Athlon 1.4gig/Assus A7A: Nightmare combination!

[AlexNichol]

I'm having loads of problems in keeping my system at reasonable temperatures:

Athlon 1.4gig@266FSB
Assus A7A266 mobo
Assus Geforce 3
512MB PC2100 DDR
Maxtor Fireball Plus 60gig HD UDMA100@7200rpm
Hauupage PVR TV tuner card
Sounblaster Live!5.1

Coupled with a modem, network card and separate DVD/CD-Writer drives, my system generates a lot of heat.

The Athlon 1.4 gig is notoriuosly hot, and it seems that the A7A266 mobo doesn't help, as it seems to run pretty hot itself, and many A7A266s have a bug which reports the CPU temp 10 degrees celsius higher than it actually is. I'm trying to arrange the case fans as best I can.

I started off with a crappy Neolec Freezer II fan/heatsink but its performance was pathetic, running at 70celsius!

What is the best fan/heatsink under $100/£60 for a 1.4gig and an A7A266? I've read Tom's guide on heatsinks, but it doesn't seem to cover a system that generates as much heat as mine.

I'm not interested in watercooling or those expensive refridgerator coolers, just a good heatsink and fan.

Its driving me crazy as I want to get my new system up and running, but at the same time I don't want to waste time and money on a crap cooling solution.

Any suggestions?

 

[Kelledin]

Right now, the best air-cooling solution we know of is the SwifTech MC462. Not all motherboards can accomodate it though.

Kelledin

"/join #hackerz. See the Web. DoS interesting people."

 

[Crashman]

Yes, even with the 10C overreading, 70C=60C=TOO DAMNED HOT! Get the Swiftec, it's available for $50 at www.nexfan.com

Video killed my Radio Card!

 

[peach]

OK, no argument, MC462 is an intensely cool hs/hsf. but there are major drawbacks, not counting expense. Weight, mobo compatibility, and noise are all considerations too and I think that according to <A HREF="http://www4.tomshardware.com/cpu/01q2/010521/index.html" target="_new">Tom's May 21st review of the CPU coolers</A>, or his <A HREF="http://www4.tomshardware.com/cpu/01q1/010306/index.html" target="_new">March 6th review</A> there a number of hs/hsf that have great performance and are a better balance with the fore-mentioned characteristics.

I personally chose a WinFop32, based on tom's guide, and have been totally happy. I run a 1.2 AXIA (9 stepping) at 1.4 with an unloaded temp 77 F (25 C), typical running temp 79 F (26 C), and burnin temp at 82 F (28 C). I do have excellent case cooling, but even with the mobo sitting loose on my desk the temps do not get out of control and proc seems to have same cap on oc/stablity. This hs/hsf I have found to be plenty quiet when the case is on and was very inexpensive.

Sure, if cooling is the ONLY consideration, I would go with the 462, but that is not the case for me nor do I think that I would get a greater amoutn of oc with one. I personally would only choose the 462 if I wanted to use a peltier w/o water cooling.

<font color=blue>On company time.....

<b>Change the Sig of the week!!!!</b>

 

[Boondock_Saint]

YAY GO SWIFTTECH FANS! #1 Weight is of no consequence because it bolts onto the holes around the socket, not the clips for it. #2 It is fairly loud, but in a case with the cover on and other fans it is not really loud then.

MC-462-A = The only option for hardcore computer fans using air cooling.

 

[AlexNichol]

I went for a Swiftech, but it doesn't support the Assus A7A266 mobo!

Any other suggestions?

 

[peach]

OK - i understand about the weight and your are right - it really should not be a consideration. BUT!!!!!!

1st, the 462 IS more expensive.
2nd, It might not be compatible (check - if it is - mute point)
3rd It is louder - even though tolerance is preferential, it <i>is</i> louder,
and
4th is there a DEFINATE beneifit? I don't think so.

Before you go off and flame me - please just listen with an open mind. I spent a huge amount of time/effort/money adding incredible cooling to my case - I will not go into details here, but lets just say, everyone I knew (not comp geeks) thought I had lost my mind. But here is the kicker, Oh yes, when it is all on, I get a case temp of 69F and a burn-in temp of like 72-73F. BUT WHO CARES!!!!!!!!!! I can pull the mobo completely out and test again (I have hd's attached to the mobo tray so it is farily convenient to do that with a spare ps sitting around.) Anyway, I have tested it at length with a very very very very cheap winfop-32 hsf as the ONLY cooling and still reached the widely recoginized peaks of my proc (AXIA with 9 stepping). I can put this sucker in the Hurricane of my case and it still isn't stable above 1.43, and not rock stable above 1.4!!!!! So, I repeat, who cares if the proc will run a <i>little</i> (emphasis on little) cooler with a 462? What tested and proven advantage have YOU got to prove me otherwise? Please, no comparison to some sh!t fan. I want a top ten contender from the latest previously linked tom's May 21st test review.

I maintain that if you have a farily good hsf, you will reach the peak performance of you proc before you will ever need the 462. Unless you are doing peltiers or a vapochill, which <i>thermally</i> accelerate the proc - these processors have speed limitations reached without the expense/hassle/noise of the 462.

<font color=blue>On company time.....

<b>Change the Sig of the week!!!!</b>

 

[peach]

See, so compatibility is an issue....


I am sure any one of the top finsihers in the previsiouly provided link would work. As mentioned - I have found the WinFop32 to be more than acceptable.

<font color=blue>On company time.....

<b>Change the Sig of the week!!!!</b>

 

[peteb]

Jeeze - you must live in an ice box to have a system temp of 69F - 20.5C. I'm guessing your room must be like 18 or 19C (64.5F-66F) right? Most people's rooms are 25 degrees and many are up to 35 degrees, which is why you get a whopping temperature reduction over others. I'm not saying your system isn't sweet, but any reasonable case setup will keep the system/mobo temp withing 1-2 degrees of room temperature.

-* This Space For Rent *-
email for application details

 

[AlexNichol]

I'm having serious second thought about my Assus A7A266. Even if I do manage to get a decent cooling system in, I have read on various boards across the web that the mobo has compatibility problems with some DDR modules, Soundblaster Live!, TV tuner cards, and the Geforce 3, all of which I have bought and hope to put into my dream system.

Its a real pisser: I checked Assus website for compatibility problems before I bought the board, and it said nothing about the above components.

The message to be learned here: always check boards like this before buying new components!!!

 

[Boondock_Saint]

Well, my Asus A7M266 w/Swifttech MC-462-A cools my Athlon 1.4GHz great. Now, you make a good argument about it not being able to go cooler than the room, but what about how fast it dissipates heat? The less time your CPU is heated to a great degree, the longer and potentially faster it runs. This is what makes the Swifttech great. It dissipates heat VERY fast. But dissipating heat at low temperatures becomes more and more minute. So, for my next point: holy cow man your room is cold! No wonder it doesn't show as much of a difference. 69o F holy cow. What the hell is your cooling bill? $1000 a month?! I'm sorry, but most people usually run it quite a bit higher. We don't enjoy meat locker temperature. In any other normal room, the Swifftech will show HUGE results over others.

 

[AlexNichol]

I've decided to go with the GlobalWIN CAK 38, with the copper heatsink.

The thing is, this heatisnk looks likes it weighs a ton!!
Apparently it is one of the heaviest heatsinks around.

Has anybody had any problems installing this? I'm afraid of cracking my Athlon with it.

Also, do you use thermal paste with this sink? What about Artic Silver II?

 

[Guest]

I've had some cooling problems too, but aside from that let me ask you a few questions since I'm using an Asus A7A too.

Though my system isn't quite at the level yours is, I'm also using almost all of my PCI slots and have a ton of cards in there. Ensonique audio card, DVD decoder, Ethernet card, and Hauppauge TV Tuner.

Here's my question. Are you getting weird IRQ's for all your stuff? Like having your sound card and video card sharing the same IRQ. Your TV Tuner on IRQ 3.

Also, the big question. When you start up the Hauppauge software does your computer just completely freeze up? So far I don't have any RF cables hooked up, but that shouldn't make my comp freeze up if I try to use my TV Tuner? Right?

 

[Guest]

Yep they are heavy, and yes it is easy to crush a cpu core if you are not careful. Easier with MB flat or even out of the case, sit it on the cpu hold it in position and then attatch that clip try not to let it rock over the CPU core coz thats bad! I have the Dragon orb 3 ($53.00 AUD)and thats just as heavy, and dare I admit it, I crushed a cpu core :frown: and learnt a hard lesson in patience. But I have my HSF nicely mounted now and doing a great job on a new T'bird 1.4g cpu.
Artic silverII is the best.

<font color=blue>THG needs 2 change the sig' of the week errrr century!</font color=blue>

 

[killall]

SwifTech MC462, easily the best out there... if a little (ok not a little but very) heavy

if in doubt blame microsoft...

 

[AlexNichol]

I haven't actually booted my system yet, as I don't think there's any point until I get my CPU temperature down, so I haven't tried my TV card. I was reading some message boards over at Anandtech and lots of people have had problems running TV cards on their A7A266.

By the way, has anybody tried installing the GlobalWIN CAK38 on an A7A266? Will there be enough room?

 

[peach]

Alrighty then.... I have been very busy working so I haven't had the time to respond properly to this. Although a subect not fully developed, it is not one that I think is *overly complicated*.

You see, it al comes down to why you modify things. Let's say you had a Mustang Cobra. Say you wanted to modify it to go faster and/or quicker. Say you put something* that costs $200 on the engine that gives it an extra 50 horsepower and that in turn makes you car pull .07 off your ET and gives you an extra 5 mph. Cool. Let say they make another part like yours, but is ABSOLUTELY better, but it is $700. Say a buddy of yours puts it on his Cobra that was identical to yours and his goes the exact same 0.07 ET faster with the same 5 mph gain. Let us just say that there are about 30 of you and all of your buddies have the same Cobra and all of you vying for more hp go with either of the two choices and all of you get the same results. Would you not agree, that maybe the $200 is the smart purchase??

Now about me living in a freezer and paying Trump's light bill. I have a small engineering office with one other partner and our office supposedly had a thermostat setpoint of 72 F. My office usually ran 72-73 on a temp gauge. But, my case temp always ran 71. Last months bill was extrodanarily high - of course it was July in Mobile AL. Now it happens to be that we are mechanical engineers and had a vendor replace our thermostat last week (nice perk!) with a programmable one. Now the setpoint is 74F during the work day and 78F on weeknights and weekends. But still, my case temp runs about 73 during the day, and typical running load 77F. Of course, I only mention this to refute this rumor of me paying out the ass to live in an ice box.

To be honest this clouds my point, for I believe that I could run this coputer outside where it is a nice 99% RH at 85F. As muggy and hot it is out there, I bet you my computer would still run stable at 1.4 and crap out above 1.43. So again, why spend more?

*I can not specify what kind of engine mod or everyone would flame me saying it a more expensive part <i>would</i> help and in turn they would miss the point.

<font color=blue>On company time.....

<b>Change the Sig of the week!!!!</b>

 

[peteb]

peach - in your post you said you had a case temp of 69 degrees, and now you are saying your room temp is 72-74 (but higher).

Anyway - my point is that your case temp _has_ to be no cooler than your room temp - the thermostat is probably the issue here. It is not possible to cool a lump of dry metal below ambient temperature using air only. Cool it to ambient temperature is possible, not below, that's my point.

-* This Space For Rent *-
email for application details

 

[peach]

my point is that your case temp _has_ to be no cooler than your room temp

This implies average room temp. Although i understand your root theory here, this is not actually true. I explain below.

the thermostat is probably the issue here

It most certainly is. It is difficult for me to be complete without being long winded; I apologize. I design air conditioning systems for a living and could easily get side tracked. Set point temp and actual temp are two TOTALLY different things. First there are hot spots and cold spots in rooms. The air coming out of the vent is usually about 55F. Therefore, depending on your case location relative to the air path from the supply air vent to the return air grille, the case temp could actually be considerably cooler or warmer than the setpoint on the thermostat. The thermostat is measuring the air only where it is at, and that is generally near where the return air grille is or the most important room in the zone.

in your post you said you had a case temp of 69 degrees, and now you are saying your room temp is 72-74 (but higher).

In my office, our setpoint temps were becoming erratic, and at one time my case temp was actually down to 69F when I did some benchmarks. I have done them at much higher temps as well and confirmed the same chip limitations. Our office temp is now generally 2F higher with the new tstat than it used to be.

So anyway, it happened that where my case sits, it is usually a degree or two lower than the setpoint. Now, in many homes, to save on cost, ducts are routed down the center of a house and all the vents are toward the interior. This is of inferior design as the air tends to short circuit the room and go right back out each room’s door to the return air closet. In this case, the rooms will tend to be a little warmer than the setpoint, as you figured. So that is the confusion of temps being lower than the setpoint.

It is not possible to cool a lump of dry metal below ambient temperature using air only

Of course not, a hs that is dissipating heat generated will <i>never</i> be lower or even be the exact same temperature than the air around it. This would violate the 2nd law of thermodynamics. Now if the comp was not on, obviously the hs would acclimate to the surrounding temp.

I am <b>not</b> trying to say that a lesser expensive WinFop32 will dissipate as much heat as a 462 <i>or</i> that room temperature does not influence heat dissipation. In fact, they both have everything to do with heat dissipation. The delta t (room temp related), convection coefficient (related directly to fan and hs material), and surface area (also hs design related) are major contributors. There are many others, too. I have taken an entire course on Fin Theory - this is a subject of high interest to me and I have some expertise in it. In fact, I have designed a completely different hs design that I wish to someday manufacture. It is a shame so many HS (not all - but most) are designed by people who have not a clue what they are doing except for common sense and trial and error.

****I am looking for either acknowledement or quality debate on the following*****

<b>What I am suggesting</b> is <i>the performance as a function of heat dissipation efficiency plateaus</i>. This relationship varies for different computers but I insist that it will plataea for all computers. Now, as far as an air-cooled Duron/Athlon is concerned, the Winfop32 and a Swiftech MC42 may have widely different heat dissipation rates, but they both lie on the flat part of the curve -- they are relatively equal. Therefore, if there is no performance gain - WHY SPEND THE EXTRA MONEY AND DEAL WITH EXTRA NOISE?

The art of engineering isn't designing something that will work - but designing something that is just a little more than will ever be necessary, not douible, triple, etc. There is another term for that....

<font color=blue>On company time.....

<b>Change the Sig of the week!!!!</b>

 

[peteb]

Hmm - well I guess we are mostly on the same page then, however your closing statement lost me totally....

What I am suggesting is the performance as a function of heat dissipation efficiency plateaus. This relationship varies for different computers but I insist that it will plataea for all computers. Now, as far as an air-cooled Duron/Athlon is concerned, the Winfop32 and a Swiftech MC42 may have widely different heat dissipation rates, but they both lie on the flat part of the curve -- they are relatively equal. Therefore, if there is no performance gain - WHY SPEND THE EXTRA MONEY AND DEAL WITH EXTRA NOISE?

Having used both a WinFop with Delta Fan and a MC370 with the same delta fan, and a 462 with the 80mm delta fan - -I fail to get your arguement about them performing identically.

In comparison of the Fop and 370 with identical fan, I found the Fop (on my setup tested some 30 minutes apart with stable room temps) to be 5 degrees (C) different. For final loaded temperatures, on my setup, the 462 was another 3-4 degrees cooler at slightly faster rating.

Using delta fans on all I found the noise to be roughly equivalent (to my ears if not to a measuring system) but the performance to be dramatically different.

Now this as I see it (jump in any time if I really look dumb arguing cooling and heatsinks with a guy who designs aircon setup and studies fin theory for a hobby) is the crux - given equal ambient conditions, the effectiveness of a heatsink is totally dependent on the following criteria.

Surface area of cooling fins
Thermal conductivity of material used
Distance travelled to get to cooling fins.
Efficiency of airflow and heat transfer in the heatsink.

Now I am lost as to what you are saying - are you saying the the performance of each sink is roughly equivalent on the scale of things? I know through experience that the cpu performace available rises from the WinfFop up to the Swiftechs.

Please elaborate on your arguement!!

-* This Space For Rent *-
email for application details