*Discussion Time* What has more effect on CPU temp? Increasing HSF speed or increasing case cooling?

[MichaelD]

OK gents; it's time for some intelligent discussion. Do we have our thinking caps on? Good.

This is not as easy a question as it seems. This question is for those who have "normal to quiet cases." If you have five, 5,000rpm 80mm case fans, this is not your thread.


The following things are assumed:

1. You have a nice CPU HS like an AX-7, SLK800, Alpha8045, etc. and are using AS2/3.
2. You have a mid-to-quiet fan on the CPU HS, not a Delta screamer.
3. You have decent case cooling, at least one intake, one exhaust fan.

I'm one of many who've recently gotten on the Overclocking as an Everyday Thing bandwagon. I've noticed that when my CPU temp breaks 50C, Prime95 errors. But as long as we don't break 50C, it runs for days straight and is happy.

I know that CPU temp is greatly affected by:

1. How efficiently the CPU HS removes heat from the CPU
2. How efficiently the case ventilation system removes heat from around the CPU HS. I.E. a hot plate will stay hot in a hot oven longer than a hot plate in a cold oven will. (hey, that was pretty good! )

So, the question/scenario has popped into my thick head:


You set up two identical systems and put them in the same room and run the same stress programs on them. After a few hours, both the CPU and Case temps are identical.
Then, in System #1, you put in a faster CPU HS fan and in System#2 you leave the CPU HSF alone, but make the case fans spin faster/replace them w/faster fans.

Theoretically, which system (#1 or #2) would have the lower CPU temperature after a few hours of running stress programs? If I had the money and the time (in that order ) I'd do this myself.

What say you, Anandtech Wizards?

 

[Jackbike]

I am no wizard, but here goes. The question poses an either/or scenario, implying that one is superior, asking which. I don't buy either/or, as interactions occur which cannot be ignored. So much for the philosophical.

My sense of logic asks, does the in-case air flow pattern change by increasing case fan speed? If it does not in the area where the cpu cooler gets its air from then changing either case fan speed or cpu fan speed would get the same result -- but only if the pattern of air flow is constant through the cpu cooler giving unchanged efficiency. As the air speed through the cpu cooler increases there may be points at which greater turbulance/resistance to air flow occur, reducing the efficiency of the cooler. Room temperature is your lower limit (given no ducting of house air conditioned air to the case) for in-case temps; after that, fan speed on heat sink is what you have left.

Again, either-or questions ain't fair if other factors are involved!

 

[MichaelD]

Thanks for the reply, Jackbike. AFA case temp/cooling goes, my logic is that air is moving faster thru the case, thereby taking more of the hot CPU air with it...at least that makes sense to me. Now, with the faster CPU HS fan, the faster fan is pushing more air "off" the HS, thereby keeping the CPU cooler...but that extra hot air has to have someplace to go...

 

[pillage2001]

You cannot always push air through the fins. You need to push COOL air and not any hot air. Also remember, moving air cause friction and rushing air cause more friction and friction cause heat. It'll be a time when high air flow will increase temps or won't decrease temps at all.

 

[Budman]

You can have the best heatsink in the world,but if you're pusing hot air on it it's not going to run cool.

I say a cool case = a cool cpu.

 

[MichaelD]

Originally posted by: Budman
You can have the best heatsink in the world,but if you're pusing hot air on it it's not going to run cool.

I say a cool case = a cool cpu.

Wow..you had the same gorilla icon since I got here...now you changed...all for a clown? Oh, the humanity!!

*ahem* Back to the topic...

I currently have two case fans, but have provisions for four. I have considered adding a fan in front to pull in more room air. My intake/exhaust fans are speed regulated by the PS..they turn very slowly. The powersupplys' two exhaust fans actually suck most of the hot air out of the case....it would be a simple thing to add another intake case fan...I may just do that tonight.

I just don't want to wind up with "a screamer case" again, you know?

 

[Budman]

Yes I am test driving the Icon.

I will probably go back,when you're a clown nobody takes you seriously.

 

[Budman]

I feel much better now.


Where's those Bananas at .


I was getting tired of those cream pies in my face.:Q

 

[MichaelD]

Ah, the old Budman is back...you scared me....clowns are evil.

 

[Aiden]

Either change should lower your temps with the scenario you describe. im not an expert,but it seems to follow the basic rules of physics,which has very little to do with computing

scenario #1- upgrading to better heat sink and powerful fan should allow you to make the transfer of heat from the cpu to the heatsink work more effeciently. this would probably be the best improvement you can make all other things being equal. reason is the faster you can disperse the heat buildup,the better of you will be.

scenario #2- should still see an improvement because your moving more air through the case,thus removing the hot air faster,while introducing more air that is at room temp. while it is true,that moving air causes friction,realistically the ammount of air we are talking about moving will cause minimal friction ,and heat buildup i would think.

think of scenario #1 as direct cooling on the cpu,while scenario #2 is indirect cooling on the cpu.


you still wont see a drastic change,because unless your doing some extreme cooling,with water/antifreeze etc etc,you wont cool your computer below room temps.


for my antec 1080 amg i bought 2 vantec stealths to go with the 3 antec stock fans that came with it. put the vantecs on the front. have 3 intakes <2 front,1 side> 2 exhaust<rear>. to be honest,i cant tell the difference between this computer i have now,and my old 600mhz amd with 2 fans. personally i think all this "loud" case stuff is overrated,my antec hums a bit,but thats it. btw, everything but my CPU fan,and my PS fan are run through molex connectors.

 

[MichaelD]

Aiden, nice post.

I agree w/you that "the loud case stuff is overrated." In fact, I used to have FOUR of those same Antec Ball Bearing fans for case fans. The case sounded like a damn hovercraft, to be honest. After creating a quiet system, my case temps went up by maybe 1-2C. Big deal, you know?

I use Vantec Stealth fans for all my case fans. I currently have three systems running, using six of these fans as case fans. I have also done the power-supply-fan-swap using two of the Vantecs in my "lesser two" systems...it makes a difference.

My main rig has an Antec Trupower in there and it is quiet. What I think I'm going to do is add an intake fan to my case. If you count the PS as an exhaust fan (and it is, technically) I currently have two exhaust fans and one intake. Granted, the front of the case is very "open" (Antec 1040 clone) but "forcing" more air in there via another fan can only help...I guess.

*sigh* Too bad there's no such thing as a "leakproof" water-cooling setup. I'd be on that in a minute.

 

[chizow]

I would say, after a few minutes, if it didn't hit its "break point" of 50C in your case, that the system with the higher powered fan would lower the CPU temps closer to the system temps. If you started both, and 1 system had a better fan, and good exhaust, I would say that the one with the better fan may very well have lower system and CPU temps.

Chiz

 

[Drsignguy]

My name should kind of tell you all in what I do for a living. Think of a BIG message unit as a very large computer. There are many now from the basic black and white to your now up-to-date LED units. They all put out a great deal of heat.
So for starters, the first units had just a few fans in them with some kind of open exhaust. Not very good. Then we started to make sure there was a pressure system, more air in and forcing the air out by the drivers (electronic componets) and the lamps, thereby cooling each. Not to bad. Then a system that took in as much as it takes out. In with the good and out with the bad. Equaling the cfm.
Most of these systems had exhaust at the top of every unit.
Well now they are putting air cond. units into them, completely self containing each unit like a refrigerator, not adding cold air, just removing the heat. This is what all air cond. units do. So, by having enough cool air forced into your system, it will have a better effect.
CFM is better than RPM. To answer your question as my own opinion, I would prefer to have a quiter system with a good amount of airflow rather than have a system to sound like an f-14.
This is a great thread

Docs operations



3DMarks

 

[MichaelD]

Thanks, Drsignguy, nice reply. I am strongly leaning towards more CFM throughput. I guess I"ll have to run some tests to see.

 

[Drsignguy]

No problem glad to have an input.

 

[pelikan]

I think that if you remove the side panel from your case and your cpu temp lowers more than 2-3c then you need better airflow in your case. If the temp doesn't go down much then more airflow over the heatsink is the best bet.

 

[MichaelD]

Originally posted by: pelikan
I think that if you remove the side panel from your case and your cpu temp lowers more than 2-3c then you need better airflow in your case. If the temp doesn't go down much then more airflow over the heatsink is the best bet.

Hmm, now there's an experiment worth doing. I can tell you this:

When I was in the "testing it out after just building it" phase, I had the case sitting on the floor (as opposed to inside my main PC desk) and the side panel was off. My CPU temps were lower. Well, I guess that means I need more airflow.

That conclusion begets the question: More intake or more exhaust? Currently, I have one intake fan (but the front panel has provisions for two, i.e. plenty of open grill in front) and technically, two exaust. One actual exhaust fan and the powersupply, which has two fans and actually exhausts most of the hot air from the case.

Do I add another intake fan or exhaust?

 

[THUGSROOK]

Originally posted by: pelikan
I think that if you remove the side panel from your case and your cpu temp lowers more than 2-3c then you need better airflow in your case. If the temp doesn't go down much then more airflow over the heatsink is the best bet.

the man knows his stuff
and he is 100% correct!
my temps actually go up if i remove my side panel :Q

 

[KF]

Maybe people could do better cooling with more measuring. Use one of those digital indoor/outdoor thermometers with a probe and place it in different parts of the case (with a stiff wire). You will see that it is hotter around the CPU than elsewhere. (Duh!) There is no such thing as a single case temperature. The strength of the airflow created by the CPU fan dominates and overwhelms the gentle flow created by the usual case ventilation fans, and drives the cool air away from it. The CPU fan basically recirculates a large part of its own hot air, just because of its own strength. Really strong CPU fans have a tremendous wind at the base exiting from the heat sink which helps drive the hot air further away before it recirculates. But the space of the case is very confined, and the CPU fan's own air still winds up circulating back into the fan. It sucks as much air in as it blow through the heatsink.

The specific flow of the air around the CPU counts as much as the CFM. I have found that even with a wide open case, there is still a halo of hot air surrounding the CPU fan. I use a big and relatively quiet 120mm fan blowing cool room air directly at the CPU fan to make the hot air around it move out through the back vent for an 80mm fan. It actually works better at a 45 degree angle and about a foot away from the CPU fan, but that is difficult to arrange.

 

[MichaelD]

Bump my own thread b/c KF's reply is right on the money. Someday I'll have enough dough to setup multiple idenitical rigs and benchmark all the changes...until then, I count on you guys!

KF; 500 posts since 1999? Where'ya been buddy?

 

[rogue1979]

The worlds greatest heatsink with a 1000cfm fan will only cool as good as the case temps! Superior case cooling with a weak heatsink/fan will usually cool better than a weak case cooling and great heatsink/fan. The two most important factors for case cooling are the correct intake to exhaust ratio and the placement of the fans themselves. Poor design/planning for case cooling will make a noisy monster that doesn't do much. I prefer a cheap case with a good power supply and alot of modding. Buying an expensive case and the doing further modding will not improve results over a modded cheap case if you know what to do. I don't have a degree in thermal dynamics, but I am willing to try a thousand different configurations for case cooling and test thoroughly. It is a lot of work, but my efforts have paid off. I have five case fans all runnning on a 7-volt mod. Intake blowholes to feed the heatsink and video card cool air. My full load case temps are 26c (ambient temp). I actually run a degree or two higher with the case covers off. Quiet and cool, with an AX7 running a 3300rpm my 1600+ @ 1800MHz runs 47C under full load. The cost was very little, but I spent many man hours and several cases perfecting the case cooling. Highly overclocked, reasonably cool, and very quiet, it was all worth it.

 

[lookin4dlz]

Perhaps this experience will help.

With my stock hs/fan vented to the exterior of the case (using flexible tubing sealed to the hs fan and to a case fan so that I have a fan on each end of the tube and fresh, cool air from outside of the case). This would be similar to a fast, quiet hs fan.

If I remove the case fans and leave the external CPU hs venting, the system temp increases and so does the CPU temp, but not proportionately i.e. some of the system heat gets transferred to the CPU but the extra external cooling helps.
If I remove the case fans and remove the tubing (so that I have only the hs fan running), the CPU temp increases to where it's about what the system temp is.
If I leave the case fans and remove the tubing, the CPU temp increases proportionately with the system temp.
If I leave the case fans and leave the tubing, the CPU increases slightly less than proportionately with the system temp.

So, it appears to me that there is a great deal of heat transfer from the components (video card/ram/hard drives) to the motherboard to the CPU. It also appears to me that the case cooling affects changes in CPU temp more than CPU cooling (this is with a P4 which desn't run very hot to begin with, in fact the hs only feels warm to the touch). Therefore, it seems that CPU temps could be overcome by either extreme (and expensive) CPU cooling or by a good combination of case cooling and CPU cooling. What is interesting is that the overclocking community has, through trial and error, come to a darwinian conclusion that combination cooling (since this is what the majority use) offers the best efficiency/effectiveness.