• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Possible Bending in 12700F. Any Cure? Opinion Please.

Joined
Feb 24, 2013
Messages
276 (0.06/day)
Location
Assam (India)
System Name myLastOne
Processor Intel Core i7 12700F @ PL1=PL2=150 Watts
Motherboard MSI B660M MAG Mortar
Cooling Deepcool AK400 + CM Sickleflow 120mm (Push|Pull)
Memory 2x 16GB Crucial Ballistix 3600MHz 16-18-18-38
Video Card(s) Asus TUF RTX 3060 Ti Gaming OC
Storage TeamGroup MP33 Pro M2 1TB + (1+2)TB Seagate + 2TB WD
Display(s) BenQ EW2775ZH
Case CM Elite 371 USB3 With Arctic F14 (Side+Front) + Arctic Bionix F120 (Back) + CM 120mm (Top)
Audio Device(s) Realtek ALC 1200 + Swans D1010-IV
Power Supply MSI MPG A750GF
Mouse Logitech G102 LightSync
Keyboard Logitech G100
VR HMD None
Software Windows 11 PRO
Dear, Techpowerup forumers,

I have owned my new system for two years now. It consists of an i7 12700F on an MSI B650M Mortar motherboard with a cooler, Deepcool AK400 with another CM SickleFlow fan in push-pull config.

My PL1=PL2 are both set at 150W. My CPU temps. used to be really low. Even with Prime95 max heat preset it would be under 74°C; even if I turn my room into a hot box all fans off, ventilators,windows and doors closed at noon in the harshest summer day possible with stress test for 50mins! My boost used to be really good at ~ 4.9GHz (Now just 3.8GHz). But of late my temperature by 12-15°C for same PL1,2 settings and no bios update. I suspect a bent cpu (?) Here are pictures after disassembly:
CPU:
cpu1.jpg


Heatsink base:
hs1.jpg


What do you think? No, I do not have any socket mods installed, CPU looks ok, flat. But today after watching this very short video, I am not sure, think the left side of the longest side where the Intel etching starts maybe is a bit bent (not sure). Dont have any special equipment. Used a clear plastic scale just seems ok and not sure at the same time.

IF I assume my CPU is BENT, but just a little bit. Can it be saved/repaired diy. Say I use the Thermalright CPU bracket, now, would it straighten the bent overtime? Any other way?

Thank you and Please, help!
 
TIM is dried out, probably just needs a thorough cleaning and new application.
 
First, it is not "bent". Bent would suggest damage - even a crease in the metal. It may be slightly curved in a concave (inward) or convex (outward) manner, but not bent.

Second, this would not happen years after use - unless the CPU was removed and subject to some extreme physical abuse. So if you have a concave or convex IHS, it came that way from the factory.

Third, you could "lap" the surface but frankly, the very purpose for TIM (thermal interface material) is to fill tiny imperfections in the mating surfaces of the IHS and heatsink to displace and prevent any insulating air from getting in there. So I agree with Arcutcas and you should thoroughly clean the mating surfaces and re-apply a fresh new layer of quality TIM.

BTW - you do not need to replace TIM just because it is x number of years old, even if it has dried out. The ONLY reason TIM comes in a liquified format is so we can squeeze it out of the tube and spread it around evenly. If it dries, the solid materials remain behind and are still doing their jobs.

TIM can easily last 10, 15 years or even longer AS LONG AS the cured bond is not broken. So the only time TIM needs to be replaced is when the cured bond is broken, allowing air in. And since TIM should NEVER be reused, any time you remove the cooler, the surfaces need to be thoroughly cleaned and a new application applied.

Note the most common reasons for a broken bond is abuse (bouncing the computer off the floor or rough handling during transport) or, ironically, by the user twisting the cooler too hard to see if loose, breaking it loose in the process. :rolleyes:

Having said all that, unless you broke the bond somehow, I suspect something else is happening here. While it is true the efficiency of TIM decreases over time, it typically is only by 3 - 5°C. Not 12 - 15°C. And so a fresh new layer of TIM will typically only improve cooling by that 3 - 5°C.

And to the point, if you "need" those 3 - 5°C to prevent crossing over thermal protection thresholds, you are already too close, clearly indicating another problem. Case cooling is the most common cause - assuming default clocks and voltages.

Are all case fans spinning properly? CPU fan? Do you have good flow of cool air through the case? Is the case interior, vents and filters clean of heat trapping dust? Do you have good cable management?
 
Doesn't the socket tend to flex on LGA1700? That's why there is a contact frame for the socket. Maybe one of those will help prove better cooler contact to the IHS.

Sorry for the US Amazon link, I see you are in India.
 
If the chip itself doesn't look bent, then bending issue did not occur. Buy a BCF anyway - it's $10 or so and peace of mind is worth it. The LGA1700 (and presumably, 1851 as well) ILM is dogshit. I use Thermalright's - Thermal Grizzly's also works, buy whatever you can find cheaper (or faster locally).

Once the CPU is physically damaged, there is no going back.
 
I've seen no credible suggestion that the default LGA 1700 retention mechanism causes direct physical damage to the CPU. It can warp the IHS enough to reduce cold plate contact, thus raising temperatures, but that isn't quite the same as bending the CPU. In the OP's case, I would assume something else went wrong; maybe your heatsink was slightly loose to begin with and your dog bumped into the machine. Maybe a fan is on the fritz. Maybe your fan curve got screwed up somehow, or your power limit settings changed.

There's no harm in grabbing a Thermalright contact frame now that you have to reseat/repaste the CPU anyway. They're certainly affordable, adding one probably will help a bit with cooling. But aftermarket contact frames aren't really supposed to address some sort of critical physical threat to the health of your CPU; they're designed for people looking to eke out every last bit of thermal headroom. IIRC, derbaur himself (owner of ThermalGrizzly, which sells contact frames) said that most people shouldn't bother.

I'd guess that if you put the machine back together correctly, double check all the fans, double check your BIOS settings, and clean any intake filters, you'll see temps get back in line.
 
Get a heatsink with a baseplate
 
TIM is dried out, probably just needs a thorough cleaning and new application.
Might try only after cpu bracket arrives. Really afraid now. If there is very little bending will the bracket fix it? Any other way? Please read my reply to Mr. Bill_Bright for more details. Thank you.
First, it is not "bent". Bent would suggest damage - even a crease in the metal. It may be slightly curved in a concave (inward) or convex (outward) manner, but not bent.
There is no physical damage, but I just have a clear plastic scale/ruler.
I included a tomshardware linked video, just 22 secs (please look at it). The proper way is to put the cpu back into the socket of the mother and then use an illuminated source in the back and on the front a straight opaque object is used to see if there is deformity and light passes through. I was not able to take off the motherboard. But tested with just the cpu put onto a straight place, using phone torch and using a thick card that came with my TUF graphics card. Seems like little arch is the on the extreme left side of the cpu (concave, not tested on the socket, reminding!). Not that much in the video, but less than half, very little. If I do this (both) centre to centre and both diagonals no light comes out. Really confuse??? Its normal? Like you said. I also tested on the extreme right of the cpu (long side) this time no light comes out. Normal? Also I had to use opaque card assumed straight. I do not have calibrated metal scale at the moment. What do you think?

Second, this would not happen years after use - unless the CPU was removed and subject to some extreme physical abuse.
No, never! I never rough used. I treat my system like a newborn baby. Extremely aware of these things.This is the first time I have removed this cpu because no other software solutions were working.

Second, this would not happen years after use - unless the CPU was removed and subject to some extreme physical abuse. So if you have a concave or convex IHS, it came that way from the factory.
My did. I always monitor my cpu and other components temps regularly. Because, previously I had a Haswell locked quad core i5. This is my most expensive setup ever, I am not getting another. I went from 4 cores to 12 cores hdd to nvme ssd. I was really paranoid of damage. Chcked regularly only in August of previous after checking for 1 year, across every season did I stop.
Secondly, you forget one thing 12th gen had/has this bending/deforming issue (esp.(less) i7,(more) i9). Thermalright bracket was invented 3-4 months after I assembled my system (?) Atleast not available at all after invention for 2 months. I also did not pay much attention because my DC AK 400 cooler was lightweight enough.

Having said all that, unless you broke the bond somehow, I suspect something else is happening here. While it is true the efficiency of TIM decreases over time, it typically is only by 3 - 5°C. Not 12 - 15°C. And so a fresh new layer of TIM will typically only improve cooling by that 3 - 5°C.

And to the point, if you "need" those 3 - 5°C to prevent crossing over thermal protection thresholds, you are already too close, clearly indicating another problem. Case cooling is the most common cause - assuming default clocks and voltages.

Are all case fans spinning properly? CPU fan? Do you have good flow of cool air through the case? Is the case interior, vents and filters clean of heat trapping dust? Do you have good cable management?
Firstly: Yes, drastic change in temp. Also, 1GHz less boost. These are drastic changes. TIM used was Kingpin Kpx (already 2 yrs old when applied. Also, please notice there's hardly any tim on the hs base. Never happened before removing cpu.
Secondly: Yes case is clean all fans working properly tested really well continuously for a years. What do you suggest, now?
These are really drastic changes, i have not even put the cpu back out of fear. Its been 8 days now.
If I get the cpu bracket now will it cure any bending (assumed there is bending, yes), reduce temp overtime any other way? Or no hope left. Please, reply. Thank you for replying thoroughly, you always do. Thanks again.

Doesn't the socket tend to flex on LGA1700? That's why there is a contact frame for the socket. Maybe one of those will help prove better cooler contact to the IHS.

Sorry for the US Amazon link, I see you are in India.
It wasn't invented when I assembled my PC. Please, do not apologise.
If there is very little bending will the bracket fix it? Any other way? Do you know of diy repair. Please read my reply to Mr. Bill_Bright for more details. Thank you.
If the chip itself doesn't look bent, then bending issue did not occur. Buy a BCF anyway - it's $10 or so and peace of mind is worth it. The LGA1700 (and presumably, 1851 as well) ILM is dogshit. I use Thermalright's - Thermal Grizzly's also works, buy whatever you can find cheaper (or faster locally).

Once the CPU is physically damaged, there is no going back.
Please read all my replies to Mr. Bill_Bright for more details and please reply. Is there no way to un-bend or mitigate for just a little, very little bend on one side maybe (cannot be sure now, unless calibrated scale arrives). No other way, none, to save the cpu? no physical damage cracks or anything to the pcb or anything! Thank you.
I've seen no credible suggestion that the default LGA 1700 retention mechanism causes direct physical damage to the CPU. It can warp the IHS enough to reduce cold plate contact, thus raising temperatures, but that isn't quite the same as bending the CPU. In the OP's case, I would assume something else went wrong; maybe your heatsink was slightly loose to begin with and your dog bumped into the machine. Maybe a fan is on the fritz. Maybe your fan curve got screwed up somehow, or your power limit settings changed.

There's no harm in grabbing a Thermalright contact frame now that you have to reseat/repaste the CPU anyway. They're certainly affordable, adding one probably will help a bit with cooling. But aftermarket contact frames aren't really supposed to address some sort of critical physical threat to the health of your CPU; they're designed for people looking to eke out every last bit of thermal headroom. IIRC, derbaur himself (owner of ThermalGrizzly, which sells contact frames) said that most people shouldn't bother.

I'd guess that if you put the machine back together correctly, double check all the fans, double check your BIOS settings, and clean any intake filters, you'll see temps get back in line.
Please, read my replies to Mr. Bill_Bright to get all answers and more info. Please do not mind. Thank you for thorough reply.
Is there no way to un-bend or mitigate for just a little, very little bend on one side maybe (cannot be sure now, unless calibrated scale arrives). No other way, none? no physical damage cracks or anything to the pcb or anything! Thank you, again. Please reply, if any solution found.

Get a heatsink with a baseplate
Mr. eidaraman, of you had read it already, then I apologise. But could you read all my replies to Mr. Bill_Bright, for more information and reply back. Please .
 
Perhaps the weight of the cooler (I presume motherboard is vertical, and cooler is horizontal) is causing the motherboard to flex slightly?

A backplate, such as the Heatkiller will alleviate it.
 
That the CPUs are bending in the LGA 1700 socket is a well known fact. You can see a lot of information in this thread:


You may want to renew the paste and then mount a cooler which has a bulge on the base, that fits better in the CPU shape. The coolers with direct contact heatpipes tend to be flat and will have a thicker layer of paste between the base and the CPU heatspreader.
 
That the CPUs are bending in the LGA 1700 socket is a well known fact. You can see a lot of information in this thread:


You may want to renew the paste and then mount a cooler which has a bulge on the base, that fits better in the CPU shape. The coolers with direct contact heatpipes tend to be flat and will have a thicker layer of paste between the base and the CPU heatspreader.
Wow and i mentioned earlier about using a heatsink with a base plate...
 
Oh you guys know me.
Tell this what I see.
This has no bending!

Issues

1. Used 2 years old shelf paste
2. application looks too thick
3. mounting pressure looks low (unsure, maybe just bad paste)
4. dirt and dust built up around the cpu
5. close up pictures to hide the dirt??
6. AK400 cooler is inadequate. 3.8ghz would be all I'd expect

Cures

1. New paste
2. apply several times practice different applications of paste
3. double check mounting pressure for a week after first mount.
4. Periodic cleaning will help prevent the build up of earth bacteria
5. pictures say 1000 words. more the merrier
6. Larger cooler to dissipate 253w of power OR sell it and get a 12400F instead. (this would put money in your pocket)

:p
 
Oh you guys know me.
Tell this what I see.
This has no bending!

Issues

1. Used 2 years old shelf paste
2. application looks too thick
3. mounting pressure looks low (unsure, maybe just bad paste)
4. dirt and dust built up around the cpu
5. close up pictures to hide the dirt??
6. AK400 cooler is inadequate. 3.8ghz would be all I'd expect

Cures

1. New paste
2. apply several times practice different applications of paste
3. double check mounting pressure for a week after first mount.
4. Periodic cleaning will help prevent the build up of earth bacteria
5. pictures say 1000 words. more the merrier
6. Larger cooler to dissipate 253w of power OR sell it and get a 12400F instead. (this would put money in your pocket)

:p
My sig rig I had initially warped, i backed off some and this was before first boot ever, no warp no problems.
 
My sig rig I had initially warped, i backed off some and this was before first boot ever, no warp no problems.
What warped?
My GPUs warp without a stand. Never seen an issue from it.
Motherboard, I've seen the CPU pull away from the contacts (LGA = no post)
IHS plate, had to install into a vice and hit it with 32oz hammer to bend it. :)
 
First, it is not "bent". Bent would suggest damage - even a crease in the metal. It may be slightly curved in a concave (inward) or convex (outward) manner, but not bent.

Second, this would not happen years after use - unless the CPU was removed and subject to some extreme physical abuse. So if you have a concave or convex IHS, it came that way from the factory.
Seems, you may be right afterall (?) The slightest of slight bend is the on the upper side of cpu ihs (the shorter side of the rectangle on one side.) Yes concave but very slight, you cannot feel it but see it in certain conditions. Never paid any attention (or measured), never been an issue since 12th gen. Can manufacturing defect cause this? Arent cpu manufacturing by default is "perfect" bleeding edge cannot be copied top-notch?
1. Used 2 years old shelf paste
2. application looks too thick
3. mounting pressure looks low (unsure, maybe just bad paste)
4. dirt and dust built up around the cpu
5. close up pictures to hide the dirt??
6. AK400 cooler is inadequate. 3.8ghz would be all I'd expect
1. Its Kingpin KPx, very well kept.
2. No, very thin, I am not the verge guy, I apply little and spread and spread and spead and spread.
3. Mounting pressure perfect. Bad paste, possible.
4. I am from a dusty and very humid place.
5. Close up pic to show cpu detail (cpu + paste). Also, very bad picture taker.
6. Techpowerup, did test AK400, very good (esp on intel). Also, Pl1=Pl2=150w. And it worked before. ...And its an i7 12700"F" (Max, 180w). Tested from pl1=pl2=95w, until I was comfortable with perf and temp. Why would I lie? Or are you insinuating that I did? Please, I would never. Dont do that please.

Oh you guys know me.
Tell this what I see.
This has no bending!
You might be right (?) I became depressed even after the brackets came I didn't reinstall. Did not use PC, all this time. Only day before yesterday that I did. Tested p95 with hw info on 2021 oct version Windows 11 and latest. Temp. at 63C (how much it will increase during summer dont know). But temp is not the issue now, just boost! Stuck at 3.8ghz! If I remember correctly I did a bios update prev year november, I think. That might have messed up, maybe? Cannot say. Now when temp fixed, tested on different os versions and different p95 versions used, culprit seems the new bios, wouldnt you say so too?
I could go back but ups acting up dont know whats up, batteries are 7mnths old only. Repair possiple only next month. Besides I am not going out.
What do you suggest I do (concerning the cpu boost issue)?
I tried putting pl1=pl2=180w but no change in temp or boost. Earlier in 2021 when I tested it it reached 87C in 3mins boost 4.89ghz. I stopped the test. Now no change.
But when I reset bios boost increases but only to 4.1ghz and temp. actually drops 1 degrees. Can the bios update cause manual setting pl1 and pl2 values to become redundant? Is that even possible? But why? What purpose does it serve?
Do you have any idea?

@ShrimpBrime Would you like to see all settings of my uefi bios?

Sorry for wasting everyones time for 2 months!! Only re-assembled day before yesterday.
 
Every heatsink is unique while every model has a fairly specific profile it ground to. At least you are chasing the enthusiast route to solving this with an open mind.

Past hardware concerns you have to accommodate what a mess Intel processor behaviors are with mobo and chipset firmware as wildcards + OS as a multiplier. What your latest description sounds like is a need to reconsider the newly shuffled deck.
 
.........What your latest description sounds like is a need to reconsider the newly shuffled deck.
Sorry, English is my 4th language. I am slow, sorry. If I get it right, you are saying here is I have deviated (or changed completely) from my original post and we/ (you?) will have to look at my problem from a new
perspective, right? With also a little sour tone, right? Sorry if I got it wrong! I just want to revive my cpu. I was may unclear on some posts. Please, say. Should I create a new thread?
 
New cooler with a strong, robust baseplate and the Thermalright/Thermal Grizzly reinforcement frame should be all that's needed.

I don't think any modern CPU is good for the old direct-contact heatpipe coolers any more; Ignoring the poor contact between pipes and poor heat distribution to the outer pipes, the current-gen LGA sockets have too much mounting pressure for heatpipes IMO. They aren't that strong and deformation is definitely possible.
 
New cooler with a strong, robust baseplate and the Thermalright/Thermal Grizzly reinforcement frame should be all that's needed.

I don't think any modern CPU is good for the old direct-contact heatpipe coolers any more; Ignoring the poor contact between pipes and poor heat distribution to the outer pipes, the current-gen LGA sockets have too much mounting pressure for heatpipes IMO. They aren't that strong and deformation is definitely possible.
Thanks but thats solved now (for the time being, will report back in summer, also installed bracket). Please read, post 15 (3rd paragraph). Its now about the boost now.
 
Seems, you may be right afterall (?) The slightest of slight bend is the on the upper side of cpu ihs (the shorter side of the rectangle on one side.) Yes concave but very slight, you cannot feel it but see it in certain conditions. Never paid any attention (or measured), never been an issue since 12th gen. Can manufacturing defect cause this? Arent cpu manufacturing by default is "perfect" bleeding edge cannot be copied top-notch?

1. Its Kingpin KPx, very well kept.
2. No, very thin, I am not the verge guy, I apply little and spread and spread and spead and spread.
3. Mounting pressure perfect. Bad paste, possible.
4. I am from a dusty and very humid place.
5. Close up pic to show cpu detail (cpu + paste). Also, very bad picture taker.
6. Techpowerup, did test AK400, very good (esp on intel). Also, Pl1=Pl2=150w. And it worked before. ...And its an i7 12700"F" (Max, 180w). Tested from pl1=pl2=95w, until I was comfortable with perf and temp. Why would I lie? Or are you insinuating that I did? Please, I would never. Dont do that please.


You might be right (?) I became depressed even after the brackets came I didn't reinstall. Did not use PC, all this time. Only day before yesterday that I did. Tested p95 with hw info on 2021 oct version Windows 11 and latest. Temp. at 63C (how much it will increase during summer dont know). But temp is not the issue now, just boost! Stuck at 3.8ghz! If I remember correctly I did a bios update prev year november, I think. That might have messed up, maybe? Cannot say. Now when temp fixed, tested on different os versions and different p95 versions used, culprit seems the new bios, wouldnt you say so too?
I could go back but ups acting up dont know whats up, batteries are 7mnths old only. Repair possiple only next month. Besides I am not going out.
What do you suggest I do (concerning the cpu boost issue)?
I tried putting pl1=pl2=180w but no change in temp or boost. Earlier in 2021 when I tested it it reached 87C in 3mins boost 4.89ghz. I stopped the test. Now no change.
But when I reset bios boost increases but only to 4.1ghz and temp. actually drops 1 degrees. Can the bios update cause manual setting pl1 and pl2 values to become redundant? Is that even possible? But why? What purpose does it serve?
Do you have any idea?

@ShrimpBrime Would you like to see all settings of my uefi bios?

Sorry for wasting everyones time for 2 months!! Only re-assembled day before yesterday.
I recommend to clear cmos. Do the testing at defaults, make sure it's boosting and then set up your XMP.

There is no reason to under-volt or limit power settings with coolers that are believed adequate.
 
I recommend to clear cmos. Do the testing at defaults, make sure it's boosting and then set up your XMP.

There is no reason to under-volt or limit power settings with coolers that are believed adequate.
Not even turn un XMP? XMP is off at default. Plz reply
 
Not even turn un XMP? XMP is off at default. Plz reply
He's talking about testing the system. Start with XMP off after a BIOS reset and test to see if boost problem is corrected.
 
Not even turn un XMP? XMP is off at default. Plz reply
Yes, clear cmos, test defaults without xmp.

XMP enables and increases voltages like system agent for example which increases cpu thermals.
 
Sorry, English is my 4th language. I am slow, sorry.

No sourness was intended. There is also no need for a new thread.
 
...There is no reason to under-volt or limit power settings with coolers that are believed adequate.
There was a reason for me. I came from i5 4590 so performance was enough for me. And tests have shown that even pl1=pl2=125w, easily gets you 90%+ of the perf, pl1=pl2=150w was a bonus for me! Maybe buy a better cooler after 6 months (I never did). I did not under volt. Just pl1=pl2=150w and enabled XMP.

Heres, the test result 6mins test small fft prime95:
Untitled.png

No sourness was intended. There is also no need for a new thread.
Sorry.
 
Back
Top