mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Hardware

Reply
 
Thread Tools
Old 2007-02-03, 22:00   #1
kdelisle2005
 
Mar 2006
Michigan

11 Posts
Default Laptop - why does colder equal slower??

I have a P.III laptop running Prime95. My timings are about 0.240 seconds per iteration (in the house).

I put the laptop in my Michigan garage for an overnight run (about 5-10 'F) and my timings increased to 0.269 seconds per iteration.

The laptop is not overclocked in any way. I was curious on if lower temperatures would affect the speed. It did but in the wrong direction.

Does anyone have an explanation on why colder means slower??


I performed the same experiment with my desktop (P4 HT) and the same thing happened (i.e. timings increased).


thanks
kevin
kdelisle2005 is offline   Reply With Quote
Old 2007-02-04, 00:30   #2
moo
 
moo's Avatar
 
Jul 2004
Nowhere

809 Posts
Default

are you running a ll or are you running a benchmark
moo is offline   Reply With Quote
Old 2007-02-04, 00:44   #3
kdelisle2005
 
Mar 2006
Michigan

10112 Posts
Default Laptop colder>slower

I am running LL as described below:



exponent bits iteration run / to go / exp date updated date assigned computer ID Mhz Ver
-------- -- ---- --------- ----------------- --------------- --------------- ------------ ---- ---
17980763 D 67 1441600 12.5 38.6 98.6 26-Jan-07 14:30 22-Jan-07 11:56 Laptop 1002 v19/v20
kdelisle2005 is offline   Reply With Quote
Old 2007-02-04, 06:53   #4
ixfd64
Bemusing Prompter
 
ixfd64's Avatar
 
"Danny"
Dec 2002
California

7×331 Posts
Default

Aren't colder temperatures actually supposed to speed up the processor?
ixfd64 is offline   Reply With Quote
Old 2007-02-04, 14:56   #5
S485122
 
S485122's Avatar
 
Sep 2006
Brussels, Belgium

2×797 Posts
Default

Colder temperature should not affect the speed of a computer. But now we have evidence that it can ?!? I think only of one explanation, see below.

It is not true that colder means faster, what is true is that if a computer is colder you can have it run faster via overclocking. Another consequence of colder temperatures is that components wear less.

There could be a problem with disks drives and fans if the temperature gets really cold because the lubrificants of moving parts would not be operating to specifications.

I do not think that -15°C (5°F) is good for disks or fans, the fans might have stopped altogether resulting in higher internal temperatures, and thus throtling and thus higher iteration times ! You should check this.
S485122 is offline   Reply With Quote
Old 2007-02-04, 15:08   #6
dsouza123
 
dsouza123's Avatar
 
Sep 2002

10100101102 Posts
Default

Possible scenarios

Scheduled programs run ie defrag, checks for updates overnight.

Screen saver starts maybe using an hour or so delay
versus normally mouse or key activity so it is never activated.
The screen saver would be one that uses a fairly small amount
of CPU resources but would still cause extra task context switches
and maybe some extra page faults.

Difference between running plugged in or on battery.

Screen closed so less heat escapes from the keyboard surface.

Temp sensor detecting low temp so the fan(s) stop
temp builds then speed step or equivalent kicks in.

Hard drive temp sensitive in some way.

Change in FFT size, not sure if it can normally happen during a LL,
maybe just if an rounding error because near border size
followed by a retry with the next size.
dsouza123 is offline   Reply With Quote
Old 2007-02-04, 15:33   #7
dsouza123
 
dsouza123's Avatar
 
Sep 2002

66210 Posts
Default

There are a few motherboards that have options for dynamic overclocking
for CPU intensive programs, ie ASUS Striker Extreme,
but the far more prevalent situtation is a CPU, may need MB or driver support,
that underclocks/adds wait states on increased temp,
ie Intel's SpeedStep or AMD's Cool'n'Quiet and PowerNow!.
dsouza123 is offline   Reply With Quote
Old 2007-02-04, 16:58   #8
kdelisle2005
 
Mar 2006
Michigan

11 Posts
Default

Here is a little more data:

I don't think the HDD ever stops because I hear a whirl sound and it is not the corner fan (cooling).

The corner cooling fan will cycle on and off as the as needed - about half each.

I also removed the 5" x 5" underside coverplate where the memory modules are - this allows a lot of heat to escape.

I have the laptop in the closed position (so no screensaver) with the bottom facing up.

Memory modules are hot when touched (inside house)
Memory modules are much cooler and not warm - maybe 60-70 F (outside garage)

I have the "iterations between screen outputs" set so I get an update every 30 minutes. I also let the machine run in one location for a least 1/2 day.

There are no scheduled background tasks.

All the settings in Prime95 remain asis regardless of location.

The laptop is a very typical Dell business laptop. It is a spare at work so I run Prime on it (at work). The funny part is that I use it to keep my hands warm because my cube is so damn cold. With the near 0 'F temps this weekend, I thought it would be a good time to try this experiment out at home.

We already know that there are operating temp. ranges but typically with PCs that is more of a concern on the upside temps; and I would never want to start a PC from a dead cold freeze.


I am curious which what factors (e.g. mechical and/or electrical) are at play here.


Try it yourself for a day - do your timings change??


I will try a few Benchmark tests and see what I get.



thanks
kevin
kdelisle2005 is offline   Reply With Quote
Old 2007-02-05, 06:50   #9
TehPenguin
 
TehPenguin's Avatar
 
Jun 2006
Perth, Western Australia

2·5 Posts
Default

Most electrical components do actually have a recommended running tempature. Although, the tempatures mentioned do not sound very extreme, it is possible that your CPU could be designed to work around the 100°F mark (plus or minus a few tens of degrees).

Although it may appear counter intuitive, having the CPU that cold may result in the resistance being too low for the circuits and so they're not running at an optimal speed, since the chip was designed to have a resistance at a set range (which is why that 100°F range is its optimum)

See if you can dig around the Intel website for design specification for your CPU
TehPenguin is offline   Reply With Quote
Old 2007-02-06, 09:12   #10
PCZ
 
PCZ's Avatar
 
Jun 2006
Chertsey Surrey UK

1011001102 Posts
Default

The reason for your increased iteration times is most probably due to the low temperature decreasing the resonant frequency of the timing circuitry.

The frequency that quartz crystals oscilate at is affected by temperature.

If you want to take advantage of the low temperature then you will need to overclock the CPU to a point it wouldn't run reliably at inside the house.

Last fiddled with by PCZ on 2007-02-06 at 09:13
PCZ is offline   Reply With Quote
Old 2007-02-08, 22:48   #11
nngs
 
nngs's Avatar
 
Jun 2004

22·3·5 Posts
Default

Quote:
Originally Posted by PCZ View Post
The reason for your increased iteration times is most probably due to the low temperature decreasing the resonant frequency of the timing circuitry.

The frequency that quartz crystals oscilate at is affected by temperature.

If you want to take advantage of the low temperature then you will need to overclock the CPU to a point it wouldn't run reliably at inside the house.
If it's true, we should expect the computer clock runs slower instead of Prime95 timing slower (Prime95 uses longer 'second' to calibrate its work).

Last fiddled with by nngs on 2007-02-08 at 22:50
nngs is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
square root equal to a negative LaurV Homework Help 34 2013-03-03 09:35
Derivative of Sum not Equal to Sum of Derivatives jinydu Math 2 2008-11-24 10:22
Is version 25 a lot slower? Jud McCranie Information & Answers 3 2008-11-12 15:21
Special exponents with equal digits in series hbock Lone Mersenne Hunters 0 2004-01-03 20:51
Why is my PC slower than comparable PCs? markhl Software 15 2003-07-22 18:47

All times are UTC. The time now is 15:52.

Tue Nov 24 15:52:35 UTC 2020 up 75 days, 13:03, 4 users, load averages: 1.80, 1.74, 1.78

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.