Hi SETI people,

I have noticed that a recent batch of CPU WUs eventually calculate but have a much bigger than usual Run time v CPU time. Here is an example.

Name 07ja12ag.624.13319.9.10.99_2
Workunit 943242023
Created 11 Apr 2012 | 4:42:20 UTC
Sent 12 Apr 2012 | 4:55:15 UTC
Received 15 Apr 2012 | 18:51:13 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x0)
Computer ID 5880149
Report deadline 18 May 2012 | 7:13:10 UTC
Run time 52,327.46
CPU time 18,250.49
Validate state Valid
Credit 95.18
Application version SETI@home Enhanced v6.03
Stderr output

<core_client_version>7.0.25</core_client_version>
<![CDATA[
<stderr_txt>
setiathome_enhanced 6.02 DevC++/MinGW
libboinc: 6.3.6

Work Unit Info:
...............
WU true angle range is : 0.826361
Optimal function choices:
-----------------------------------------------------
name
-----------------------------------------------------
v_BaseLineSmooth (no other)
v_vGetPowerSpectrumUnrolled 0.00011 0.00000
sse3_ChirpData_ak 0.01533 0.00000
v_vTranspose4ntw 0.01086 0.00000
AK SSE folding 0.00118 0.00000

Flopcounter: 18935091682409.402000

You will see that in this case the run time is about three times the CPU time. This is much longer than usual. I run these on an E4700 Intel core duo and my system is not running anything more than normal at the time.
Does anybody have any ideas why this should occur?

Thanks for any contributions

John

Could it be that the temperature of your CPU is too hot, an therefore your CPU has changed to a lower frequency?

---

Goodbye Seti Classic

Could it be that the temperature of your CPU is too hot, and therefore your CPU has changed to a lower frequency?

I don't believe a change in CPU frequency would affect this - CPU time would stretch out also in such case.

---

Hi SW@NL_XP_Freak and jravin,

thanks for your suggestions. I couldn't find out what frequency the processors were working at directly so I ran the Performance Information and Scores from the Vista Control panel Administrative tools and the CPU score has not changed since I got the pc.

Any other ideas?

John

Hi SW@NL_XP_Freak and jravin,

thanks for your suggestions. I couldn't find out what frequency the processors were working at directly so I ran the Performance Information and Scores from the Vista Control panel Administrative tools and the CPU score has not changed since I got the pc.

Any other ideas?

John

CPUZ will read the processor and memory speeds directly.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

Hi msattler,

I am not familiar with CPUZ. Can you please tell me where to find it and how to use it.

Say hi to the kitties

John

In the US we've developed a "search" engine. It's called Google. See if it works in the UK

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

You can find and download the latest version of CPU-Z at http://www.cpuid.com

Select the Full install version or the zip stand alone run time in 32 or 64 bit version.

What is CPU-Z

CPU-Z is a freeware that gathers information on some of the main devices of your system.
CPU

Name and number.
Core stepping and process.
Package.
Core voltage.
Internal and external clocks, clock multiplier.
Supported instruction sets.
Cache information.

Mainboard

Vendor, model and revision.
BIOS model and date.
Chipset (northbridge and southbridge) and sensor.
Graphic interface.

Memory

Frequency and timings.
Module(s) specification using SPD (Serial Presence Detect) : vendor, serial number, timings table.

System

Windows and DirectX version.

---

In the US we've developed a "search" engine. It's called Google. See if it works in the UK

Do you mean that s/ware developed by Сергей Михайлович Брин

Hi msattler,

I am not familiar with CPUZ. Can you please tell me where to find it and how to use it.

Say hi to the kitties

John

Hiya John...
I was out for the evening, but I see that Dr. Who Fan posted a link for the download.
It's a very simple, small program that does not even need installation.
I usually save it to my temp folder, unzip it, and then put a shortcut to it on my desktop.
You just run it, it collects the data, and then there are a number of tabs on the top for various categories of information.
Very simple, nothing to learn to use it.

The kitties say meow back at ya.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

Thanks to Dr.Who Fan and msattler again. I will download CPUZ and see if that is detects any problems with my CPU. I thought that it was something on BOINC or SETI not a prog from the net.

Skildude you are correct I could/should have Googled it and would have if I had realised that it was an external source but it was 04:00 BST and I was not thinking straight.

I will get it and see how healthy my CPU is.

My original problem was with a batch of wu's that had Run time = 3xCPU time if my CPU turns out to be healthy then I will need to think again.

In the meantime thanks to all who offered help.

John

Hi again,

I now have the tool and it says that my wee core duo is running at 2.6GHz which is what it should do. So S@NL_XP_Freaks suggestion that my CPU was overheating doesn't seem to be true.

I am continuing to crunch and I have had one normal one since my report so maybe I just had a batch of funny ones.

I will keep an eye on things and report back if anything worse happens.

Thanks again peeps
#John

In the US we've developed a "search" engine. It's called Google. See if it works in the UK

Do you mean that s/ware developed by Сергей Михайлович Брин

ROFL :D

Hi SETI people,

I have noticed that a recent batch of CPU WUs eventually calculate but have a much bigger than usual Run time v CPU time. Here is an example.

Name 07ja12ag.624.13319.9.10.99_2
Workunit 943242023
Created 11 Apr 2012 | 4:42:20 UTC
Sent 12 Apr 2012 | 4:55:15 UTC
Received 15 Apr 2012 | 18:51:13 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x0)
Computer ID 5880149
Report deadline 18 May 2012 | 7:13:10 UTC
Run time 52,327.46
CPU time 18,250.49
Validate state Valid
Credit 95.18
Application version SETI@home Enhanced v6.03
Stderr output

<core_client_version>7.0.25</core_client_version>
<![CDATA[
<stderr_txt>
setiathome_enhanced 6.02 DevC++/MinGW
libboinc: 6.3.6

Work Unit Info:
...............
WU true angle range is : 0.826361
Optimal function choices:
-----------------------------------------------------
name
-----------------------------------------------------
v_BaseLineSmooth (no other)
v_vGetPowerSpectrumUnrolled 0.00011 0.00000
sse3_ChirpData_ak 0.01533 0.00000
v_vTranspose4ntw 0.01086 0.00000
AK SSE folding 0.00118 0.00000

Flopcounter: 18935091682409.402000[/color]

You will see that in this case the run time is about three times the CPU time. This is much longer than usual. I run these on an E4700 Intel core duo and my system is not running anything more than normal at the time.
Does anybody have any ideas why this should occur?

Thanks for any contributions

John

Hi John.
I'm not sure what you're getting at here, according to that stderr_txt file the unit was crunched with V6.03 the CPU app. I also note you are running the latest V7 BOINC client. You have a pretty heavy load on what I presume is an old laptop

Looking at your computer, your GPU only has 256MB of memory and you are running V285.62 drivers. These drivers require more than 256 MB of video RAM to run.

What is happening is that BOINC detects the low value of of VRAM and the unit "falls back" to being run on the CPU.

There are combinations of Drivers and GPU apps that will run with 256MB GPU's but off hand I don't know them. Maybe if you do a search of the forums or someone else can post them.

A good place to start though, is to roll back to V191.07 drivers and the stock V6.09 CUDA App. To help save memory, disable your screensaver too

T.A.

Hi SETI people,

I have noticed that a recent batch of CPU WUs eventually calculate but have a much bigger than usual Run time v CPU time. Here is an example.

Name 07ja12ag.624.13319.9.10.99_2
Workunit 943242023
Created 11 Apr 2012 | 4:42:20 UTC
Sent 12 Apr 2012 | 4:55:15 UTC
Received 15 Apr 2012 | 18:51:13 UTC
Server state Over
Outcome Success
Client state Done
Exit status 0 (0x0)
Computer ID 5880149
Report deadline 18 May 2012 | 7:13:10 UTC
Run time 52,327.46
CPU time 18,250.49
Validate state Valid
Credit 95.18
Application version SETI@home Enhanced v6.03
Stderr output

<core_client_version>7.0.25</core_client_version>
<![CDATA[
<stderr_txt>
setiathome_enhanced 6.02 DevC++/MinGW
libboinc: 6.3.6

Work Unit Info:
...............
WU true angle range is : 0.826361
Optimal function choices:
-----------------------------------------------------
name
-----------------------------------------------------
v_BaseLineSmooth (no other)
v_vGetPowerSpectrumUnrolled 0.00011 0.00000
sse3_ChirpData_ak 0.01533 0.00000
v_vTranspose4ntw 0.01086 0.00000
AK SSE folding 0.00118 0.00000

Flopcounter: 18935091682409.402000[/color]

You will see that in this case the run time is about three times the CPU time. This is much longer than usual. I run these on an E4700 Intel core duo and my system is not running anything more than normal at the time.
Does anybody have any ideas why this should occur?

Thanks for any contributions

John

Hi John.
I'm not sure what you're getting at here, according to that stderr_txt file the unit was crunched with V6.03 the CPU app. I also note you are running the latest V7 BOINC client. You have a pretty heavy load on what I presume is an old laptop

Looking at your computer, your GPU only has 256MB of memory and you are running V285.62 drivers. These drivers require more than 256 MB of video RAM to run.

What is happening is that BOINC detects the low value of of VRAM and the unit "falls back" to being run on the CPU.

There are combinations of Drivers and GPU apps that will run with 256MB GPU's but off hand I don't know them. Maybe if you do a search of the forums or someone else can post them.

A good place to start though, is to roll back to V191.07 drivers and the stock V6.09 CUDA App. To help save memory, disable your screensaver too

T.A.

I have been using the 280.26 release with my 256MB 8500GT for some time without any issues.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

Hi John.
I'm not sure what you're getting at here, according to that stderr_txt file the unit was crunched with V6.03 the CPU app. I also note you are running the latest V7 BOINC client. You have a pretty heavy load on what I presume is an old laptop

Looking at your computer, your GPU only has 256MB of memory and you are running V285.62 drivers. These drivers require more than 256 MB of video RAM to run.

What is happening is that BOINC detects the low value of of VRAM and the unit "falls back" to being run on the CPU.

There are combinations of Drivers and GPU apps that will run with 256MB GPU's but off hand I don't know them. Maybe if you do a search of the forums or someone else can post them.

A good place to start though, is to roll back to V191.07 drivers and the stock V6.09 CUDA App. To help save memory, disable your screensaver too

T.A.

Those are 6.03 tasks. They are CPU tasks NOT GPU tasks. They already run on the CPU they can't fall back to the CPU!

A 256 MiB GPU can be run with optimised apps and new drivers if you can free up enough VRAM.
I haven't looked at the 6.08 memory footprint with new drivers to be able to say if it can be run - I need a ToDo list :( - but most likely with underreported VRAM SETI won't issue GPU tasks in the first place. You probably have to bin 2 years of driver development and go back to 195.62 if you want to be issued stock 6.08 on a 256 MiB GPU.

@John do the longer run times coincide with BOINC version change?

---

I'm not the Pope. I don't speak Ex Cathedra!

Hi SETI people,

1. Winterknight/Raistmer: I am not sure about Googles' Russian antecedence but maybe Skildude can help.

2. Terror Australis/HAL 9000/Lady L: Perhaps my original post was not clear enough. BOINC says my GPU is not fit to crunch having too little memory so I have never tried to use it being happy to stick with the wee dual core CPU. That said my usual results show that Run time is only slightly higher than CPU time. With this wu and a few others the Run time is much higher than the CPU time and is not normal for my results.
(I have an Acer Aspire M1640 desktop Intel E4700 CPU running MS Vista HP SP2)

Since my post I have had two more typical results with the run time just above the cpu time so maybe it was just a funny batch. This change does not seem to be connected with the move to BOINC 7.025

Thanks to all for the advice.
John

Sorry thought a bit of morning humor was due. No need to be grumpy ... back to our regularly scheduled flogging

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

Hi Skilldude,

no problems as to the humour even though your early morning is my middle of the night. I was genuinely scrabbling through BOINC and S@H to find CPUZ as I had never heard of it let alone that it was a freebie download available through Google from the net. Your contribution together with Dr Who Fans' eloquent post and msattlers'post gave me that information. You will see from the above that CPUZ revealed that my processor was working ok so that was/is not the source of my problems which seem to have gone away at the moment.

As far as the allegedly Russian source of Googles' omnipotence I learned that at least it was a matter of debate.

I enjoyed what we call in Scotland "the banter" or what the Irish call " the craic" and I am not grumpy, well not at most times.

Thanks
John

Hi John,

When comparing the difference between elapsed (Run) time & reported CPU time, some basic understanding is needed first to track down what's going on.

There is some small part of elapsed consumed by OS Kernel operations, typically on the order of a few seconds (That alone is not enough to explain the difference of course). The remainder would likely be some other process on the machine pre-empting the processes. The CPU science applications typically run at 'below normal' priority, to minimise machine usability impact. This means if you have scheduled periodic antivirus scans, disk de fragmentation, or other intensive processes, then the science applications can spend much of their time waiting... so while using little CPU themselves, elapsed is naturally extended.

What I would suggest, in your case, is to check windows task manager periodically, to see if some unexpected processes are using excessive CPU. If you want to delve deeper, there are many diagnostic checks & more advanced tools to track down what may be consuming CPU, if you don't already know what that might be,

Jason

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.