At last, new CPU cooler

purrkur · Post by **purrkur** » Tue May 25, 2004 6:22 pm

(Post edited 3 August 2004 - became a BP6.com member and moved my pics over to by bp6.com webspace!)

Hi everybody,

As you know by now, I have been looking into changing my CPU coolers for something that will give me better cooling while being quiet at the same time. Because of the caps lying on both sides of the ZIF-socket, space is minimal. Oh, and I was also looking for a pair of cheap coolers. I can't justify buying two coolers for a price comparable to that of a new motherboard and a new Athlon CPU either

I had been looking at the Spire WhisperRock IV but I was pretty uncertain that it would fit. Here is the one I am talking about:

http://www.spire-coolers.com/fcc.asp?ProdID=105

I found them here in Sweden for a measly 119 SEK ($15.5 / £8.5) so what I did was to go and buy one to fit it in. Let me bring in some eye candy...

Here is the board with the Intel coolers in place before the operation:

As you can see, the chassi isn't all that big and I don't have any fans blowing steam out of it either, which probably the reason why my system temp is always 8-10 degrees hotter than my cpus

Mounting the cooler was quite an advanture! I think I have never used so much force to mount a cooler on a motherboard before. I was pretty scared I was destroying my BP6! Anyhow, statistics: There is space for a cooler with a dimension of 60mm between the capacitors on either side of the ZIF-Socket. That is exactly what I had on the Spire cooler! Needless to say, a few of the caps are literally touching the heatsink.

More eye candy

Here the Spire cooler is mounted on the board.

Different angle:

I set the computer up again and started running it like shown on the pics above. With the two Intel coolers I was getting a reading of 39/40/48 for cpu1/cpu2/system respectively while idle. I should also note that if I started an application that used up 1 cpu for 100% while the other was maybe doing 20%, both processors would heat up and not just the one doing 100% work. Is this what others have seen as well?

Anyhow, cpu1 was consistently showing 5-6 degrees less than cpu2 when using the Spire heatsink and if I started a program doing 100% cpu work on cpu1 then the increase in temperature was less than when I had the Intel stuff. Another very interesting development is that my system temp went down by quite a lot as well! I think that the reason for this is that the large 80x80mm fan is moving the air around the box. A fan blowing out the excess heat might make things even better.

Encouraged by what I had seen, I went and bought the second cooler today. It was slightly easier mounting this one than the first one but I was still a bit shaky doing it because of the tight fit. I modified the cooler a bit because with the way things were, the cable from the second cooler would have come out between the two coolers. Because space was so tight, I had to unscrew on of the screws holding the fan to the heat sink, thread the cable into the space found behind the screw and put it back into its place again. I think that there is maybe a millimeter or two between the two fans

And another angle:

You are probably wondering how things are looking now as I sit here writing this on my BP6. Well, I am running my two 533MHz cpus in turbo mode, giving me 544MHz, while using 1.9 Volts on em as well. Here are my sensor readings:

VCore 1: +1.89 V (min = +1.81 V, max = +1.98 V)
Vtt: +1.49 V (min = +1.81 V, max = +1.98 V)
+3.3V: +3.31 V (min = +3.14 V, max = +3.46 V)
+5V: +4.97 V (min = +4.73 V, max = +5.24 V)
+12V: +12.40 V (min = +10.82 V, max = +13.19 V)
-12V: -12.03 V (min = -13.18 V, max = -10.88 V)
-5V: -5.05 V (min = -5.25 V, max = -4.75 V)
V5SB: +5.13 V (min = +4.73 V, max = +5.24 V)
VCore2: +1.92 V (min = +2.40 V, max = +3.60 V)
fan1: 0 RPM (min = 8881 RPM, div = 2)
fan2: 0 RPM (min = 5273 RPM, div = 2)
fan3: 0 RPM (min = 3497 RPM, div = 2)
temp1: +35 C (high = -68 C, hyst = -108 C) sensor = thermistor
temp2: +34.5 C (high = +80 C, hyst = +75 C) sensor = thermistor
temp3: +42.5 C (high = +80 C, hyst = +75 C) sensor = thermistor

The best thing is that I can load the cpus hard now and the temp doesn't rise to the scary temp readings I was getting the other day when I ran it as an ET-server in a room where the temperature was maybe 25 degrees or so.

Other minor modifications I have done is to apply paste under the green heatsink and I have also lifted the sensors under the cpus high enaugh so I know that they are now touching the cpu. That is about it. What I would recommend to people attempting to mount such large heatsinks on the BP6 that are not used to doing such work is to remove the motherboard from the box in order to have easier access to the area around the ZIF-Socket. the BP6 was never meant to have these type of heatsinks which makes it even harder getting them there.

Now all I gotta do is to try and set my FSB to 75MHz

Dave Rave · Post by **Dave Rave** » Tue May 25, 2004 7:00 pm

something I always do with a new cooler is,
bend the retaining clips a fair bit so that you don't need so much force to attach the cooler to the cpu socket.

nothing like dispair after pulling the socket off of the mboard.

purrkur · Post by **purrkur** » Wed May 26, 2004 2:47 am

Dave: I didn't describe everything that I had to do, especially to get the first cooler mounted, but it did involve bending the clip. I must admit though, although I have had adventures mounting heatsinks on motherboards, I have never had the pleasure of pulling the socket off the motherboard

davd_bob · Post by **davd_bob** » Thu May 27, 2004 1:14 am

if the cooling is good enouth why not try 72fsb.
unlikely but maybe even 75fsb. wouldnt it be cool to run dual 600s.

purrkur · Post by **purrkur** » Sat May 29, 2004 4:01 pm

update:

I tried overclocking to 75MHz fsb with the voltage at 2.0. The machine started booting but when it was almost done it hung hard. Not much to do so I reduced it down to 72 MHz fsb. It has been running stable at that for almost 48 hours now. Strangely enaugh, I am almost positive that the temperatures on the cpus is lower now than when I was running at 533MHz + turbo =544 MHz. I have been thinking about increasing the voltage to 2.05 or 2.1 volts to see if I can get it stable at 600MHz....

Remember, this is an unmodified BP6 built in 1998. The EC10 cap is the small one as well. The board is very stable and everything is just humming along...

BCN · Post by **BCN** » Sat May 29, 2004 5:22 pm

good luck! the board should give you those 75MHz if the CPUs support it and put something on the greenie

purrkur · Post by **purrkur** » Sun May 30, 2004 5:12 am

Another update

Yesterday I bumped the board up to 75MHz fsb. I upped the voltage on the cpu's to 2.05 and thats it! My cpu's have been running at 600MHz since yesterday without so much as a hickup!

Code: Select all

[davidt@loki]:~$ more /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 6
model name      : Celeron (Mendocino)
stepping        : 5
cpu MHz         : 600.225
cache size      : 128 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 2
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr
bogomips        : 1183.74

processor       : 1
vendor_id       : GenuineIntel
cpu family      : 6
model           : 6
model name      : Celeron (Mendocino)
stepping        : 5
cpu MHz         : 600.225
cache size      : 128 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 2
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr
bogomips        : 1196.03

And here are the figures for heat and voltage at idle and room temperature of 24 degC...

Code: Select all

VCore 1:   +2.05 V  (min =  +1.94 V, max =  +2.14 V)
Vtt:       +1.47 V  (min =  +1.94 V, max =  +2.14 V)
+3.3V:     +3.31 V  (min =  +3.14 V, max =  +3.46 V)
+5V:       +5.03 V  (min =  +4.73 V, max =  +5.24 V)
+12V:     +12.10 V  (min = +10.82 V, max = +13.19 V)
-12V:     -11.78 V  (min = -13.18 V, max = -10.88 V)
-5V:       -5.05 V  (min =  -5.25 V, max =  -4.75 V)
V5SB:      +5.11 V  (min =  +4.73 V, max =  +5.24 V)
VCore2:    +2.06 V  (min =  +2.40 V, max =  +3.60 V)
fan1:        0 RPM  (min = 8881 RPM, div = 2)
fan2:        0 RPM  (min = 5273 RPM, div = 2)
fan3:        0 RPM  (min = 3497 RPM, div = 2)
temp1:       +36 C  (high =   +60 C, hyst =   +20 C)   sensor = thermistor
temp2:     +35.5 C  (high =   +80 C, hyst =   +75 C)   sensor = thermistor
temp3:     +42.5 C  (high =   +80 C, hyst =   +75 C)   sensor = thermistor
vid:      +2.050 V

Two 533MHz Celerons running at 600MHz in an unmodified BP6 from Abit! I have only been doing "normal" work on it though. No stress testing so far. However, I feel I should update my sig again

hyperspace · Post by **hyperspace** » Sun May 30, 2004 8:17 am

I'll try OC'ing my 533's to 75 when I get home, tonight. It will be interesting to see how the Golden Orb's handle the increase compared to your new Spire coolers.

hyperspace · Post by **hyperspace** » Sun May 30, 2004 10:03 pm

@ 75Mhz, it would lock up after opening two or three applications. Had VCORE's at 2.0 volts. CPU temperatures were 39° C, loaded. Will try again when I get better cooling. Want to try 2.05 volts. Ambient temperature was 20.5° C.

purrkur · Post by **purrkur** » Mon May 31, 2004 1:00 am

Hyperspace: The only thing I did was raise the voltage to 2.05 and it worked without problems. I only tried to boot the machine once at 2.0 volts per cpu and it failed so I didn't try again (if it fails once then it is bound to fail again..).

At the moment the room is really warm where the computers are because the door and window were both shut. The room termperature is now 27 degrees C and my cpus are reporting 39 and 40 degrees C, which is idle. When the room temperature is at 22-23 degC then my cpu's idle at around 34-35 degC.

I would say that it is pretty good for nearly silent cooling! How are your Orbs when it comes to noise?

hyperspace · Post by **hyperspace** » Mon May 31, 2004 8:10 am

purrkur wrote:I would say that it is pretty good for nearly silent cooling! How are your Orbs when it comes to noise?

My computers are in our bedroom (apartment). The Orbs are not noisy enough to keep me from sleeping.

But, I would like to find some fans that run with less noise.

davd_bob · Post by **davd_bob** » Wed Jun 02, 2004 12:46 am

purrkur,
i have a theory id like tested, you too hyperspace if neither of you mind. if you get stable at or near 600mhz note the date and after running for a week try dropping the voltage to 1.9v.
it is my personal belief that mendocino core cpus will run at 1.8-1.95 in their effective range no-matter what the speed(oc or underclocking) also IF it is necessary to increase over 2.0 then the chip has hit its limit and will NOT be stable under any serious loading. Im am wondering though if kicking up to 2.1 volts for a while will somehow 'burnin' the chip at that clock speed allowing the voltage to later be dropped to 1.9 or lower?
what do ya think?

BCN · Post by **BCN** » Wed Jun 02, 2004 8:28 am

I do not believe there is such a burnin (there would have to be an explanation for that, right?)...
there may be a burnout though....

hehehehe

davd_bob · Post by **davd_bob** » Thu Jun 03, 2004 12:58 am

burn in=married male that is so trained by wife that she no longer has to remind him to take out the trash

burn out=collage student with a bp6(ie holodeck2 maybe)

burn-in as to cpus...i don't know. maybe if the electrons flow over the perscribed transistor paths for a time they will lower the resistance on those paths... just wishful thinking.

purrkur · Post by **purrkur** » Thu Jun 03, 2004 2:25 am

Davd_bob: I gotta go with BCN on this one. What happens to a transistor inside the cpu under high current situations is that electrons might "leak" through in places where they are not supposed to. This is probably the biggest risk factor for overclockers, both when they increase voltage and when they increase the speed.

So an update on my system: It has been running at 600MHz now since I last wrote and it has been very stable and nice to work on! The sun shines into my office room the better part of the day so the room temperature can get pretty high. Yesterday when the room temperature was at 27 degrees, the cpu's were running at 40/41 degrees respectively under a mild load.

I haven't done any continuous-load stress testing but I have done work on it which has seen high intermittent load without any issues. I can also say that while I like overclocking, I would give it up in a second if it was either unstable or I felt it being harmful to my system. There is a difference between running at 533MHz and 600MHz but it is not that great that I will sacrifice stability or my system either.

I can keep you updated on my progress and testing at this speed if you want. I haven't had much time to do stress tests but one of these days I might just turn off everything and run a game of Enemy Territory for half an hour or so to see what happens

Some figures at this point in time...

From my Linux Logo (at login):

Code: Select all

Linux Version 2.6.5
Compiled #2 SMP Wed Apr 28 20:35:13 CEST 2004
Two 600MHz Intel Celeron Processors, 512M RAM
2379.77 Bogomips Total

Uptime:

Code: Select all

[davidt@loki]:~$ uptime
 08:04:02 up 3 days, 16:07,  3 users,  load average: 0.18, 0.11, 0.08

Heat (from sensors):

Code: Select all

temp1:       +39 C  
temp2:     +38.0 C 
temp3:     +46.0 C

I am at work at the moment so I don't know what the room temperatures are but I would guess somewhere at 25+C since we close the room (computer noise

) in the evenings.

jaybird · Post by **jaybird** » Tue Jun 08, 2004 12:57 pm

Just thought I'd add my 2 cents worth.

I did a "burn-in" with my 2 366's and it does work as described.

They now will run at 605 Mhz (yes, and air-cooled!) @ 2.0 volts!

But, you MUST run active cooling on the BX, I can't stress this enough, and even better is a "mini-Peltier" for $4.99 US as sold on EBay.

Again, I may just have "lucked-out" with my BP6 but I am more than willing to share everything, settings, hardware, etc. so that others may also benifit.

Regards,

jaybird

davd_bob · Post by **davd_bob** » Wed Jun 09, 2004 12:39 am

purrkur and BNC
HA...I told ya so...

see jaybird is willing to think OUTSIDE the box like me. maybe he thinks he is smart as i think i am
(just kidding)

um jaybird, are you saying you had to over-volt the cpus to get them to run at 605 then later you were able to drop to 2.0v?

purrkur · Post by **purrkur** » Wed Jun 09, 2004 1:58 am

Good stuff! It is always fun to hear of other peoples experiences! I guess an update on my setup is in order....

After running stable at 600MHz for about 3 days (without any stress testing) the BP6 finally locked up hard while I was working on it. At that time I simply decided to drop the FSB back to 72MHz because with the work I do on the BP6 I can't have it lock up on me. However, I am still willing to experiment

After dropping the old girl down to 72MHz, I started a game of Enemy Territory and played for about 30 minutes. The system was rock solid at that speed and didn't cause me any troubles. So that is what I am running now (2x576MHz) without any worries at all.

Jaybird: The difference between your BP6 and mine is that you got active cooling on the BX. From what I understand, you are saying that removing the active cooling and your system won't run at 605MHz? That would put us in the same boat wouldn't it?

Planned: As soon as I find the time I intend to open up the system again and do some work. I want to add a fan to the BX, an 80mm fan to the box to draw out the heat and an extra harddrive to test the HPT controller. Once I do that I might try 75MHz FSB/600MHz CPU speed again!

davd_bob · Post by **davd_bob** » Wed Jun 09, 2004 3:00 am

for cooling the BX, get a video chipset hsf. some come with more then one spacing setting where the clips hold it to the board. the first one i got ohly had one setting for the gap and i had to modify it to replace the greenie. it looks cool, dark blue and circular but the modification looks nasty.
my BX sure is happy though and runs within 2 degrees of the cpus or even cooler sometimes.
Of course, i did add hs-grease.

purrkur · Post by **purrkur** » Wed Jun 09, 2004 3:07 am

Thanks davd_bob. I gotta check my parts bin since I think I may have a small fan lying around somewhere, although I am not sure if it can be used. I already have the hs-grease in place though. That was something I fixed before I fired up the board for the first time!

Char · Post by **Char** » Wed Jun 09, 2004 2:33 pm

Oh man.. you answered my biggest quest.. I've been looking for two weeks now for some modern coolers that might fit.. have visited a zillion websites...

THANKS!

Char · Post by **Char** » Wed Jun 09, 2004 3:18 pm

found and ordered two of those babies.. there really affordable

purrkur · Post by **purrkur** » Wed Jun 09, 2004 4:11 pm

Char: Just be careful when you mount them on our BP6. Remove the memory, and start with the one farthest from the memory sockets. So far they have been working very well for me.

Char · Post by **Char** » Thu Jun 10, 2004 3:20 pm

Thanks for you tip, much appreciated.
I'll take the board out the case anyway, and reserved half a day.. no rushing.

How did you align the clips? Large clip part (the hinge) on the outsides?

My case is a tower, so the board will be vertical. I'm thinking of supporting the heatsinks, I expect them to be heavy- I don't know if the board was designed for that.
Did you do something like that or do you use a desktop?

purrkur · Post by **purrkur** » Thu Jun 10, 2004 3:54 pm

letseehere... You can only mount the unit in one way in the box. What I did (without removing the board from the chassi) was to mount the left side first and then I used a screwdriver to force it down the other side. On on of the heatsinks I actually bent the mounting part a bit because I was having problems getting the right side to catch on. It turned out fine in the end. It was easier doing the right hand cooler, probably because I had more space. You will be just fine if you remove the board from the chassi.

Oh, and don't worry about the weight. They look heavier than they really are (I got a tower as well). Once they are sitting there they really sit tight so vertical mounting is no problem. The only thing you have to do is mount the clips onto the socket and you're set.

Let us know how it goes...

BP6.Com

At last, new CPU cooler

At last, new CPU cooler

High there!