View Full Version : XP and beyond. Is cooling going to be a problem?
08-14-2002, 05:19 PM
I've been thinking about my new XP system and how the XP2100+ cpu I have installed generates as much heat as my 1.4 Thuderbird did.
I'm getting a 2200 Thoroughbred in a couple of weeks and was concerned about cooling it!
I use an SK6 with a delta fan at the moment which does a good job, however even the best heatsinks are going to struggle when the 2600 and 2800 CPU's come out.
Even if the sinks can cope with the heat its a sure bet that we need to have higher CFM fans to provide enough air to remove the heat.
So.... Bigger sinks, faster louder fans.... when will it stop?
Thankfully I'm going H2O soon :)
What do you guys think about this the XP heat issue?
08-14-2002, 05:31 PM
Enigma what are your temps at ?
08-14-2002, 05:33 PM
My 1.4 Thunderbird ran at 44*C
The 2100XP runs at 42*C
These are full load temps.
08-14-2002, 05:39 PM
Damn, those are good tempts. I get about the same w/ my Alpha 8045 + 39CFM 80mm fan on a 1.7G XP.
08-14-2002, 05:42 PM
I have slightly extreme case cooling AmStoned.
I've been modding my poor old case for years to get the right balance of air.
I must be honest though, these temps are recorded off the motherboard sersor. I've just recieved a Digidoc 5 in the post today, so when I get my new case I'll let you know the 'real' temps.
08-14-2002, 05:52 PM
I see... Yeah, I get my temps by sticking the flat thermal probe head between the HSF and CPU (not the core), w/ the tip placed and taped next to the core.
08-14-2002, 05:58 PM
Thats what I plan to do with my Digidoc Amstoned. I'm sick of the reading that the motherboard sensors give. Hopefully the Abit board will give a better reading.
Well I'm off to bed now. Too much computering for me today :)
08-14-2002, 06:27 PM
Yup, very decent temps Enigma, I don't think you have much to worry about.
08-14-2002, 07:44 PM
the number one factor (in my mind) to reducing CPU temps is oddly enough case temperature. if your ambient surrounding is hot, no CPU fan can do any magic better then the air its sucking from.
i have seen people buy massive delta fans to cool the cpu and totally disregard their case cooling. where does all the hot air go - you guessed it.
from the looks of it, you have done your homework and are getting decent temps with your current setup. FWIW, my case temps are always within 1-2C of ambient temperatures. i currently use an Alpha 8045 with a 92mm Panaflo M1A fan and I get about 40-42C using a CompuNurse at 1.7G XP1800+ The IWILL thermistor always registers 4-6C cooler then that.
Those newer XP models will be plenty hot no doubt - higher heat/unit area. but keep in mind AMD does state the operating specs of the core to be able to handle greater then 60C (forget the actual limit) so we should be ok...although we don't like it. they also probably will try to optimize the core a bit to prevent from using more elaborate heatsink desings to satisfy their OEM customers.
08-14-2002, 09:30 PM
I'll agree that those are excellent temps Enigma, you must have your case air flow figured out really good. I'd like to know if you are OCing your XP and what case you're using.
08-15-2002, 03:56 PM
I've got a generic full tower case from my local Maplins store (Electrical and gadget shop in the UK)
It cost £39 and has loads of pre made fan attachments. So far I haven 1 X 80mm at the bottom (front) blowing in, 1 X 92mm side blow-hole (blowing in) and 1 X 80mm at the back sucking! + the PSU fan.
I'm still on the lookout for a new case to build my new system in. I like the look of the Chieftec but their too pretty to hack up and cost a small fortune.
Maybe I'll stick with old faithful.
08-15-2002, 07:50 PM
The genral point you raise about where all this heat goes to is an interesting one. There are two issues to consider, efficiency of the chip design, and the moving of heat away from it.
Chips are getting smaller so that they can go faster. They can go faster because they draw less power when they are smaller and therefore generate less heat. Of course there is a limit to how small these chips can go which is one restriction of the computing power they can generate. By making chips smaller you also create the problem of moving heat away from it.
To transfer heat, you need a large surface area, smaller chips have smaller surface areas and it is therfore more difficult to move heat away from them. The problem of cooling a chip can be split into two parts, taking the heat away from the core, and dissapating the heat. Moving the heat from the core involves getting the maximum heat transfer from the core to the cooler. This is boosted by the use of termal pastes and a good mating surface. Disapating the heat is a more comlex problem. The heat needs to be moved away from the mating area. Most people use a heat sink and a fan to draw heat away from the core. Decent airflow to remove hot air is also important. As temprature increase more and more people will move over to liquid cooled system simply because these can move a greater deal of energy from the core, much faster. However these solution also end up transfering the heat to a radiator(heatsink) and then use air to disapate the heat in the coolant. Eventually even this solution will not be enough as the radiators will become impractically large and the heat they give off will also become a problem.
Basically we can't just continue to ramp up the speeds to get the power we want, we need to think about other ways of solving problems since the current technology can only go so far.
Ya, Intel really f---ed up the industry with the P4 architecture. Hopefully they wont do the same with the 64-bit cpus.
Im hoping for a cross of Transmetas low wattage hardware with AMDs simple architecture.
08-16-2002, 12:20 AM
one thing AMD needs to use is the heat spreader like the P4. I believe they are doing that with the Hammer chips. No clue why they didn't do that with the Athlons.
08-16-2002, 06:00 AM
Enigma, I hate like hell to be the "other" side of the story, but save yourself a big freakin' headache and aircool. Should you decide different, let me know, I'll get you what you need to start for nothing, I machine my own parts. BTW, I guarantee my copper waterblock to perform as good, or better than what you may buy online, and that's partly due to a COPPER heater core I use for inline cooling, use nothing else.
08-16-2002, 06:18 AM
Rando - I've been 50 / 50 about watercooling for a long while now and always stayed with aircooling. And upgraded heatsink and fan every six months or so seemed to do the trick!
I'd be very interested to see your blocks if thats possible and your WC system.
Thanks for all the interesting comments and thoughts on this subject guys. It's a subject that I think some manufacturers have forgotten about, for now at least.
08-20-2002, 12:36 AM
my hopes are in super conductivity, almost resistanceless computers, there will be some major speed, and little resistance=little heat, it might be a simplistic idea of a big problem, but it would be nice, sinc. lurch
08-21-2002, 09:08 PM
Actually lurch your right on there. There is research under way to create superconductors at room temprature, along with many other lines of research into different approaches to takling the issue of increasing computer power. Brute force just won't do it this time.
08-21-2002, 10:24 PM
yeah the more i think about the origonal statement the more it irks me, i mean hardware can not be the only solution to increased computing power, another thing that will have to be done is to increase the efficiency of the software design, it might mean making less broad range computers to more job specific computers which can drop processes and work more efficiently, all in all it can't be just on the hardware end, someone may have to rethinkg the whole system of software, which might mean losing tons of old compatibility, but then again compatibility has been a problem ever since the 286 and the 8086, slowing down and restricting the new just to compensate for the old, when do you say enough is enough, ??? lurch
08-22-2002, 12:08 AM
and its puzzling why the hell the mobo manufacturers aren't taking advantage of the thermal diodes built into the XP to give us the most accurate CPU temps like the P4's. cmon the xp's have been out for over a half year now and barely a handful of mobo makers offer it.
08-29-2002, 09:15 AM
a little OT here but do you know where I could find the lists of mobo's that support it?
Powered by vBulletin® Version 4.2.0 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.