DC machine - odd behaviour and temps?

[Turbonium]

Long story short: I have a T9500 running at max load (I would assume), yet HWMonitor says it's only at around 50°C.

Task Manager never shows either core going below 70%, and I've set each of the DC projects to use 50% of the CPU resources each (50% x 2 = 100%). The two projects are Folding@home and SETI@home.

Furthermore, the air coming out of the laptop isn't as hot as I would expect it to be.

The system is a Vostro 1400 (integrated Intel X3100 graphics motherboard) - it's using the stock cooler, but I used Noctua NT-H1 thermal compound when installing the CPU.

Is this normal? I feel like something is wrong.

 

[GLeeM]

Does Intel Burn Test give higher temps?

 

[Turbonium]

What is Intel Burn Test? I Googled, but I'm getting lots of results (some seem a bit dodgy).

Please link me to the one you were referring to. I'll try it.

Or can I just try and stress-type test?

 

[sangyup81]

Prime95 can test your temps but Intel Burn Test runs even hotter.

http://www.majorgeeks.com/files/details/intelburntest.html

 

[Turbonium]

Here's what I don't get...

Assume I do a max load type of test on the system, and the temps stay at 50°C. What would that mean?

What if the test yields much higher temps? What would that mean?

I think I already know; just need some clarification.

 

[GLeeM]

If IBT temps are like 80-90 then FAH is not running like it should.

Does anyone know if the T9500 is a very cool running CPU? I'm more used to the hot i7 920 temps

 

[Turbonium]

If IBT temps are like 80-90 then FAH is not running like it should.

Does anyone know if the T9500 is a very cool running CPU? I'm more used to the hot i7 920 temps

Running IBT now.

Task Manager doesn't go below 100%, and my temps max out at 55-60°C.

This leads me to believe that the DC programs aren't running properly. But still, 55-60°C seems incredibly cool for a T9500 under full load with stock cooling.

...or maybe I'm just really really good at applying thermal paste?

 

[Turbonium]

Ok, so I closed SETI, and I'm currently running only Folding (at "Full"), yet Task Manager doesn't really go above 50-55% for either core, and my temps are actually below 50°C.

This is weird.

EDIT: both cores are now around 75-80% (each). Temps are still just as low as before. My PPD estimate is 550. Does that sound right for a T9500?

 

[Assimilator1]

If you right click on F@H>advanced control, go to configure>advanced, is CPU usage at 100%?

 

[GLeeM]

Ok, so I closed SETI, and I'm currently running only Folding (at "Full"), yet Task Manager doesn't really go above 50-55% for either core,

You must be running the FAH uni-processor WUs. You would have to set two of those going after stopping Seti to get 100% usage (at least it used to be that way).

As a general rule DC should peg your CPU at 100%. Although different projects will affect temps differently. FAH used to give highest temps.

 

[Turbonium]

If you right click on F@H>advanced control, go to configure>advanced, is CPU usage at 100%?

Yup.

Also, here is some of the stuff from the main window:

Folding Slots
ID = 00
Status = Running
Description = cpu:2

Selected Work Unit
Folding Slot ID = 00
Work Queue ID = 01
FahCore = 0xa4

Task Manager is constantly jumping from 50% to 100%, for each core. Temps are stable at around 50°C per core.

Thoughts? I want to get the basics down before I post on the actual FAH forums.

Anyway, I was thinking perhaps it's not properly running in SMP mode, despite it being the V7 client. Shouldn't there be two "Folding Slots"?

Again, it's a Core 2 Duo, so two physical cores.

 

[Turbonium]

Perhaps my BIOS settings are somehow set to throttle at ridiculously low temps?

 

[GLeeM]

Shouldn't there be two "Folding Slots"?

If you are doing uni-processor (classic) WUs, yes there should be two slots.

If you have SMP set up in one slot then the program will use up to 16 CPUs (I don't know maybe up to 64 CPUs?) and all should use 100%!

It all depends on how you set it up.

Tell me the Project number of the WU you are crunching and I will tell you if it is SMP or classic uni-processor.

 

[Turbonium]

If you are doing uni-processor (classic) WUs, yes there should be two slots.

If you have SMP set up in one slot then the program will use up to 16 CPUs (I don't know maybe up to 64 CPUs?) and all should use 100%!

It all depends on how you set it up.

Tell me the Project number of the WU you are crunching and I will tell you if it is SMP or classic uni-processor.

Note that I only opened the visualization to get the info. I normally don't have it open (I just didn't know where else to get the info you requested).

 

[GLeeM]

If the number is 2974, then it is a classic uni-processor using only one core.

Find the Project number here: http://fah-web.stanford.edu/psummary.html and note the server number.

Then find the server number here: http://fah-web.stanford.edu/pybeta/serverstat.html
for lots more info.
http://fah-web.stanford.edu/pybeta/serverstat.html

 

[Turbonium]

The weird thing is, both my cores in Task Manager virtually mirror each other in reported load, so it seems to be acting like SMP (at least to my understanding).

Anyway, I thought V7 automatically configures things properly?

 

[Pokey]

I am running Client Control V 7.3.6

Another thing to check is:
open client control in expert mode
click on configure
click on slots tab
highlight cpu
click on edit
at the top look at CPUs and verify that the value is set to -1 (default)
this lets Client Control set the clients. I only use a specific number if I want to cut back on the number of cores.

You also might try reinstalling Client Control.

 

[Turbonium]

It's definitely set to -1.

Client version is 7.3.6 by the way.

This is weird. I'm going to try posting on the FAH forums soon enough if I can't figure this out. Maybe it's a bug?

Also, does my OS version matter? It's Vista Business 32-bit, if it matters. The client correctly identifies it as well.

 

[Philippart]

The weird thing is, both my cores in Task Manager virtually mirror each other in reported load, so it seems to be acting like SMP (at least to my understanding).

Anyway, I thought V7 automatically configures things properly?

You are running windows 7 I suppose. I noticed the same strange pattern on one of my computers: I was not running a distributed computing project, I only ran some Matlab code I wrote (believe me I don't write multi-core software )but all 4 cores displayed the same load pattern??? while no other program was running. I guess it must be a bug in the taskmanager since running the same code on a windows 8 machine only showed one core as active.

 

[Turbonium]

You are running windows 7 I suppose. I noticed the same strange pattern on one of my computers: I was not running a distributed computing project, I only ran some Matlab code I wrote (believe me I don't write multi-core software )but all 4 cores displayed the same load pattern??? while no other program was running. I guess it must be a bug in the taskmanager since running the same code on a windows 8 machine only showed one core as active.

Well, the 2 cores in Task Manager aren't mirrored exactly I don't think. Rather, they're just both fairly busy at similar levels on first glance. You might still be right though. Anyone else know of a bug such as this?

Note, however, that I'm running Windows Vista 32-bit, not Windows 7.

 

[GLeeM]

If all cores are not 100% busy with threads that use 100% then Task Manager will show all cores being used at something less than 100%. So on a dual core if only one thread (that would normally use 100%) is active - Task Manager will have two graphs on the Performance tab show about 50% each.

You are not running SMP WU. Look at "CPU Usage:" at the bottom of Task Manager window, it says 50 %.

it is a classic uni-processor using only one core

 

[Turbonium]

I agree, but how do I get an SMP WU?

I'm going to try to see what WU it acquires automatically after this one is done. If it still goes uni-processor again, I'll have to manually configure it for the next round.

It must be a bug, right?

 

[GLeeM]

It must be a bug, right?

I doubt it is a bug.

how do I get an SMP WU?

On my a little older V7 client I have to go into "Configure" and then "Slots" tab and click the "Add" button where I have options to add uni-processor or SMP or GPU.

 

[Turbonium]

I doubt it is a bug.

On my a little older V7 client I have to go into "Configure" and then "Slots" tab and click the "Add" button where I have options to add uni-processor or SMP or GPU.

What happens if you change the settings during work on a WU? Like say it's working on a uni-processor WU, and then halfway through you change it to an SMP setup?

 

[biodoc]

Not a problem. Change the setting before your current WU finishes. Historically, you can get SMP WUs on dual cores but I'm not sure what the minimum requiresments are these days.