CPU release

[hunterjwizzard]

This is a Socket 462 Pentium IV:



In the past, every time I have taken this HSF off it has ripped the CPU out of the socket. I really do not want today to be the day I ruined my favorite Pentium IV, but I'm mothballing the machine and need to get the HSF off.

As you can see, it has a lovely little plastic shroud ALL around the socket. Does anyone have any suggestions for how I can remove the HSF without wrecking the CPU?

 

[tech58761]

Are you able to unscrew the fan from the heatsink? That way you're not trying to move so much when handling the heatsink assembly?

 

[hunterjwizzard]

I could remove it but that's not going to help much. The issue is the HSF is stuck to the CPU by the grease. I need some way to detach the two so the CPU doesn't get ripped out of the socket.

 

[Uniballer]

Turn it on and let it heat up. Then turn it off and try to remove it. No warranty expressed or implied, but it might work.

 

[Eudimorphodon]

The heating strategy is probably worth a try, but I think it's going to depend a lot on the exact grease you used. Some of that stuff stays "gooey" forever, that's the kind I would expect heating to help with. If it's the stuff that dries out hard then I would suspect the heat's not going to do much. If you can remove the metal clip and apply just a *teeny* bit of friction to "wiggle" it (in a really, really gentle rotary motion instead of pulling) that might crack the bond between the chip and the heatsink more effectively than pulling.

In all fairness they never really designed these things to be put together more than once.

 

[Zare]

It is a socket 478 board, not 462.
Now that you know, you will be able to remove the CPU.

 

[hunterjwizzard]

In all fairness they never really designed these things to be put together more than once.

That does explain a lot. Although the very next iteration, socket 775, had the lovely little CPU arrester. I'm still staggered it took AMD so long to invent the same.

Turn it on and let it heat up.

Sadly turning it on is a no-go. The reason this system is being mothballs is because there is some play on ATX power connector, I need to explore whether or not I can remove and re-solder that(or replace it). Otherwise I would probably just sore the damned thing with the HSF in place.

It is a socket 478 board, not 462.
Now that you know, you will be able to remove the CPU.

The sarcasm has not gone unnoticed.

 

[tech58761]

My line of thinking when I suggested taking off the fan was to create a better area to hold onto or work around while trying to release the heatsink from the CPU. Hope you find a solution.

 

[GiGaBiTe]

Remove the metal clips and rotate the heatsink from side to side, it will break free pretty quick, unless you used thermal adhesive.

You haven't seen a stuck CPU until you've seen an OEM Dell with the grey factory heatsink compound that turns into concrete.

 

[Zare]

If it's stuck stuck, maybe the last resort is to cut off the joints of the black frame from the backside of the board.

 

[hunterjwizzard]

I'm gonna try the twisting today. It shouldn't be too badly stuck. I used the standard white heat sink grease and its only been assembled for a few years.

 

[Quagmire]

I'm gonna try the twisting today. It shouldn't be too badly stuck. I used the standard white heat sink grease and its only been assembled for a few years.

You may want to give a safety razor blade a try. With a lot of patience it worked for me once. I used a set of forceps clamped tightly to the blade and took my time 'poking' at it.

 

[krebizfan]

Might be simpler to leave the CPU on the motherboard and track down another 478 board for any other matching CPU to be tested. 478 boards seem to still be comparatively affordable.

 

[hunterjwizzard]

Success!

 

[hunterjwizzard]

You may want to give a safety razor blade a try. With a lot of patience it worked for me once. I used a set of forceps clamped tightly to the blade and took my time 'poking' at it.

There's not a way to work a razor in on this one. The plastic shroud blocks all access to the underside.

Might be simpler to leave the CPU on the motherboard and track down another 478 board for any other matching CPU to be tested. 478 boards seem to still be comparatively affordable.

I'm tempted to buy one. The CPU here is a 3.2ghz, so far as I know the fastest single-core CPU ever made. The board has had some damage, I broke one of the RAM clips and as I mentioned there is some play on the power connector. Probably a new board would be easier.

What I WANT is actually an Intel 845GV board. If you've never seen one, its an industrial board. Socket 478, AGP, but with 3 ISA slots. Unfortunately those are still about $200 and not likely to come down.

 

[Quagmire]

There's not a way to work a razor in on this one. The plastic shroud blocks all access to the underside.


I'm tempted to buy one. The CPU here is a 3.2ghz, so far as I know the fastest single-core CPU ever made. The board has had some damage, I broke one of the RAM clips and as I mentioned there is some play on the power connector. Probably a new board would be easier.

What I WANT is actually an Intel 845GV board. If you've never seen one, its an industrial board. Socket 478, AGP, but with 3 ISA slots. Unfortunately those are still about $200 and not likely to come down.

If all else fails, just be very careful and just 'lift' the whole thing up and out of the socket. Done that one too and the pins were okay. Got lucky.

 

[GiGaBiTe]

I'm tempted to buy one. The CPU here is a 3.2ghz, so far as I know the fastest single-core CPU ever made.

It is not. Socket 478 went up to 3.4 GHz. There are also two extremely rare steppings of Prescott Pentium 4s for PGA478 that supported EM64T (x86_64). These were only ever made for OEMs. Two examples are the SL7QB (3.2 GHz) and SL7Q8 (3.4 GHz). I've only ever seen one myself. It's rather strange to see a Socket 478 motherboard running a 64 bit OS.

As for raw performance, they're not the fastest. Intel's last single core CPU was the Celeron G465, and AMD one upped them with the Sempron 150. These were the final and fastest single core CPUs in terms of performance. The Celeron G465 would probably be a slightly better experience because it has Hyperthreading.

One more honorable mention is the Sempron 130, which was a soft locked Sargas dual core that could often be unlocked. I bought one of these CPUs back in the day and unlocked it to a dual core with no issues, and it chugged along for nearly a decade. Still have the CPU, but the motherboard got blown up in a lightning strike. CPU still works strangely enough.

 

[hunterjwizzard]

It is not. Socket 478 went up to 3.4 GHz.

Interesting! Thank you for sharing, I'm really interested in the end of the single core era.

I liked the idea of getting windows 98 up on this fast single core CPU. Pretty much once you hit the HT/multi core world yeah the performance gets better but thats the end of stuff like 98. Windows 2000 can boot on later socket 775 boards.

One more honorable mention is the Sempron 130, which was a soft locked Sargas dual core that could often be unlocked. I bought one of these CPUs back in the day and unlocked it to a dual core with no issues, and it chugged along for nearly a decade. Still have the CPU, but the motherboard got blown up in a lightning strike. CPU still works strangely enough.

Thats quite impressive for an AMD.

 

[Zare]

I liked the idea of getting windows 98 up on this fast single core CPU. Pretty much once you hit the HT/multi core world yeah the performance gets better but thats the end of stuff like 98. Windows 2000 can boot on later socket 775 boards.

And you probably witnessed that performance of Windows 98 comes due to architecture of Windows 98 and not lack of MHz in CPU.
I like running lesser OSes on fast hardware, like Windows 95 on Pentium MMX or P2, but you're not getting much out of Win98 on P4 unless you have ISA in equation.

 

[hunterjwizzard]

And you probably witnessed that performance of Windows 98 comes due to architecture of Windows 98 and not lack of MHz in CPU.

Yes, that's pretty much what happened. Didn't make the experiment any less fun.

but you're not getting much out of Win98 on P4 unless you have ISA in equation.

So the specific system I built out of this board had (one of) the fastest CPUs that could natively boot '98, one of the fastest AGP cards, DDR RAM, and fast storage. Its more power than 98 could handle, but that was the point. It wasn't about actual performance. It was about having some dumb fun.

but I never could get that system to be stable.