That's an interesting project. I briefly tried watercooling one of my 3200+ machines using a modern all in one cooling solution, and the results are promising - 35-36C in idle, 41 tops under load. I used an Abit AN7 maiboard witch allows installing the cpu block / pump by use of a bolt-trough kit, although the hole spacing is like no other socket - maybe a bit like early intel 423 p4 sockets. To attach the water block I made new holes in one of the mounting plates that come with it, and used a sheet of aluminum to make a backplate for it.

There are two things to worry about: mounting pressure and making sure the cpu block is not mounted at an angle, because there is a high chance of cracking the die. I used some adhesive feet - the type you can get at most hardware stores - witch I cut out to make sort of a protective ring around the cpu die (like the metal ones used for radeon gpu dies), making sure the material is ~1 mm above the die. The material is paper/textile like and is a little compressible under pressure, so tightening the four mounting bolts makes sure that the two surfaces are in contact with each-other, while at the same time making it easy for the cpu block to stay level so it won't crush one of the cpu die's corners.

I'm planning of building such a machine myself, when I can get a good deal on an all in one water cooler, since the one I used for the experiment is the same one I need for my daily driver PC.

kanecvr wrote:

That's an interesting project. I briefly tried watercooling one of my 3200+ machines using a modern all in one cooling solution, and the results are promising - 35-36C in idle, 41 tops under load. I used an Abit AN7 maiboard witch allows installing the cpu block / pump by use of a bolt-trough kit, although the hole spacing is like no other socket - maybe a bit like early intel 423 p4 sockets. To attach the water block I made new holes in one of the mounting plates that come with it, and used a sheet of aluminum to make a backplate for it.

There are two things to worry about: mounting pressure and making sure the cpu block is not mounted at an angle, because there is a high chance of cracking the die. I used some adhesive feet - the type you can get at most hardware stores - witch I cut out to make sort of a protective ring around the cpu die (like the metal ones used for radeon gpu dies), making sure the material is ~1 mm above the die. The material is paper/textile like and is a little compressible under pressure, so tightening the four mounting bolts makes sure that the two surfaces are in contact with each-other, while at the same time making it easy for the cpu block to stay level so it won't crush one of the cpu die's corners.

I'm planning of building such a machine myself, when I can get a good deal on an all in one water cooler, since the one I used for the experiment is the same one I need for my daily driver PC.

I'm having terrible flashbacks just thinking about uneven heatsink pressure cracking dies.

I'm so glad that all desktop CPUs have integrated heat spreaders now. Its scary enough trying not to crack a semi-rare 13 year old CPU die that goes for $40 on eBay. Mounting a brand new $350 i7 without an IHS would be torture...

I wonder if its possible to take the lid off of a K6-2, use some high end thermal compound (like Conductonaut) and use it on an Athlon XP. They actually look very similar without a heat spreader.

kanecvr wrote:

That's an interesting project. I briefly tried watercooling one of my 3200+ machines using a modern all in one cooling solution, and the results are promising - 35-36C in idle, 41 tops under load. I used an Abit AN7 maiboard witch allows installing the cpu block / pump by use of a bolt-trough kit, although the hole spacing is like no other socket - maybe a bit like early intel 423 p4 sockets. To attach the water block I made new holes in one of the mounting plates that come with it, and used a sheet of aluminum to make a backplate for it.

There are two things to worry about: mounting pressure and making sure the cpu block is not mounted at an angle, because there is a high chance of cracking the die. I used some adhesive feet - the type you can get at most hardware stores - witch I cut out to make sort of a protective ring around the cpu die (like the metal ones used for radeon gpu dies), making sure the material is ~1 mm above the die. The material is paper/textile like and is a little compressible under pressure, so tightening the four mounting bolts makes sure that the two surfaces are in contact with each-other, while at the same time making it easy for the cpu block to stay level so it won't crush one of the cpu die's corners.

Thats why shims were invented back then, I still got one lying around. My A7N8X-deluxe board has a 4 way mounting system so I can mount it using 4 screws, so aslong I tighten the 4 screws evently without to much pressure and/or add some springs to the mount there shouldn't be a problem.
Anyway thats not something I am concerned about truthfully, the main issue to me is the condition of the pump (bearings and gaskets should be in good condition) and the blocks (my least worry aslong its not fully oxidized or calcified)

Ozzuneoj wrote:

I'm planning of building such a machine myself, when I can get a good deal on an all in one water cooler, since the one I used f […]
Show full quote

I'm planning of building such a machine myself, when I can get a good deal on an all in one water cooler, since the one I used for the experiment is the same one I need for my daily driver PC.

I'm having terrible flashbacks just thinking about uneven heatsink pressure cracking dies.

I'm so glad that all desktop CPUs have integrated heat spreaders now. Its scary enough trying not to crack a semi-rare 13 year old CPU die that goes for $40 on eBay. Mounting a brand new $350 i7 without an IHS would be torture...

I wonder if its possible to take the lid off of a K6-2, use some high end thermal compound (like Conductonaut) and use it on an Athlon XP. They actually look very similar without a heat spreader.

derbauer has a CPU delider tool which he sells and linus did a video about this recently with a 7960K 😲 i believe, it didn't seem to hard but ruining another cpu for that is questionable.

Richo wrote:

Ozzuneoj wrote:

I'm planning of building such a machine myself, when I can get a good deal on an all in one water cooler, since the one I used f […]
Show full quote

I'm planning of building such a machine myself, when I can get a good deal on an all in one water cooler, since the one I used for the experiment is the same one I need for my daily driver PC.

I'm having terrible flashbacks just thinking about uneven heatsink pressure cracking dies.

I'm so glad that all desktop CPUs have integrated heat spreaders now. Its scary enough trying not to crack a semi-rare 13 year old CPU die that goes for $40 on eBay. Mounting a brand new $350 i7 without an IHS would be torture...

I wonder if its possible to take the lid off of a K6-2, use some high end thermal compound (like Conductonaut) and use it on an Athlon XP. They actually look very similar without a heat spreader.

derbauer has a CPU delider tool which he sells and linus did a video about this recently with a 7960K 😲 i believe, it didn't seem to hard but ruining another cpu for that is questionable.

Usually with modern CPUs delidding is done to replace the TIM and get rid of any gaps between the die and IHS (which is reapplied). Only really crazy people leave the IHS off after delidding.

and abit more back on topic 😀
My chieftec dragon is limiting my block choice, I only have 1 120 mm fan port, the rest are 92/100? mm or 80 mm. Doubt I can find those nor would I want them since they don't make high static pressure and still silent fans for those 😲
now I was running the numbers on the heat output I would need to cool approx 80W (Barton 3200+), 60W (9800 XT) and I gues 20W (NForce2 chipset northbridge) which is around 140W total at maximum heat output. my old AIO was rated for 140W with a 120x120x30 radiator, but would that be sufficient for full custom loop?
another issue is my 9800XT, since I would include it in my loop my ram will be to be cooled passive, would this be sufficient without a dedicated fan?

I like this idea. A good thicker 120mm rad in push pull should handle what you have without loud fans.

Also to note, performance PCs has a lot of older WC'ing stuff on there site. The have 80 and 92mm rads and good 80/92mm fans too.
If I were you I'd go for a 120mm and a 92 or 80mm rad. even if you can't find good fans the extra rad space will help alot.

Right now I have about 1200w being cooled by 2 120mm and 1 240mm rads with 800rpm and 500rpm fans in my c2q setup.

Jade Falcon wrote:

I like this idea. A good thicker 120mm rad in push pull should handle what you have without loud fans. […]
Show full quote

I like this idea. A good thicker 120mm rad in push pull should handle what you have without loud fans.

Also to note, performance PCs has a lot of older WC'ing stuff on there site. The have 80 and 92mm rads and good 80/92mm fans too.
If I were you I'd go for a 120mm and a 92 or 80mm rad. even if you can't find good fans the extra rad space will help alot.

Right now I have about 1200w being cooled by 2 120mm and 1 240mm rads with 800rpm and 500rpm fans in my c2q setup.

guess a single 120 mm with 2 fans attached to them should be plenty then for a maximum of 150ish Watts 😎 , can always add a 80 mm fan later on and knowing me I will take the loop apart atleast 3 or 4 times before I am happy with it 🤣
Performance PC's is located in the USA, here in the Netherlands we still have Highflow and another company which seems to have NOS from alot of stuff. when I get everything the first thing I will probably be busy with is making some support brackets for my GPU/cpu/chipset and general mounting of the loop since for the first time I will be using flexible tubing.