Posts by MarkJ

1) Message boards : Number crunching : Fundraising time....Green stars for sale! (Message 1833315)
Posted 5 days ago by Profile MarkJProject Donor
And I'm in too.

You gave $50.00 to SETI@home on November 30, 2016 via Visa

Your confirmation number is 250730
2) Message boards : Number crunching : Monitoring inconclusive GBT validations and harvesting data for testing (Message 1823698)
Posted 12 Oct 2016 by Profile MarkJProject Donor
When I will get feedback from already provided build I could try to disable this pragma too.

Sorry for the delay. Work intervened.

Another set running using new app. Same hosts as before. I also snagged some guppies this time.

Looking through the results it would seem v8.19 supplied by beta validate most of the time. The r3525's are almost all inconclusive or invalid.
3) Message boards : Number crunching : Monitoring inconclusive GBT validations and harvesting data for testing (Message 1823524)
Posted 11 Oct 2016 by Profile MarkJProject Donor
When I will get feedback from already provided build I could try to disable this pragma too.

Sorry for the delay. Work intervened.

Another set running using new app. Same hosts as before. I also snagged some guppies this time.
4) Message boards : Number crunching : Monitoring inconclusive GBT validations and harvesting data for testing (Message 1823266)
Posted 10 Oct 2016 by Profile MarkJProject Donor
So try test iGPU build.

8 cooking on beta using 8.19 app as we speak. Will take a couple of hours for them to be done. All are Arecibo at the moment, no sign of guppis.

Tasks
5) Message boards : Number crunching : Monitoring inconclusive GBT validations and harvesting data for testing (Message 1823244)
Posted 10 Oct 2016 by Profile MarkJProject Donor
If it helps I have a bunch of i7-6700 (Skylake) machines with HD Graphics 530. All are running Win7 x64. Currently have GPU disabled in BOINC due to this issue. Happy to help test on here or beta. Currently have driver 4501 on them but have older drivers as well.

After Christian Beer changed the beta validator for BRP4's I found them all pretty much validating. Unfortunately when one selects beta apps at Einstein it gives all different work types which didn't work on HD Graphics 530.
6) Message boards : Number crunching : Thought(s) on changing S@h limit of: 100tasks/mobo ...to a: ##/CPUcore (Message 1812841)
Posted 26 Aug 2016 by Profile MarkJProject Donor
Most of my machines are i7's so they get 100 / 8 threads = 12.5 WU per thread.

If we expand on that by saying we get 12.5 x number of threads then it could work for the smaller (single thread) machines as well as the larger (56 thread) machines. I would suggest we make it something like 15 per thread. Its simplistic and achievable with the current infrastructure.

A more long term approach might be to increase that number based upon the average turnaround if the host is considered reliable. It could also be applied the other way to reduce the number if the host is unreliable.
7) Message boards : Number crunching : Thought(s) on changing S@h limit of: 100tasks/mobo ...to a: ##/CPUcore (Message 1812467)
Posted 25 Aug 2016 by Profile MarkJProject Donor
Do we really need to differentiate between processors and thread counts? I mean what would I want to do differently in this case? From a task scheduling point of view I have the same number of available threads that want feeding regardless of my physical processor count.
8) Message boards : Number crunching : I've Built a Couple OSX CUDA Apps... (Message 1773873)
Posted 25 Mar 2016 by Profile MarkJProject Donor
Yes, that's basically why I've been working on configuration and dispatch infrastructure. Without that, the base code has to be able to run on a G80 class Pre-Fermi GPU (e.g. 32 max registers) and blocking sync only, minimising CPU use (limited to no usefulness of Cuda streams).

Jason, do we really have to support back to the G80, given Nvidia have thrown them in the legacy bin? Can we convince the powers that be (Eric?) that the lowest supported GPU should be a Fermi for future developments. It sounds like a lot of effort to support old hardware.
9) Message boards : Number crunching : Intel HD Graphics 530 - Keeps "Postponing"... (Message 1764591)
Posted 13 Feb 2016 by Profile MarkJProject Donor
According to Downloadcenter.intel.com there is a 4380 beta driver available. No mention of 4360 on the Intel site so it's probably a Microsoft Win10 special version.

The 4380 driver lists the following fixes
KEY ISSUES FIXED:

Just Cause 3 - game crash
Microsoft Enhanced Video Renderer - API crash
Project CARS - game crash
F1 2015 - game crash during built-in benchmark
Angry Birds - corruption after using “Maximize” icon in game window

Quake IV - corruption resuming from standby mode
Fallout 4 - game crash when game is loading
LEGO Minifigures Online - command list reset lock takes long time (4th & 5th Gen)
Wolfenstein The New Order - game crash


As you can see it's all game stuff, no mention of OpenCL so I wouldn't hold your breath on this one fixing anything, nor the 4360 that you've got.

I've raised a bug with Intel about their OpenCL not working properly (too many validation errors) along with another user eta over at Einstein. Intel told me it's the OEM responsibility, so they don't want to know about it. My tasks run fine, just fail to validate, but then I'm running Win7 x64 on my machines.
10) Message boards : Number crunching : New app? (Message 1753502)
Posted 3 Jan 2016 by Profile MarkJProject Donor
The only thing I can think of, I suspended an active task to start my first v8 task, maybe the stot was not emptied properly ... no one will ever know.

You might want a more recent BOINC version (than 7.4) to get the fixes for emptying out the slot directories. The current release version is 7.6.22.
11) Message boards : Number crunching : TIME FOR A NEW FUND DRIVE. (Message 1752644)
Posted 31 Dec 2015 by Profile MarkJProject Donor
You gave to: SETI@home

We received your gift of $100.00

via Visa on 12/31/2015.

Your confirmation number is 174555.
12) Message boards : Number crunching : Windows 10 - Yea or Nay? (Message 1707317)
Posted 1 Aug 2015 by Profile MarkJProject Donor
Installed on a Intel cruncher with iGPU.

It downloads the latest drivers when you install. After rolling back the iGPU driver to the one I had before it couldn't detect my screen so I got stuck at 1024x764 res. You have no control over the drivers, it just downloads the latest and installs it. After rolling back to original it downloaded the latest one again.

Edge was quick. Pity about the driver crap. I've gone back to Win7.
13) Message boards : Number crunching : Using BOINCView. (Message 1704589)
Posted 24 Jul 2015 by Profile MarkJProject Donor
I have a couple of Parallella's (they run Ubuntu) and a few Raspberry Pi's (they run Debian). I use BOINCtasks with them. While BOINCview will work it can't do half of what BT can do.

From what I gather the author of BV passed away so it hasn't been updated and the source code was not made available.

To manage machines via either one you'll need to setup a gui_rpc_auth.cfg and remote_hosts.cfg. They go in the BOINC data folder. The remote_hosts tells it which computers (or IP addesses) are allowed and the gui_rpc_auth has a password, although it can be blank. You need to set the same password in BV or BT when you configure the computers that its going to control.

If you have a firewall configured on your number crunchers you'll need to allow the BT machine through.
14) Message boards : Number crunching : Windows 10 - Yea or Nay? (Message 1703712)
Posted 21 Jul 2015 by Profile MarkJProject Donor
Ah, only optional: https://support.microsoft.com/en-gb/kb/3064209

There seem to be a number of Windows 7 users complaining they can't boot after installing this one, so tread carefully.

Another gripe seems to be its getting offered to AMD-based machines as well.
15) Message boards : Number crunching : Curious about cc_config.xml (Message 1696721)
Posted 29 Jun 2015 by Profile MarkJProject Donor
BOINCtasks also has the ability to edit the cc_config (even of a remote machine) and the end result will be a file with all the tags even though you may only have changed one of them.

Personally I prefer to edit the thing by hand so you don't get all the tags.
16) Message boards : Number crunching : Windows 10 - Yea or Nay? (Message 1695131)
Posted 24 Jun 2015 by Profile MarkJProject Donor
A question for you guys testing it.

From what MS is telling us is the Pro version will be able to run 2 VM's and the Enterprise version has no limit. So does that mean you can only run 2 VirtualBox tasks at a time on Pro?
17) Message boards : Number crunching : HD 2500/4000 graphics driver 10.18.10.4226 from intel (Message 1692618)
Posted 17 Jun 2015 by Profile MarkJProject Donor
Second batch of results are in for Einstein BRP4 tasks. They too show a faster run time but slightly more CPU usage. I did the same approach as for Seti (see previous message).

Before test, 16 samples with average run time of 813.88 and CPU of 29.00.

After test, 18 samples with average run time of 774.41 and CPU of 29.29. There was however one task that took almost 3 times longer so it skewed the averages.

Host is an identical i7-3770. Link to host is here. Tasks up to and including 7Jun2015 13:06:18 UTC were using old driver. Tasks after that are using the newer driver.
18) Message boards : Number crunching : HD 2500/4000 graphics driver 10.18.10.4226 from intel (Message 1692610)
Posted 17 Jun 2015 by Profile MarkJProject Donor
First batch of results are in for Seti (they ran out of work units to send). I took a simplistic approach. Get the run time and CPU time off the task detail on the website and average them. Compare old with new. I realise that angle range will effect how long they take but I don't have the time or patience to check the AR for each one.

Old was a small sample of 11 MB run while the CPU also had 7 CPU tasks going. Average run time was 3,728.82 and CPU time 169.04.

New got a slightly bigger sample of 15 MB again run while the CPU had tasks going. Average run time was 3,475.52 and CPU time 150.23

Make your own conclusions but that looks to be slightly faster and less CPU time to me. As they say your mileage might vary.

That's on an i7-3770 running at stock speed, no tweaking but was running optimised app r2489 single GPU task at a time. The installer requires .NET runtime 4.5 or later.

Link to host here. Tasks done up to and including 16Jun2015 10:46:42 UTC were the old driver and after that its the new driver.
19) Message boards : Number crunching : HD 2500/4000 graphics driver 10.18.10.4226 from intel (Message 1692132)
Posted 16 Jun 2015 by Profile MarkJProject Donor
Has anyone tried the Intel HD 2500/4000 graphics driver 10.18.10.4226 dated the 5th of June 2015?

The third generation (2500/4000) drivers are usually OK, but I'm still not going to upgrade mine. It's the fourth generation ones (4200/4400/4600 and upwards) which they usually break. Use with care, and check validation rates (especially at Einstein), whichever.

I found 3621 to be the best one so far on my i7-3770 machines.

I am running a test on this one now. I suspect it will be slower due to all the WDDM stuff they keep adding but we shall see. I am also running down tasks on another rig so I can run Einstein tasks on it. They quite often will fail unless I use 3621 or 4176 driver.
20) Message boards : Number crunching : HD 2500/4000 graphics driver 10.18.10.4226 from intel (Message 1691595)
Posted 15 Jun 2015 by Profile MarkJProject Donor
Has anyone tried the Intel HD 2500/4000 graphics driver 10.18.10.4226 dated the 5th of June 2015?


Next 20


 
©2016 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.