A warning for AM4 socket watercooling

Smoked Brisket

Smoked Brisket

Gawd
Joined
Feb 6, 2013
Messages
786
I have been watercooling for many years on both Intel and AMD platforms. This is the third board I have installed and removed a CPU block on the AM4 platform. The specific board is the latest revision of the Gigabyte VISION D b550. I had intended a few months ago to start a build with this board and upgrade from a 3700x to a 5900x that i got on sale. I installed the chip and a heatkiller IV pro, then I got sidetracked and did not get back to it until yesterday. I Wanted to reseat the block so I took off the screws and started to pull the block. I have done this many times, it is a watercooling block so there is not much twisting you can do. I assure you I was gentle. The cpu came out stuck to the block, I was pretty stunned but was able to get the cpu off the block with no bent pins. I did another seating with brand new paste, pulled the block gently off and the cpu came out again with the secure arm down. I have since reseated the cpu, booted the board on the table and have a post, so it seems it still works. I have built a lot of pc's so I was able to stay calm and just put the cpu back in and test it. After searching on the internet it seems that this is common on AM4. Most threads end with the advice to "twist" the cooler side to side to loosen before removing the block. This does not work with water blocks. This is just a friendly warning. Has anyone else had this happen? Is it fixed for AM5?
 
I've had that happen with a lot of CPUs, pretty much anything with a PGA Socket and a heatspreader. Pentium 4 were notorious for it with the "bubblegum" phase change thermal pads that were common back in the day combined with the 423/478 HSF retention that didn't allow any twisting.
Never damaged pins that I can remember but it's terrifying every time!
 
Yeah I had it with P4, 754, 939, AM2/3/4. I use arctic silver so that tends to get a good adhesion to the cooler, and without twisting it, it almost always will pull it out of the socket and sometimes even after twisting it.
 
I almost forgot...
re: "is it fixed for AM5?"
TL;DR = Yes

the stock Loading Mechanism on modern LGA sockets like AM5 / Intel / etc mechanically press and latch the CPU into the socket. As long as some sort of socket/mobo-level loading mechanism is there the CPU will stay very much secure, with default mobo config 100% will stay in the socket.

If one does a block mount or something that requires removing the stock ILM, there's various solutions for securing the CPU & regulating pad-to-pin pressure. Still evolving with AM5, lots of new socket-level mod products and hax out there lately.

Running an LGA socket with no CPU retention at all (other than the cooler) requires a lot of caution because the CPU will definitely come out. And socket pressure may be wack and cause I/O to glitch out.
 
Last edited:
I have done this a couple of times with various Ryzen builds I have owned and this has happened to me even after "twisting" if the chip wasn't warmed up first.
Totally agree this isn't an AM4 issue but what did change is that the heat spreader on these Ryzen chips was a lot bigger than many in the past and the larger surface area makes it more "sticky" so to speak. I have never had this problem with my water-blocked AM4's, but more so with large tower coolers in a tight case where there is not much room to twist the cooler prior to trying to remove it.
 
As I stated in the OP, if you are using a waterblock with screwed in posts, like the Heatkiller IV pro, you CANNOT twist back and forth. Intel has been using a locking frame to secure the cpu for well over a decade. This is not meant to bash amd, I have been on the platform for a few years now and have added and removed blocks without issue. However, when looking at it, it is a weak retention design. I believe that possibly because the Heatkiller has a massive surface area compared to blocks like the optimus I was using before, it is creating an airtight seal and really sucking on to the cpu.
 
Like others have said, this is often an issue with PGA sockets. It was an issue loooong before AM4 came around. I remember it happened to me with Pentium 4s and Socket 462 Athlons. It's happened to me multiple times for various different generations of CPUs from both brands, as well as to a lot of my PC-building friends, over the past 20-something+ years. None of us have killed a CPU or bent pints yet, but yeah it can get your heart racing when it happens. LGA CPUs don't come out with the cooler.
 
I'd actually be surprised if I didn't find the CPU stuck to the cooler, but I still prefer PGA over LGA. LGA can be messed up by just dropping the cpu when trying to seat it.
 
I had this happen with my old socket 939 Toledo 4400+ X2 and the "classic" Thermaltake flower-shaped cooler....

CPU turned out to be ok and the motherboard was already dead sooooo - guess I got lucky

Damn, I LOVED that rig back in its day:

Athlon 64 4400+ X2
4GB RAM
ATI 3850 512mb (AGP variant)
SB Audigy 2
 
Why all the drama. Is there anything wrong with just opening the latch and pull? Alternatively how about just unthread the threaded rod and play twister?
 
Top Nurse said:
Why all the drama. Is there anything wrong with just opening the latch and pull? Alternatively how about just unthread the threaded rod and play twister?
Click to expand...
You have to remove the block to access the nuts or latch and the rods are attached to the backplate.
 
I would ditch the watercooler and drop a Peerless Assassin 120 SE on it for $36buxs.
 
You must log in or register to reply here.
Back
Top