not detected your iGPU. I think you might have mistaken this line for the iGPU.
27/03/2015 20:38:56 | | OpenCL CPU: Intel(R) Core(TM) i7-3770K CPU @ 3.50GHz (OpenCL driver vendor: Intel(R) Corporation, driver version 3.0.1.10878, device version OpenCL 1.2 (Build 76413))

Which is just the CPU. Also a good way to tell is by the messages stating there is no iGPU.
27/03/2015 20:38:56 | | app version refers to missing GPU type intel_gpu

If BOINC had detected the iGPU you would see something along the lines of this:
3/27/2015 3:13:02 PM OpenCL: Intel GPU 0: Intel(R) HD Graphics 2500 (driver version 10.18.10.4061, device version OpenCL 1.2, 1298MB, 1298MB available, 53 GFLOPS peak)
3/27/2015 3:13:02 PM OpenCL CPU: Intel(R) Core(TM) i5-3470 CPU @ 3.20GHz (OpenCL driver vendor: Intel(R) Corporation, driver version 3.0.1.10878, device version OpenCL 1.2 (Build 76413))
9) Message boards : Number crunching : I need some help to find a software (Message 1657697)
Posted 3 days ago by Profile HAL9000
Well thank everybody, I will check what you give to me ^^ Thank you!

Also, about my pretty far away computers, they are all into my home network, but I use far away because if I want to use a KVM switch, I would need a very very long cable ^^

I use VNC software at the OS level for my other machines to run them headless. You can also get hardware devices that allow a VNC sessions & give you full remote management. Allowing you to power on, reboot, or shutdown the system from your remote session.

For controlling BOINC I mostly use the built in function of BOINC manager & boinccmd, from command line, for my needs.
10) Message boards : Number crunching : Multi GPU physical support (Message 1657400)
Posted 4 days ago by Profile HAL9000
For that you could get a threaded rod from the hardware and then use a pair of plastic washers between some adjustment nuts for each card.

If there are hard drive bays, or the mounting points for removed drive cages, by the ends of the cards. That might be the easiest way to rig up some other form of brackets.
11) Message boards : Number crunching : Multi GPU physical support (Message 1657357)
Posted 4 days ago by Profile HAL9000
From what you said I think you are looking for a PCIe card retention bracket. So that the end of the card will fit into the card supports in the case?
One issue I see you may run into is that a "Full-Length PCIe card" is defined as having a maximum length of 312.00mm & your GPU may already be exceeding that. I do see a design in my head that would work for an adjustable bracket, but I can't say I have seen anything like that. Something like

But with a 2nd piece that slides for adjustment.

EDIT: Also I remember something called a cardkeeper, but it's main function was to keep the cards in the slot when the machine moved around.

I kind of laughed when I saw this, but I imagine it is similar to the device you found. Power Jack Which you could use if you put a support between each of your 4 cards before using it.
12) Message boards : Number crunching : SSD and Bionic??? (Message 1657237)
Posted 4 days ago by Profile HAL9000
I just recently bought and installed a 1 TB Samsung 850 PRO, and I went from 20-25 minute boot up times to a complete restart in less than 3 minutes. I wish they had included the 850 in their tests, but this drive has a 10 year warranty. Now that I have made the jump to SSD, I will never go back!

This thing is fantastic!

Steve

You had a 20-25 minute boot up time? What the heck is your machine doing at start up. Loading the matrix? And I mean the actual matrix not the movie or game. :P

Is the 850 the replacement product for the 840 or just a higher end series? In either case I would hope for the same or better out of it vs the 840.
13) Message boards : Number crunching : Bandwidth consumption of Seti@home (Message 1656823)
Posted 5 days ago by Profile HAL9000
hello!

I just want to know how many Bandwidth one Seti@home task take.

I know this is not a big amount of Bytes, but I want to know do do some statistic :D

Thank you in advance, your are awesome!

For SETI@home Multibeam one task task is ~377KB to download & then upload is much smaller with a max size, IIRC, of ~30KB. In general it is easy to use 400KB for 1 task total when calculating. Perhaps even 450KB to figure in overhead.
For Astropulse a task is about 8MB to download.
14) Message boards : Number crunching : AstroPulse not being assigned to GPU (Message 1656763)
Posted 5 days ago by Profile HAL9000
You can run AP on GPU only if you install the Lunatics optimized apps.
When running the installer, you do not select to install the AP app for CPU.
It's been so long since I have run stock, I am not sure if you can tweak things there, but you probably could manually remove the stock AP CPU app and accomplish the same thing.

You could manually remove the stock CPU app, but BOINC would just download it again.
For the most part. The only way to specify which app on which hardware is still via an app_info.xml. Which is easiest to do by using the lunatics installer.
An app_config.xml might be able to be used to prevent an app from running on one type of hardware using the <app_version> settings, but that would probably be overly complicated. Also I think BOINC would still download tasks & just not run them.
15) Message boards : Number crunching : SSD and Bionic??? (Message 1656757)
Posted 5 days ago by Profile HAL9000
Tech Report's endurance experiment finally ended. It had a surprising winner.

And the winner is... us. The consumer.

I was surprised at the longevity of them, given the past warnings I have read occasionally about SSDs.
The lifespan shown would outlast most computers in daily use for many many years.

In my main gaming machine I'm wearing the 240GB SSD about 1% a year according to the Sandisk app. However I'll probably be upgrading to a 480-512GB in the next year or so. Then rotate the 240GB into my HTPC which is running an older 120GB SSD now. I have been running BOINC on the SSDs since I got them.
16) Message boards : Number crunching : SSD and Bionic??? (Message 1656678)
Posted 5 days ago by Profile HAL9000
Tech Report's endurance experiment finally ended. It had a surprising winner.

And the winner is... us. The consumer.
17) Message boards : Number crunching : GPU driver version lists (Message 1656674)
Posted 5 days ago by Profile HAL9000
Looks like I would need to have a 2nd look.

Yes, it is known, e.g. by "Computers belonging to Richard Haselgrove":
http://setiathome.berkeley.edu/hosts_user.php?userid=5509

... that GeForce GTX 750 Ti is supported on Windows XP
The NVIDIA page lists ~15 drivers from 335.04 to 347.88

Richard uses 335.28 (obviously successfully) but if he tried other versions I don't know (or remember)

Maybe easier is to look in 'SUPPORTED PRODUCTS' tab of a driver page:
http://www.nvidia.com/download/driverResults.aspx/74421/en-us

In some fields of your table you may need 'fuzzy' words like Most/Some/Few or some ~% of supported GPUs

It does looks like it is a bit more complicated than I first thought.
The main objective is to source out which driver(s) are a ones we don't think should be used for the project. Which may mean there should be an indication of CUDA: OK & Open CL: NG.

Is there even an issue with Fermi & newer generation NV GPUs & driver combo nations at this point? Should this chart be focused only on Pre-Fermi GPUs?
18) Message boards : Number crunching : GPU driver version lists (Message 1656428)
Posted 6 days ago by Profile HAL9000
Besides the ATI driver list I have been keeping. I recently put together one for Nvidia drivers. The NV list needs lots of information filled in at the moment, but if it is useful then I'll populate it as I gain information. Something like the current NV list might be needed for Intel GPU drivers at this point as well.
ATI Driver Version Cheat Sheet
Nvidia Driver Version Cheat Sheet

On your XP version list, you have all drivers listed as "card not supported" for 600, 700, 900 series cards. My XP box has been running a GT 630 on 335.28 for, well about a year now, according to the release date you list for that driver. It has *not* been returning errors.

I was using http://www.nvidia.com/Download/Find.aspx & it looks like for the 600 & 700 series leaving Product: at the default value for XP was my problem. I had changed it to a few different ones when I first started & everything was showing the same. So I figured they would all be like that. Looks like I would need to have a 2nd look.
19) Message boards : Number crunching : 963 Multibeam tasks with "Validation Pending" (Message 1656190)
Posted 7 days ago by Profile HAL9000
But, on a 10 day cache ... ok a slow computer that takes 4 days computing time to complete a task, that's 14 days to get it back, fine whew. So to get to 6 weeks that is 96 hours out of 1008 hours, not bad, roughly 10% daily computing time.

THAT IS ONE TASK AND THEY HAVE 99 MORE THAT THEY GOT THE SAME DAY. Ok, being conservative, they only got 30, but still ... 29 or 99 lost tasks.


Slower hosts are not going to have 99 tasks in a 10 day cache. Remember that the cache is time based, so if the host takes over 240 hours to crunch a single workunit, it will only get a single workunit in its 10 day cache. 120 hours and it will only have 2 tasks in a ten day cache. etc. etc.

These slower systems not clogging up the database as much as you think. It is the faster systems that download several dozen, then the owner loses interest and never bothers to complete them that are a bigger problem (relatively speaking, as I don't think there are many of those either).

I was going to say my slowest host runs about 5-6 days for a normal task & caches 4-5 at a time. So a task it downloads today may not make it back for 20 days. However I do have all of my hosts check in at all of their projects once a day at 2PM (local time). That is mostly so that I get a nice stats graph on each machine.

Shorter limits don't really make any sense for this project. Given none of the data is time sensitive. Any form of host or user reliability metric used on the server side to limit host work would take time to develop & wouldn't make up for reliable hosts that just vanish.
The only real way to effectively reduce the number of pending tasks the OP is seeing would be to further reduce the limits for hosts. Perhaps allowing no cache at all and only the number of tasks for the number of devices in their system. I think many of us remember the days prior to tools, such as SETI Spy, for caching how fun only having 1 task per CPU was.
20) Message boards : Number crunching : GPU driver version lists (Message 1656146)
Posted 7 days ago by Profile HAL9000
There were a total of three 295.xx and 296.xx drivers that had the Nvidia missing GPU when monitor goes to sleep Bug, from my driver stash:

295.51-desktop-win7-winvista-64bit-international-beta.exe

295.73-desktop-win7-winvista-64bit-international-whql.exe

296.10-desktop-win7-winvista-64bit-international-whql.exe

Edit: The Bug isn't apparent when connected via a VGA connection,

Claggy

I added all the drivers that are currently available to download from the Nvidia website. I could cut the list down to start with the 3xx.xx drivers if that makes sense to that. Perhaps it should start the oldest working driver for the OpenCL AP app? I'm unsure what that might be off the top of my head. It may be in one of the app readmes from the lunatics installer. However from suggestions in the @Pre-FERMI nVidia GPU users: Important warning 266.xx drivers were being suggested.
I only have a 8500 GT to muck around with & I have been running 306.23 on it probably since 2012. Which may not be the most ideal driver, but it isn't any more than any previous driver I used. Then again I haven't tried to run AP on that card either.


Next 20

Copyright © 2015 University of California