How much of a damage can ripping the CPU out of socket do?

[gal-m]

26 minutes ago, Hymenopus_Coronatus said:

heat the CPU up with something like Prime95 then slowly twist without pulling. That's what I did and it worked fine

Yup. Exactly what I mentioned before.

[Guest]

33 minutes ago, gal-m said:

While this suggestion helps @ShrimpBrime it also depends on the age of the thermal paste (and of course CPU).

I help a lot of my friends out with computer related maintenance and am in most cases dealing with rather older computers and of course older AMD CPU's - we're talking at least 8 years old on average. Thus, I have personally found that twisting an old and cold CPU doesn't help that much and will result in forceful separation, but heating the CPU up AND also twisting sure does.

In fact, I have a funny story here. One of my friends ripped her CPU out of the socket together with the cooler, much like the issue @Neo5Politan is dealing with and she actually decided to pry the CPU from the heatsink with a screwdriver, because of how glued on it was. The story does in fact have a happy ending as I was able to straighten using the "pen method".

Yeah most of my CPUs are lid-less or become lid-less after I acquire them. 

I have a few Cold plates I use for various cooling methods also.

The plates become rather "stuck" to the cooling block or TEC.

I use the "Vice Method" to remove IHS plates and cold plates off water blocks and pots.

 

But I can see running a CPU at load for a bit can help the situation. But most cases, TIM doesn't "dry out" for many years. But I suppose this depends on the type of paste and the environment it's been in.  

For modern 1331 pin AMD processors, I actually use my library card. It's a little thinner and fits between the rows much better. 

 

Good advice though! warm and twist!

[Neo5Politan]

You guys are all sweethearts, I didn't expect to start such a discussion here. Thanks to you all ^^ Although I've been into PC building for quite awhile, still I haven't heard of that "pen method" or of "reseating". Probably because I'm not from english-speaking country and such terms are new for me

[LogicalDrm]

-> Moved to Troubleshooting

[Amias]

On 2/5/2021 at 1:51 PM, LAwLz said:

Yes but that's not what we are talking about in this thread. If you fuck up the heat-spreader on the CPU it might make it run hotter and throttle, but that's not something that happens from ripping it out of the socket like what OP is asking about.

 

On 2/5/2021 at 10:35 AM, Neo5Politan said:

For example, a game like Battlefront II (2017) was taking from 5% to 15% of CPU usage before the accident during intense moments, but now it takes it all with whopping 95% at the start, making comfortable gaming

impossible. 

 

https://www.pcmag.com/encyclopedia/term/thermal-throttling#:~:text=Adjusting the clock speed of,IS FOR PERSONAL USE ONLY.

 

Quote

Adjusting the clock speed of the CPU based on the amount of heat it is currently generating. Thermal throttling helps cool the chip when it gets too hot by lowering the speed. See CPU throttling.

 

On 2/5/2021 at 1:51 PM, LAwLz said:

1) I have never seen or heard that happening so I would not be so sure saying it does act that way.

2) That would not be what OP is talking about. Losing a memory channel or some IO does not cause the CPU usage reported by task manager to go up during the same workload.

 

On 2/5/2021 at 10:35 AM, Neo5Politan said:

Moreover, it can now only read half of RAM, making 7,95Gb from 16 (Kingston HyperX 2666Mhz 2x8Gb). What should I do? Is there a solution to this? 

 

 

 

"I've seen instances like that where a memory channel can go bad."

 

"bent pins in a modern cpu socket, can manifest in any number of ways, failure to post, inability to detect memory inserted into slots that one is very common"

 

 

[Mark Kaine]

On 2/5/2021 at 2:48 PM, gal-m said:

Agreed. CPU's either work (calculate), or don't. It's that simple. 

Except that's incorrect... 

[gal-m]

32 minutes ago, Mark Kaine said:

Except that's incorrect

I was trying to be very general. That being said, I am by no means an expert, just an enthusiast. Basically what I was trying to say is that components like a CPU die when they die. They usually don't show signs of incoming failure, like mechanical components do (HDD's for example).

So feel free to elaborate, I'd personally love to hear what you've got to say @Mark Kaine.

[Neo5Politan]

23 hours ago, LogicalDrm said:

-> Moved to Troubleshooting

Thanks a lot, I'm new to this forum so... It feels nice to be softly transferred to the correct section, rather than being scorned for that or even have the whole topic deleted

[gal-m]

1 hour ago, SGT-AMD said:

I have seen signs of trouble before a component like a cpu fails

I have never seen, heard or experienced a CPU gradually failing.

[gal-m]

2 minutes ago, SGT-AMD said:

stand by my statement "Heat Kills Electronics". It can do it quick, or very slow

Agreed, but when there's a failure with a component like a CPU, your computer basically won't turn on anymore.

But yes, heat does kill electronics. No doubt about that.

[PwnyTheTiger]

3 minutes ago, gal-m said:

Agreed, but when there's a failure with a component like a CPU, your computer basically won't turn on anymore.

But yes, heat does kill electronics. No doubt about that.

You can damage the CPU or it's pins without it completely failing outright, which will then affect other functions of said CPU like PCIe and memory channels. So there are situations where the component can be broken but still turn on.

[LAwLz]

13 hours ago, Amias said:

Quote

Yes but that's not what we are talking about in this thread. If you fuck up the heat-spreader on the CPU it might make it run hotter and throttle, but that's not something that happens from ripping it out of the socket like what OP is asking about.

Quote

For example, a game like Battlefront II (2017) was taking from 5% to 15% of CPU usage before the accident during intense moments, but now it takes it all with whopping 95% at the start, making comfortable gaming

impossible. 

https://www.pcmag.com/encyclopedia/term/thermal-throttling#:~:text=Adjusting the clock speed of,IS FOR PERSONAL USE ONLY.

 

Quote

Adjusting the clock speed of the CPU based on the amount of heat it is currently generating. Thermal throttling helps cool the chip when it gets too hot by lowering the speed. See CPU throttling.

 

I know what thermal throttling is. I am not an idiot.

What I am saying is that damaging a CPU pin will not cause your CPU to thermally throttle. Did you even read my post or do you just jump in and want to be like "ackchyually!".

 

 

Yes, if you fuck up your heatspreader then it is possible that you will get worse heat dissipation, causing your CPU to throttle. Nobody has said that can't happen. However, that is not related to the thread. 

OP said he could not see any damage to the CPU or socket itself. Ergo, the heatspreader was not affected and the temperatures should be the same.

 

 

13 hours ago, Amias said:

Quote

1) I have never seen or heard that happening so I would not be so sure saying it does act that way.

2) That would not be what OP is talking about. Losing a memory channel or some IO does not cause the CPU usage reported by task manager to go up during the same workload.

 

Quote

Moreover, it can now only read half of RAM, making 7,95Gb from 16 (Kingston HyperX 2666Mhz 2x8Gb). What should I do? Is there a solution to this? 

 

<videos about bent CPU pins where they might as well be guessing because they haven't tested their statements that damaged pins can behave that way>

Again, did you read my post? Losing a memory channel would not cause his CPU usage to go from 5% to 95%.

[gal-m]

38 minutes ago, PwnyTheTiger said:

So there are situations where the component can be broken but still turn on

You're right @PwnyTheTiger! I have heard of missing pins but still functioning CPU.

To be honest I'm not sure anymore to which comment I was replying but I think the general idea was: you won't notice your CPU GRADUALLY become worse. When it will die, it will die, kind of like a fuse.

 

Physical damage is a different story, for sure. 

[Guest]

1 hour ago, gal-m said:

You're right @PwnyTheTiger! I have heard of missing pins but still functioning CPU.

To be honest I'm not sure anymore to which comment I was replying but I think the general idea was: you won't notice your CPU GRADUALLY become worse. When it will die, it will die, kind of like a fuse.

 

Physical damage is a different story, for sure. 

Most people don't beat on a processor to experience a sudden steep declination of a dying cpu.

 

It starts off OC is good playing game.... and then the screen freezes and .1s sound loop.

OK restart ignore, happens again, much faster.

OK lower the OC a little bit, plays a while, again.

Run it stock... happens again.

Underclock it..... 

Finally until a no post.

Was exactly how I would describe my Opteron 165's death. 

I want to say that was back in 2010.

Have experienced this with a few Phenom and FX chips as well. Very similar deaths.

Guess it's something that kind of happens after time despite the cooling used.

And sometimes a cpu will just boink out and done. *shrugs*

[Amias]

1 hour ago, LAwLz said:

I know what thermal throttling is. I am not an idiot.

What I am saying is that damaging a CPU pin will not cause your CPU to thermally throttle. Did you even read my post or do you just jump in and want to be like "ackc

Yes, if you fuck up your heatspreader then it is possible that you will get worse heat dissipation, causing your CPU to throttle. Nobody has said that can't happen. However, that is not related to the thread. 

OP said he could not see any damage to the CPU or socket itself. Ergo, the heatspreader was not affected and the temperatures should be the same.

 

 

Again, did you read my post? Losing a memory channel would not cause his CPU usage to go from 5% to 95%.

 

Lack of visual clues does not rule out damage to the heat sink. Op could have wreaked a bracket or not seated it correctly. This is a perfectly reasonable explanation for a CPUs sudden drop in performance.

 

Secondly loss of a memory channel has been proven a common problem with CPUs/Sockets. OP has lost a memory channel so that is pertinent to the problem. You claimed it wasn't.

 

Now I don't know CPU architecture that well, not it's not a stretch to assume loss of half your CPUs memory channel will impact performance. Even more so, whatever has damaged the memory channel could very well hinder the CPUs performance going forward.

 

All of this is important and viable information to help OP.

[kabil]

On 2/5/2021 at 4:22 PM, Neo5Politan said:

Oops, forgot to mention that... I was trying to change thermal compound and it was too late, when I realized that the CPU got stuck to the radiator. I just pulled a bit harder and a second later I found myself with the radiator and the CPU stuck to it in my hand.

Are you sayin that ur CPU is kinda glued to the radiator and it came off with it?

[gal-m]

7 hours ago, ShrimpBrime said:

Guess it's something that kind of happens after time despite the cooling used

Yep @ShrimpBrime. And like I mentioned before degradation of silicon is present with pretty much every CPU and it leads to instability, especially when overclocked.

I'd assume that most normal chips which haven't been OC'd won't have these issues, but you never know

[Neo5Politan]

9 hours ago, kabil said:

Are you sayin that ur CPU is kinda glued to the radiator and it came off with it?

It was from a certain point of view. I put less thermal compound than usual (for some reason) at the time before the accident so a few days later I surely felt the need to replace it with a reasonable amount. All the mass surely dried out and it unfortunately served as a great glue replacement.