Raptor Lake’s Heat Problem

[Michael-EGV]

I'm planning to build a new PC using an Intel 13'th Generation Raptor Lake proccesser.
I plan on using a normal processer when they become available ( as opposed to the ' K ', the over-clocker version ) .
And so, I've been following posts online in various places about temperature concerns.
The posts that concern me most are those about High Idle Temperatures.

Once I start the build, I certainly do plan on updating the UEFI / BIOS to the latest version.

I also plan on doing a harware modification to the motherboard itself as recommended
by Tom's Hardware, apparently as a best practice to fix the problem.
Here is the article and a [ new ] part which I plan on purchasing . . .

10-20-22 : TOM'S HARDWARE : ThermalRight LGA1700-BCF Contact Frame Review : Can it tame Raptor Lake’s heat?

$ 15 @ NewEgg for brand ThermalRight, color { Aluminum }

Given the Tom's Hardware article,
it appears to me that we have a mechanical / thermal design bug and a fix.
Let's discuss this.
Shouldn't the motherboard manufacturers be modifying their mecahnical designs to implement this fix ?

If they already are, we could simply add a few links here to refer to the motherboard
manufacturer solutions.

( And if they aren't implementing the fix, why not ? )

 

[Solaris17]

So the problem is, it seems like the majority of people, users, yourself included mix and match tribal knowledge from different CPUs and different series, and different manufacturers.

Contact Frames:

Contact frames work, but only in some cases. On some boards, and some cooler combos. They are means to correct a curvature problem that stems from a combo of, socket clam, backplate and cooler torsion. The issue came to light with 12th gen, but not every board was affected. those affected worked fine with some coolers. With the 13th gen I have found no credible widespread information regarding it as a problem, and even if I did, it would still suffer structurally from the variables mentioned above.

Heat:

That video really. We do not live in Intel 11th gen land. Or Zen 2 land. Or the Core 2 era. People do 90c to F conversion in google realize they can boil water and get scared they are going to set on fire. We are way past 60 and 70C load temps. Thermal density and physics doesnt allow this. We live in a world of element consideration and refinement when we carve and create micro architectures. We push stability to the brink of what the physical material can handle.

The 13th gen CPUs as shown above are meant to bump the limiter. They will continue to do so. This is how performance metrics are judged now. push as hard as you can for a few seconds then back off. Thats the formula. They are engineered as such.

The bigger question to ask is what you will put up with. Bigger coolers, water cooling etc serve a purpose sure. They dump heat. How hot do you want your office to get from that heat? Energy consumption is all the rage in reviews now. Much more than it was a decade ago. This is the new era. With advancements in silicon, process and architectures the temps only become an issue when pushing OCs. a stock 13th gen will happily push 90C at its target freq.

The elephant in the room is how much heat do you want to xfer, and how much energy do you want to burn? Controlling package power is not new but it is certainly getting more attention and is starting to become IMO from what it looks like, the preferred way to tame heat on the new CPU line.

Even AMD does this with Eco mode.

https://www.youtube.com/results?search_query=AMD+eco+mode

In short, no one is implementing a fix because it isn't as wide spread as you think, and silicon makers aren't implementing a fix because its the way its meant to run now. What needs to start changing is the users thinking. Much like the turbo button or FSB overclocks, 60-70c all core loads and 65w brisbanes are the way of the dinosaur. Computing has changed. and the goal posts are different now.

 

[ir_cow]

@Solaris17 I was going to share the same video. Its nice to see someone at Intel did a interview on it. My take away is whatever the warranty covers, you are good. KS 115c. Same silicon so... the takeaway here is Intel CPU can get hot.

Now what isn't talked about is how if you are at these high temps it will degrade enough that LN2 world records isn't possible anymore. But it makes no difference for air, water or chilled water setups.

Voltage kills CPUs, not thermals within the "constraints" set by Intel these days.

 

[bobbybluz]

While I did use the contact frames on my two Alder Lake builds at this time I'm not going to on my Raptor Lake project. I read (can't recall where) somebody tried installing a Thermalright frame on a Z790 mobo and they had to buy new mounting screws because the ones from the stock Intel mount were too short.

My new G.Skill DDR5 6400 Trinity Z5 (Hynix A-die) 32 x 2 will be here tomorrow so I can finally put it together.

 

[Why_Me]

I'm planning to build a new PC using an Intel 13'th Generation Raptor Lake proccesser.
I plan on using a normal processer when they become available ( as opposed to the ' K ', the over-clocker version ) .
And so, I've been following posts online in various places about temperature concerns.
The posts that concern me most are those about High Idle Temperatures.

Once I start the build, I certainly do plan on updating the UEFI / BIOS to the latest version.

I also plan on doing a harware modification to the motherboard itself as recommended
by Tom's Hardware, apparently as a best practice to fix the problem.
Here is the article and a [ new ] part which I plan on purchasing . . .

10-20-22 : TOM'S HARDWARE : ThermalRight LGA1700-BCF Contact Frame Review : Can it tame Raptor Lake’s heat?

$ 15 @ NewEgg for brand ThermalRight, color { Aluminum }

Given the Tom's Hardware article,
it appears to me that we have a mechanical / thermal design bug and a fix.
Let's discuss this.
Shouldn't the motherboard manufacturers be modifying their mecahnical designs to implement this fix ?

If they already are, we could simply add a few links here to refer to the motherboard
manufacturer solutions.

( And if they aren't implementing the fix, why not ? )

Get a locked cpu, don't get an i9 unless you do productivity work with your PC for a living, make sure you have a case with decent airflow, a decent cpu cooler and a board that has decent VRM's.

 

[ir_cow]

While I did use the contact frames on my two Alder Lake builds at this time I'm not going to on my Raptor Lake project. I read (can't recall where) somebody tried installing a Thermalright frame on a Z790 mobo and they had to buy new mounting screws because the ones from the stock Intel mount were too short.

I have two of these thermalright ones. Can't speak for all motherboards but it works fine on the ASRock Z790 Sonic up to the ASUS Z790 Apex. Only used it on 5 or so motherboards so far.

They are cheaply made, so it's possible the one that person got has a different z-height making the screws not work.

Contact frame helps with those last few degrees for me and pushing ram to the limits. Need it to go above DDR5-8200. Doesn't mean this will be the same for everyone.

 

[Michael-EGV]

I don't normally build PCs, but when I do,
I believe in using the latest CPU generation for the features they offer
( and not for how fast they can go ).
I've never done load / performance testing and have no interest doing that.

My only interest is having a decently low idle cpu temperature and
I'm still exploring what that idle temperature should be [ or could be ].

My killer application will indeed put a load on the CPU,
but only for about 60 seconds at a time.
That's it.
Most of the time, I'm idling, surfing the web or writing an email.
So, most likely, an Intel i5-13600 would suffice.

And, this is not an AMD versus Intel thing. I'm not a troll.
In addition, I think the 13'th Generation also has a ' T ' version,
the " Power-Optimized lifeStyle".
The ' T ' may be an option for many who just want the latest
features, but not the heat.

My perception in this Build-A-PC exercise is that most of the
speed / performance enhancements will actually come from
the primary storage device anyways, the NVMe devices,
( compared to legacy SATA3 devices ), and less from the CPU.

Aside from voiding the motherboard warranty,
I don't see any drawbacks ( yet ) by considering Tom's Hardware
suggestion as a best practice.
I wouldn't mind at all going to a hardware store and buying new screws.

I'm still thinking about this as a best practice and
very interested in what others have to say.

Thank you to those who have already replied.

 

[Deleted member 185158]

Hahaha!
These chips, running hot is Not a new occurrence in the PC world.
Intel has been 100c max since X58. First gen core i7 i9.

You purchase to your needs and funds availability.

Buy a low end 13400F whatever. Nobody cares.
If you put a tiny passive heat sink on it, you'll get that processors performance all the way up to 100c.

Intel engineer is like meh, I've seen 100c before. No big deal. Lol

 

[Calmmo]

While I did use the contact frames on my two Alder Lake builds at this time I'm not going to on my Raptor Lake project. I read (can't recall where) somebody tried installing a Thermalright frame on a Z790 mobo and they had to buy new mounting screws because the ones from the stock Intel mount were too short.

My new G.Skill DDR5 6400 Trinity Z5 (Hynix A-die) 32 x 2 will be here tomorrow so I can finally put it together.

I had no such problem on z790.

 

[Michael-EGV]

Once a CPU is bent, the CPU is always bent.

Here are 3 related videos.

07-26-22 :
Investigating Intel's CPU Socket Problems | Thermal Grizzly Contact Frame Benchmark

07-28-22 :
$4.35 Fix for Intel Thermal Problems | Thermalright 12th Gen Contact Frame

02-14-23 :
Thermalright LGA1700 BCF Contact Frame Plate Weird Results Please Help!
Interesting Comments Below The Video . . .
If you've already fitted the CPU with the original clamp it may have already bent the IHS.
If you over tighten the screws...others have reported Boot up and Memory issues.
RONALD : During disassembly BIOS voltage settings were reset to stock.
RONALD : You have go back & reset all your voltage settings in BIOS the way they were before.
RONALD : This happened due to motherboard detecting CPU was removed & replaced.
RONALD : It thinks it is a different CPU & set settings to guarantee booting.
RONALD : Even though processor was never completely removed electrical contact was broken.
RONALD : Given the difference in wattage your CPU was likely undervolted by .150 volts before.
RONALD : It is common that 13 gen core i5 processors can be undervolted easily by .100 volts.
I did install it on my system and it lowered my temperature around 7• but i did use the Noctua thermo paste instead.
ALEXANDER : The problem is the LGA 1700 isn't consumer ready out of the box.
ALEXANDER : You either need to be a master technician or have a miraculous stroke of luck when seating everything properly.
ALEXANDER : I suggest watching some of Yangcom Korea's videos and pay attention to how he torques screws on a LGA 1700 bend correcting frame.

 

[ir_cow]

I think people aren't doing the contact frame correctly. I followed what a MB engineer told me. He does this daily I was told. Works for me everytime. Granted I only have done this maybe like 20 times now with 2 different contact frames. Not enough to say this is the best and only method.

Put all 4 screws back in the same place. Loosely tighten in X pattern while pressing / holding the bracket evenly so one side isn't higher. Stop when screws don't turn from light pressure. Now give each screw a firm / hard twist and you have it all even.

If you tighten one down fully at a time it WIlL be uneven and probably bend some pins.

 

[Michael-EGV]

I think people aren't doing the contact frame correctly. I followed what a MB engineer told me. He does this daily I was told. Works for me everytime. Granted I only have done this maybe like 20 times now with 2 different contact frames. Not enough to say this is the best and only method.

Put all 4 screws back in the same place. Loosely tighten in X pattern while pressing / holding the bracket evenly so one side isn't higher. Stop when screws don't turn from light pressure. Now give each screw a firm / hard twist and you have it all even.

If you tighten one down fully at a time it WIlL be uneven and probably bend some pins.

" People " who don't like meticulously following written procedures +
video demonstrations should NOT be building PCs.

" People " who are not mechanically inclined should NOT be building PCs.

This is all a matter of quality { Written Procedures, Video Demonstrations }.

" People " need to write themselves written procedures for the device they purchase.

The Grizzly device's height is short, does not touch the motherboard, and
therefore is vulnerable to over-tightening
[ but only if " people " do not follow a written procedure and or video ].

The ThermalRight device's height is taller, touches the motherboard, and
probably much less vulnerable to over-tightening, if a written procedure and or
video are followed.

Science and Quality must move FORWARD (and never sideways nor backwards).

 

[ir_cow]

The problem is people are following gamersnexus videos lol

 

[Dr. Dro]

I purchased a frame for my motherboard out of pure zeal, it is quite affordable and I didn't see why not... but it's absolutely not necessary, especially with the lower TDP parts.

Having opted for the i9-13900KS, any improvement in contact and thus heat transfer is valid for me. If only those jokers deliver my CPU already... gonna have to give the importer a call and inquire about it, I ordered it on February 8(!) and it's still not here. All the parts are waiting...

 

[Calmmo]

I purchased a frame for my motherboard out of pure zeal, it is quite affordable and I didn't see why not... but it's absolutely not necessary, especially with the lower TDP parts.

Having opted for the i9-13900KS, any improvement in contact and thus heat transfer is valid for me. If only those jokers deliver my CPU already... gonna have to give the importer a call and inquire about it, I ordered it on February 8(!) and it's still not here. All the parts are waiting...

I was quite careful on my end and went with a contact frame from the start, careful to get the pressure on everything incl the cooler right so that it's neither loose nor overly tight. And so far in CB with mce on its boosting sustained 5600 all core easily at just under 350w. I do get 3 cores about 1 to two mins in Hitting 100 and throttling down to 5500 but the rest during a 15min run never got close to that. No memory problems either which is mostly what I see ppl online complaining about requiring near perfect frame+cooler mounting to avoid unnecessary problems.

Took 1,5 month to get the 13900ks here btw, guess they're really not able to keep up with demand eh?

 

[Michael-EGV]

The problem is people are following gamersnexus videos lol

"
Remove & Replace the Intel ILM, Independant Loading Mechanism.
Step 1 ) To protect the Processor's socket, insert a Processor into the socket.
Step 2 ) Remove the Intel ILM, Independant Loading Mechanism.
Step 3 ) Assemble the ThermalRight Anti-Bending Buckle.
"

Manufacturers should update their User Manuals.

If they don't like the Gamer's Nexus video, they could create their own video.

 

[Michael-EGV]

I was quite careful on my end and went with a contact frame from the start, careful to get the pressure on everything incl the cooler right so that it's neither loose nor overly tight. And so far in CB with mce on its boosting sustained 5600 all core easily at just under 350w. I do get 3 cores about 1 to two mins in Hitting 100 and throttling down to 5500 but the rest during a 15min run never got close to that. No memory problems either which is mostly what I see ppl online complaining about requiring near perfect frame+cooler mounting to avoid unnecessary problems.

Took 1,5 month to get the 13900ks here btw, guess they're really not able to keep up with demand eh?

Dr. Dro & Calmmo :

Can a ' K ' processor be tweeked to look like a ' T ' processor ?

Just lower the " Performance Core Base Frequency " ?

Michael

 

[ir_cow]

Contact frames aren not made or sold by motherboard manufactures, so why should they update the manual?

 

[bobbybluz]

When I built my first Alder Lake system in November/December of 2021 the contact frames weren't on the market yet. I initially used an Arctic Liquid Freezer II 280 mounted externally on an ASRock Z690 Steel Legend WiFi. The i7 12700K ran hot and definitely isn't isn't an overclock lottery winner. Sometimes you win, sometimes you lose.

Around six months later I built an i5 1200K system on another Z690 Steel Legend WiFi. This time the buckles were available and I bought two. The 12600K got the buckle in the initial build and I retrofitted the 12700K rig. I also went to an Arctic 420 for the 12700K while the 12600K got an Arctic 240 I already had on hand. I noticed when I pulled the pump assembly off the 12700K the thermal paste was spread perfectly. I attribute this to the massive steel backplate Arctic uses for their socket 1700 AIO's combined with the pump mounting pressure. The CPU between the backplate and pump coldplate is under higher pressure than any mounting device in the middle is likely to provide.

Also, when removing the stock LGA 1700 mounts the screws needed very little torque to remove. They weren't cranked down tightly, they were finger-tight at best. When I installed the Thermalright frames I kept that in mind when doing the final tightening of the mounting screws. No odd results afterwards and as suspected the 12600K ran much cooler than the 12700K even with a much smaller AIO and overclocked to 200MHz higher. Better silicon in the 12600K. For grins I pulled the 240 and put the 280 on it (the 240 also needed the warranty gasket kit installed so it had to come off anyway). Even with the 420 and frame the 12700K performed very close to it's previous configuration. The 12600K showed slightly less paste on the upper left hand corner of the heatspreader when I pulled the 240, other than that is was very even. Remember that was with the contact frame installed. The radiator on that rig is also externally mounted but lengthwise vertically as opposed to the 420 being mounted horizontally.

At idle both rigs run at ambient room temperature. Under load, however, the 12600K has never thermal throttled even at 5.3GHz while the 12700K does it regularly at 5.1GHz. I did order another Thermalright frame for my in-process 13600KF build but was notified last Wednesday by the Ebay vendor they didn't have any in stock and gave me a refund. I'm tired of waiting so after I finish the needed mods on a "sleeper" Raidmax case I got off the local Craigslist for free I'm finally putting it all together with another Arctic LF II 280 mounted the same way I have the other with the 12600K on a Corsair 300R that also came off Craigslist (I love cases I can get for $25 or less. No cringe factor when Dremel modding them).

In wrapping this post up it's my opinion that the cooler and backplate used MAY exert more pressure on the CPU coldplate than any contact frame or stock Intel LGA 1700 mounting system. These variables differ widely depending upon cooler manufacturers. Ideally Intel would use the same mounting system that they used on the HEDT & Xeon X79, X99 and other 2011 type CPU sockets.

 

[ir_cow]

For reference this is what I've bought twice now and recommended to others. I had a 10c drop on one MB and next to nothing on another. I think it's really just about even mounting pressure.

https://www.amazon.com/gp/product/B0B9SFG355/

 

[bobbybluz]

For reference this is what I've bought twice now and recommended to others. I had a 10c drop on one MB and next to nothing on another. I think it's really just about even mounting pressure.

https://www.amazon.com/gp/product/B0B9SFG355/

That's the same I've used. When I began considering the physics involved I came to the same conclusion that even mounting pressure on the CPU is the most important factor. When I got my first Arctic 1700 mounting kit I was surprised at the massiveness of the backplate; it's so massive and stiff it may be drop-forged. That thing isn't going to bend without putting it in a vise and hitting it with a hammer or using Vise-Grips on it. Very different from the usual plastic many manufacturers use. Torsional rigidity properly applied is a must for the best contact.

 

[Michael-EGV]

Contact frames aren not made or sold by motherboard manufactures, so why should they update the manual?

To improve quality.
To satisfy their customers.
To gain market share.

 

[ir_cow]

To improve quality.
To satisfy their customers.
To gain market share.

Can you imaged how many people would RMA stuff after damaging the MB if it was endorsed by the motherboard manufactures????

 

[Michael-EGV]

Dr. Dro & Calmmo :

Can a ' K ' processor be tweeked to look like a ' T ' processor ?

Just lower the " Performance Core Base Frequency " ?

Michael

I think I can answer may own question.
As I suspected, I do think the answer is : YES; you a tweak a ' K ' so that it idles like a ' T ' Power-Optimized lifeStyle Processor.

Some search results I found . . .
How To Underclock CPU (Downclocking Your CPU Windows)
12700k quick n' dirty undervolting/downclocking
How To Underclock CPU

So, there is no reason to wait for Intel to make ' T 's [ nor non ' K 's ] available; not that I can think of.

 

[freeagent]

Of course you can. You can go both ways.. you can have high performance, medium, and low. It depends on how well you can cool it.