Hotter = Slower???

The forum for non-component-related silent pc discussions.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
MonsterMac
Posts: 330
Joined: Sun Mar 07, 2004 10:33 am

Hotter = Slower???

Post by MonsterMac » Thu Jul 08, 2004 1:16 pm

My friend seems to think that when things run hot (computer components) they run slower... Now I know this is not true, but can anyone throw some information at me so he will finally believe me that waht he's saying isn't true? Thanks

Splinter
Posts: 245
Joined: Mon Jul 05, 2004 9:01 pm

Post by Splinter » Thu Jul 08, 2004 1:30 pm

AFAIK, when your components are running hot, they cause more errors, so commands have to be resent. This will either make your computer act slower, or crash it.

AZBrandon
Friend of SPCR
Posts: 867
Joined: Sun Mar 21, 2004 5:47 pm
Location: Phoenix, AZ

Post by AZBrandon » Thu Jul 08, 2004 3:15 pm

It will perform exactly the same until it reaches a point where it's causing errors, as was said, although typically those will only be noticed as a crash. Even incredibly hot and unstable systems usually give the illusion of working perfectly fine right up until they crash. The only exception is Intel's thermal throttling which only comes in to play at extremely high temperatures, like over 180F or something.

dasman
*Lifetime Patron*
Posts: 485
Joined: Thu Jan 08, 2004 10:59 am
Location: Erie, PA USA

Post by dasman » Thu Jul 08, 2004 3:35 pm

Hotter is relative.

If your HSF @ 12v gives you 35c load and slowing to 5v gives you <60c, you're hotter and you're not going to have problems. Now, if you're at 70c, then you've got problems.

In general, if you're under the spec'd limits for the component, you shouldn't see a problem -- keeping in mind that you may reduce HD lifespan with higher temps. I like to keep my HD's under 45c, but others don't mind going over 50c (60c is the usual spec limit on HDs).


Dave

sthayashi
*Lifetime Patron*
Posts: 3214
Joined: Wed Nov 12, 2003 10:06 am
Location: Pittsburgh, PA

Post by sthayashi » Thu Jul 08, 2004 6:50 pm

BENCHMARK!!!!!

Really. Add a fanmate and run a benchmark at one temperature and then rerun it at a higher (but still safe temperature).

peacefulpc
-- Vendor --
Posts: 31
Joined: Sun May 23, 2004 12:01 pm
Location: oregon
Contact:

Re: Hotter = Slower???

Post by peacefulpc » Thu Jul 08, 2004 7:36 pm

MonsterMac wrote:My friend seems to think that when things run hot (computer components) they run slower... Now I know this is not true, but can anyone throw some information at me so he will finally believe me that waht he's saying isn't true? Thanks
someone correct me if i'm wrong, buy my physics classes way back in the day taught me that electricity moves with less resistance at lower temperatuters (e.g. superconductors only working at ultra-low temps)

this may not translate into much/any speed increase, but there is some scientific backing for his comment.

Laurent
Posts: 25
Joined: Tue Jun 15, 2004 7:14 pm

Post by Laurent » Thu Jul 08, 2004 7:41 pm

Well, it depends what "things" you are talking about.

It is true that transistors (and wires) are slower when they are hotter. This is due to a number of things including thermal noise.

Now, in general, this doesn't directly translate into any system speed difference. The vast majority of the logic in the chips in your PC runs synchronously to a clock. It performs a certain amount of work per cycle of that clock. This means that if the clock doesn't run faster, the same amount of work is performed. If the chip is cooler, the transistors/wires might be fatser, but the only result is that they will finish their work earlier within each clock cycle. But this doesn't translate into more work being done. There are exceptions to this like asynchronous logic or partially asynchronous logic, but there is either none of it of very very little in a PC.

However, you can transorm that extra speed in actual system speed if you increase the clock frequency (now that the logic is done a bit before the end of the cycle). Changing the clock frequency is the only way to turn extra transistor speed into extra system speed.

That's why overclockers want to run their system cooler (so that they can overclock more). That's also why Intel quotes increasingly lower maximum temperatures for their fastest processors (like 75C for a 2.8C P4 versus 64C for a 3.2EE). That allows them to qualify more processors at a given speed level (i.e. a given CPU might run at 3.2GHz at 64C but not at 75C).

Also, chips tend to dissipate more power at higher temperatures, so it also allow them to stay more easily within their power budget.

Laurent

nutball
*Lifetime Patron*
Posts: 1304
Joined: Thu Apr 10, 2003 7:16 am
Location: en.gb.uk

Re: Hotter = Slower???

Post by nutball » Thu Jul 08, 2004 11:01 pm

MonsterMac wrote:My friend seems to think that when things run hot (computer components) they run slower... Now I know this is not true, but can anyone throw some information at me so he will finally believe me that waht he's saying isn't true? Thanks
Your friend isn't into overclocking by any chance is s/he? :?

Inexplicable
Posts: 226
Joined: Sat Sep 06, 2003 5:59 am
Location: Finland

Post by Inexplicable » Fri Jul 09, 2004 12:09 am

I think we can safely say that faster things run hotter.

silvervarg
Posts: 1283
Joined: Wed Sep 03, 2003 1:35 am
Location: Sweden, Linkoping

Post by silvervarg » Fri Jul 09, 2004 2:36 am

Splinter:
AFAIK, when your components are running hot, they cause more errors, so commands have to be resent.
I have never heard of a CPU for a PC that would check the results of commands and redo them if the result was faulty. If you knew the results of the commands you don't need to run them at all, right?
So, to know the results of a command you need to have another CPU that does the same work. Another way to really check if things go fine is to run Prime95 in tourture-test-mode that checks calculations results against pre-calculated values. This way we can see if things go wrong, but on the other hand we do not do any usefull work since we only recalculate things that has been done before.

When components get to hot they cause errors, but these errors and not detected or corrected. Usually they cause the system to crash, but if you are on the border-line you can get just a few minor errors, resulting in wrong calculations.

Splinter
Posts: 245
Joined: Mon Jul 05, 2004 9:01 pm

Post by Splinter » Fri Jul 09, 2004 2:57 am

silvervarg, that's kind of what I was getting at, although I guess I didnt know quite as far as I thought I did :P

I know the CPU itself doesnt check calculations, because it has nothing to check them against, but I thought a well-programmed error trapped application might notice errant data from the CPU and ask for a second opinion.

Perhaps not tho.

bchung
Posts: 24
Joined: Sun Jul 11, 2004 4:36 pm

Post by bchung » Mon Jul 12, 2004 2:10 pm

I could be totally wrong about this, but I thought Intel chips had built in thermal protection. So if it gets too hot, the CPU automagically steps down the speed to avoid a meltdown.

Perhaps this was what your friend was referring to?

Post Reply