Posts by HAL9000


log in
1) Message boards : Number crunching : Panic Mode On (95) Server Problems? (Message 1647311)
Posted 3 days ago by Profile HAL9000
Remember folks, that it is never too late to give up :-)

I'm to lazy to give up. Wait...
2) Message boards : Number crunching : Intel® iGPU AP bench test run (Message 1647020)
Posted 4 days ago by Profile HAL9000
I have an Intel® Celeron® J1900 (Quad-Core) with Intel® HD Graphics (iGPU).
The iGPU just have 4 compute units.
An AP WU lasts ~ 21 hours.
(Not freeing CPU thread/s. I saw no difference. Or is there?)

With: -unroll 4 -ffa_block 1024 -ffa_block_fetch 512 -hp (recommendation in readme file)
[-instances_per_device 1 (for MB and AP)]

For to make bench test runs, I use 'Windows AP bench 211 minimal' ...
But with which AP test WU? It should be a very fast/short WU.

http://lunatics.kwsn.net/index.php?module=Downloads;catd=44

The whole bench test run should not last days on this slow iGPU. ;-)

(IIRC BOINC is suspended during bench test run, so the whole time no crunching.)

Thanks.

As Joe stated from looking at my J1900. Running 18-24 hours is "normal" for the iGPU. CPU WU times are nearly the same for me with the CPU running at 2.41 GHz boost constantly.
3) Message boards : Number crunching : Panic Mode On (95) Server Problems? (Message 1646442)
Posted 5 days ago by Profile HAL9000
I'm not seeing a whole lot of errors on my end. Tuesday has a higher error count, but that just has to do with the everything being down for maintenance.
Personally I'm nor worried about the "HTTP gateway timeout" that I get on this machine at work. As those occur sometimes due to the nature of the MS TMG server being used. In that it sometimes decides I have to many connections open and blocks all traffic intermittently. I would have thought setting the limit to 10,000 would be enough...

As of 2015-02-25 14:20 UTC
Showing result for date: 2015-02-25 Scheduler Request Count: 26 Scheduler Success Count: 24, 92 % of requests Scheduler Failure Count: 2, 7 % of requests Failure Details: __________________________________________________________ Description Count Total % Failure % "Timeout was reached": 0 0 % 0 % "Couldn't connect to server": 0 0 % 0 % "Couldn't resolve host name": 0 0 % 0 % "Receiving data from the peer": 0 0 % 0 % "HTTP gateway timeout": 2 7 % 100 % "HTTP internal server error": 0 0 % 0 % "HTTP service unavailable": 0 0 % 0 % Showing result for date: 2015-02-24 Scheduler Request Count: 73 Scheduler Success Count: 54, 73 % of requests Scheduler Failure Count: 19, 26 % of requests Failure Details: __________________________________________________________ Description Count Total % Failure % "Timeout was reached": 0 0 % 0 % "Couldn't connect to server": 0 0 % 0 % "Couldn't resolve host name": 0 0 % 0 % "Receiving data from the peer": 0 0 % 0 % "HTTP gateway timeout": 6 8 % 31 % "HTTP internal server error": 10 13 % 52 % "HTTP service unavailable": 3 4 % 15 % Showing result for date: 2015-02-23 Scheduler Request Count: 70 Scheduler Success Count: 69, 98 % of requests Scheduler Failure Count: 1, 1 % of requests Failure Details: __________________________________________________________ Description Count Total % Failure % "Timeout was reached": 0 0 % 0 % "Couldn't connect to server": 0 0 % 0 % "Couldn't resolve host name": 0 0 % 0 % "Receiving data from the peer": 0 0 % 0 % "HTTP gateway timeout": 1 1 % 100 % "HTTP internal server error": 0 0 % 0 % "HTTP service unavailable": 0 0 % 0 % Showing result for date: 2015-02-22 Scheduler Request Count: 73 Scheduler Success Count: 71, 97 % of requests Scheduler Failure Count: 2, 2 % of requests Failure Details: __________________________________________________________ Description Count Total % Failure % "Timeout was reached": 0 0 % 0 % "Couldn't connect to server": 0 0 % 0 % "Couldn't resolve host name": 0 0 % 0 % "Receiving data from the peer": 0 0 % 0 % "HTTP gateway timeout": 1 1 % 50 % "HTTP internal server error": 0 0 % 0 % "HTTP service unavailable": 1 1 % 50 % Showing result for date: 2015-02-21 Scheduler Request Count: 71 Scheduler Success Count: 70, 98 % of requests Scheduler Failure Count: 1, 1 % of requests Failure Details: __________________________________________________________ Description Count Total % Failure % "Timeout was reached": 0 0 % 0 % "Couldn't connect to server": 0 0 % 0 % "Couldn't resolve host name": 0 0 % 0 % "Receiving data from the peer": 0 0 % 0 % "HTTP gateway timeout": 0 0 % 0 % "HTTP internal server error": 0 0 % 0 % "HTTP service unavailable": 1 1 % 100 %
4) Message boards : Number crunching : Splitting an internet feed (Message 1646273)
Posted 6 days ago by Profile HAL9000

So of course that begs the question ....

Why are these things sold and what are they used for? Blowed if I know!



This allows you to put two phones on the same line. If you have DSL you can filter one side and put a phone there with a landline phone with your modem on the other and all will work just fine. (I even have one that looks like this, basically, that separates the four wires in a phone line.) I used to have two different wired phones and used a single wire for both.

Edit: Sorry Mark, I did not see your post.

Not the one Chris is talking about, his has absolutly nothing to do with phones. His has an RJ45 plug on one side and two RJ45 sockets on the other. Not sure how it is in the US but you can't plug a phone into an RJ45 as in the UK our phone plugs and sockets look like this

https://en.wikipedia.org/wiki/British_telephone_sockets

Chris's splitter uses the fact that cat5 has 8 wires but only uses 4 so a second LAN connection can be run over the spare wires. However that needs two splitters and two sockets on router or switch.

The other type has all 8 wires connected straight through and allows 2 pieces of equipment to share a LAN connection, just NOT AT THE SAME TIME.

If you read the description on several retailers websites it explains the difference and how they are to be used.

Now there is a splitter made for phones in the US that looks similar to that but would not work here as we use the 431a plug not and RJ11

Prior to digital office phones. The use of RJ45 connections for POTS lines was not unheard of in the US. Such as with 4 line phones.
There are RJ45 jacks designed to accommodate RJ12/RJ11 connections. Anymore they are the more common type I see.
5) Message boards : Number crunching : Panic Mode On (95) Server Problems? (Message 1646055)
Posted 6 days ago by Profile HAL9000
Chance for a working AP DB and AP project after today's outage:
Less than 3 %

Oh, That high?
6) Message boards : Number crunching : ASUS AMD R7 260X 2GB (Message 1645552)
Posted 7 days ago by Profile HAL9000
Nope... Not on this card. The box has a 500 watt power supply. The board wont even post when this specific card is plugged in.

I returned the card. I remember that I had a Radeon 4870 in storage and am going to get that. Do you think a radeon 4870 will work for Seti ?

Robert

Sounds like the R7 250 may have been DOA. Perhaps you could try testing it in your other machine?

The HD4870 can be made to work, but you will have to use an older driver as support for the older HD4000 series has been removed in newer drivers. Anything from Cat 11.12 to Cat 12.4 should be good.
I'm pretty sure the current apps will still work, for now, with that card. It should be noted that the HD4000 series only had "beta" OpenCL support.
7) Message boards : Technical News : Latest Astropulse Status (Message 1645131)
Posted 9 days ago by Profile HAL9000
Gotta love those "That can't be happening... but it is happening" moments.

Thanks for the update before you get back to your other 50 things that need to all be done at the same time.
8) Message boards : Number crunching : Not sure what is happening to my completed WU's (Message 1645130)
Posted 9 days ago by Profile HAL9000
Well I thought if I cleared my cache it would straighten itself out. Unfortunately no, they are still disappearing into the ether. Il, track the ones it completes to see where they go...

Thanks,

Chris

Looks like you dumped the tasks at 21 Feb 2015, 22:19:13 UTC. That host shows 0 in progress & no new tasks downloaded since then.
If it happens to be crunching anything for SETI@home it is not associated with that host ID.
9) Message boards : Number crunching : Productivity Chart (Message 1644958)
Posted 9 days ago by Profile HAL9000
Someone on the "panic" thread claims that there is increasingly a problem with invalid results being returned (apparently by GPU's). Is there a chart or table somewhere that lists the number of invalid results being returned over time, so we can check this claim? I know my account has that information, but I'd like to track the metric for the entire project.

You might be able to pick the information you want out of http://setiathome.berkeley.edu/stats/
10) Message boards : Number crunching : Not sure what is happening to my completed WU's (Message 1644956)
Posted 9 days ago by Profile HAL9000
So a couple of weird things happened yesterday on one of my machines. First a whole slew of WU's say they were abandoned because they ran out of time. This was not the case, their deadlines were well into the future. But more concerning is that the machine is getting new WUs and completing them, reporting them and then they seemingly disappear... My event log says they are completed and reported but the machine info on their site does not indicate that it has reported any work units in two days... Any thoughts on what might be going on?

http://setiathome.berkeley.edu/results.php?hostid=7159696

That's the machine in question...

Thanks,

Chris

Note sure but the easiest way to find out what is happening to the completed work would be to track the tasks that your machine is actively working on.
These are the current two oldest tasks on that system.
http://setiathome.berkeley.edu/workunit.php?wuid=1712359086
http://setiathome.berkeley.edu/workunit.php?wuid=1712359013
If BOINC is running FIFO, as it normally does. Then they should be the next to be completed.
11) Message boards : Number crunching : Splitting an internet feed (Message 1644942)
Posted 9 days ago by Profile HAL9000
I'm not sure what this "doubler" device you have purchases is, but it doesn't sound like a standard piece of network equipment to me. Unless it is a small 2 port hub/switch I wouldn't expect it to actually work.
From what other have posted I suspect what you have purchases is meant more for splitting a phone jack out for multiple devices.

I took an old 8 port switch I had laying around & put it in my TV stand to account for the growing number of internet devices I have there. I'm currently only up to 4. However, I suspect my next TV & surround receiver will have the ability to connect to the internet as well for various functions.
12) Message boards : Number crunching : Raspberry Pi 2 (Message 1644266)
Posted 11 days ago by Profile HAL9000
The Raspberry Pi B apparently has "single precision performance of 0.065 GFLOPS" under Raspbian & the Raspberry Pi 2 is said to have "6x more processing power". I'm not sure if the "6x" is figured using all 4 cores or just 1, but 6x would put it at 0.39 GFLOPS of CPU processing power. Which is just a bit slower than a Pentium M @ 1.7GHz, but using a lot less power.
The GPU is listed at 24 GFLOPS & it looks to be the same GPU as the previous generation.
13) Message boards : Number crunching : Panic Mode On (95) Server Problems? (Message 1643504)
Posted 13 days ago by Profile HAL9000
FYI,
I had contact with Eric ...
He said 'if everything will go like it should' then at 24 Feb AP will be online again.
You see it's not 100 % sure. So please don't shoot Eric or me if AP will be still offline. ;-)

Things most often do not go as planned. So I would call anything before March 24th a success. :D
14) Message boards : Number crunching : Seti fundraising via Bitcoin Utopia (Message 1642766)
Posted 14 days ago by Profile HAL9000
The first SETI@home BCU campaign has completed & it looks like we did it in about 90 days.
http://www.bitcoinutopia.com/2014/11/setihome-annual-fund-drive/
15) Message boards : Number crunching : Whats the Best Platform to Crunch on? (Message 1642689)
Posted 15 days ago by Profile HAL9000
Hal, you forgot ... which setup gives the most reliable since results.

From what I see in my validating of other's inconclusive results ... NVidia cards suck at being reliable Scientifically. But you will bang up your credits.

Now that should start a debate!

Measuring reliability is a tricky thing to do. If you simply compare a hosts valid & invalid results the top 20 hosts run ~1-2% invalid. Which isn't to bad on most scales of reliability. Also if they were very unreliable then they would not be producing so many valid results. Granted there are a number of hosts out there producing 100% garbage. Like due to the person running them not ever monitoring the machine.
There is the well known issue when a seemingly valid CPU result will get marked invalid against two GPU results. With the two GPU results are obviously trashed. However, That is more of an issue with limits the project, or BOINC, has on validating results. That can also happen with 3 CPU results, but is probably less likely due to the much more mature nature of CPU computing vs GPGPU computing.

Error rates and issues are a given for GPGPU computing that we can only accept for the time being.
16) Message boards : Number crunching : Whats the Best Platform to Crunch on? (Message 1642524)
Posted 15 days ago by Profile HAL9000
I know the PS3's have been pretty much depreciated since the firm ware update. The Android Based Platform runs slow on mid range to low end devices, so with that in mind is the best plat form pretty much the Desktop CPU and GPU?

Is their any alternatives?

Best in which aspect?
-Best total output
-Best Performance Per Watt
-Best Performance Per Dollar/unit of currency
One option not many consider is the MoDT(Mobile on Desktop) setup. A desktop MB that takes a mobile/notebook CPU & gives the ability to add standard desktop GPUs.
Here are some examples of the current Socket G3 MBs from ASrock.
A similar, & cheaper to MoDT, way to go would be with an Intel Atom based system & adding GPUs.
17) Message boards : Number crunching : Panic Mode On (95) Server Problems? (Message 1641911)
Posted 16 days ago by Profile HAL9000
(Currently) I have just an Intel® Celeron® J1900 + Intel® HD Graphics.
The iGPU have just 4 compute units.

E.g. an AR 0.414109 WU lasts 6h:10mins on the iGPU.
Maybe a .vlar would lasts ~ 18 hours ...?

I aborted the two WUs this morning ...

AFAIK, ATI/AMD GPUs get .vlar's, or not longer?

I also have a Celeron J1900, but have it configured to only run AP at present. Its iGPU is much slower than HD2500 or HD4600 iGPU, but gets the job done just fine for me.

It was reported that Eric is pretty sure the VLAR to GPU exception is back in place. So I would not expect iGPU to get VLAR any longer.
18) Message boards : Number crunching : Panic Mode On (95) Server Problems? (Message 1641866)
Posted 16 days ago by Profile HAL9000
What about 'Intel GPUs'?
Normally they could get .vlar's, or not?

My 'Intel GPU' got .vlar's at:
14 Feb 2015, 0:39:24 UTC and
14 Feb 2015, 6:40:35 UTC

Normally no GPUs would get tasks marked as VLAR.
I have never tried running VLAR tasks on my Haswell, Ivy Bridge, or Bay Trail GPUs. So I'm not sure how they would handle them, but my gut feeling is that they would be ok. You will have to let us know how your system responded to the tasks.
19) Message boards : Number crunching : Limit iGPU to one WU (Message 1641438)
Posted 17 days ago by Profile HAL9000
This is my current app_config

<app_config>
<app>
<name>setiathome_v7</name>
<gpu_versions>
<gpu_usage>0.33</gpu_usage>
<cpu_usage>0.1</cpu_usage>
</gpu_versions>
</app>
<app>
<name>astropulse_v7</name>
<gpu_versions>
<gpu_usage>0.33</gpu_usage>
<cpu_usage>0.2</cpu_usage>
</gpu_versions>
</app>
</app_config>

I'm sorry, I don't understand the syntax of your example.

Something along the lines of this
<app_config> <app_version> <app_name>setiathome_v7</app_name> <plan_class>CUDA50</plan_class> <ngpus>0.33</ngpus> <avg_ncpus>0.1</avg_ncpus> </app_version> <app_version> <app_name>setiathome_v7</app_name> <plan_class>opencl_intel_gpu_sah</plan_class> <ngpus>1.0</ngpus> <avg_ncpus>0.1</avg_ncpus> </app_version> <app_version> <app_name>astropulse_v7</app_name> <plan_class>opencl_nvidia_100</plan_class> <ngpus>0.33</ngpus> <avg_ncpus>0.2</avg_ncpus> </app_version> <app_version> <app_name>astropulse_v7</app_name> <plan_class>opencl_intel_gpu_102</plan_class> <ngpus>1.0</ngpus> <avg_ncpus>0.2</avg_ncpus> </app_version> </app_config>

You would need to match the values in <plan_class></plan_class> with the ones you are using in your app_info.xml
20) Message boards : Number crunching : Limit iGPU to one WU (Message 1641421)
Posted 17 days ago by Profile HAL9000
One of my hosts has two GTX 970's and also an integrated GPU (intel HD4600). I run multiple WU's concurrently on the 970's but would like to limit the 4600 to one at a time. Is there a way to do that?

Also, are there any recommended settings for mb_cmdline_win_x86_SSE_OpenCL_Intel.txt?

Have you tried an app_config.xml with the <app_version> settings?

<app_config>
[<app>
<name>uppercase</name>
<max_concurrent>1</max_concurrent>
[<fraction_done_exact/>]
<gpu_versions>
<gpu_usage>.5</gpu_usage>
<cpu_usage>.4</cpu_usage>
</gpu_versions>
</app>]
...
[<app_version>
<app_name>uppercase</app_name>
[<plan_class>mt</plan_class>]
[<avg_ncpus>x</avg_ncpus>]
[<ngpus>x</ngpus>]
[<cmdline>--nthreads 7</cmdline>]
</app_version>]

...
[<project_max_concurrent>N</project_max_concurrent>]
</app_config>


Next 20

Copyright © 2015 University of California