News Intel LGA1851 socket has a new ILM that helps CPUs run a bit cooler — MSI claims 1ºC to 2ºC lower CPU temperatures

Status
Not open for further replies.
"since Intel's Arrow Lake processors use a multi-chiplet design, their bending could be devastating and kill these CPUs, so Intel had to alter the socket design"

As though Intel cares about that.
The greater the failure rate, the greater the sales.
 
>integrated lever mechanism -
ILM
OK, the little socket locker widget thingy.

>the new ILM offers better compatibility with liquid coolers
liquid coolers, OMG

You mean, the chip runs so cold I can chill a beer on it, that would be awesome.
 
The RL-ILM is basically just a washer mod so if Intel really had a problem with the ILM they'd use one of the superior designs they used to have or design something entirely new. I imagine that this mechanism was largely driven by motherboard manufacturers who would rather people didn't remove the ILM.

I doubt this will make any sort of substantial difference in operation on any of the boards it'll be installed on. The most likely place this would make a big difference would be on lower end boards that don't have as many layers/reinforced backplate. It will of course also make a difference when using a cooler that has a flatter coldplate though I imagine most people in that situation would have zero issues installing a contact frame.
 
  • Like
Reactions: KyaraM
Yeap, 1-2c. Major flaw, lol
You see, you adding your imagination again, nobody said it’s more than 1-2c average, from beginning, when igor invented washer mod, it’s all about the hot core hotspot throttling when load ramp up being unsuppressible, and it is something out of oridinary without any benefit, AKA, flawed.

Fun fact is all your claims about memory issues comes from ppl cannot tighten up the contact frames evenly and properly or over tighten one corner excessively, a lot of ppl successfully get same or better memory OC after using one. And basically none have issues using washer mod instead of the CF, which is what now Intel does. If the washer mod is only able to reduce the temp by 1-2C, IHS comes back to historical minor bend, and no adverse effects on OC, isn’t that a flaw fixed? MSI even showed the old ILM with stock base plate (part of ILM) under old design bends itself, so the old part, premium or cheap board all bends
The RL-ILM is basically just a washer mod so if Intel really had a problem with the ILM they'd use one of the superior designs they used to have or design something entirely new. I imagine that this mechanism was largely driven by motherboard manufacturers who would rather people didn't remove the ILM.

I doubt this will make any sort of substantial difference in operation on any of the boards it'll be installed on. The most likely place this would make a big difference would be on lower end boards that don't have as many layers/reinforced backplate. It will of course also make a difference when using a cooler that has a flatter coldplate though I imagine most people in that situation would have zero issues installing a contact frame.
Could be partly, but there are small differences here and there, more than the "simply a washer mod", like "Perhaps more importantly, the new ILM also adopts a flat load plate (compared to a slightly curved load plate in the case of LGA1700 POR-ILM)", and when I first learned about the 1700 bending (wasn't suspecting that sort of silly issue for Intel back then, I upgraded directly from Sandy Bridge to Alder Lake), what I was aware was it takes a lot more force to clamp it down compared to Sandy bridge. But on the various videos I looked, someone filmed the base plate, which is a 1mm thick metal visiabaly bends, if that bends, the extra 5 layers of PCB won't help, and even with a undeformable backplate on, the clamping force focused on the 2 fins on the IHS would be even higher as now the base plate didn't absorb part of the force applied, so the IHS and CPU PCB would defore more, not less.

Into why they basically slightly modified the ILM vs re-do the design, it's an easy to make commercial choice.

1) The old LGA1700 ILM, though flawed, isn't something majorly impacting performance, momentarily throttling one core during sub second peak is more of a user comfort issue and say, maybe reduce things like R23 scores by 100 or so, so it isn't an major issue, but they can't roll out this back then, coz if they do, they will have to recall and replace all boards out there and the CPUs with extra deformation yet works.

2) As shown in the MSI slides, the RL-ILM did changed 3 (hinge frame insulator, load fram insulator and the load plate) out of 10 component (including the will be removed cover), so it's essentially re-designed or modified 30%, as for why not re-speccing all parts, it's simple, cost. both re-tooling and re-inventing the wheel, when during the 2.5 years+ of 1700 smart guys like Igor have did the R&D for them and in internal testing, it works, showing a constant improvement in temperature, less bending and no adverse effect on the CPU performance, why not just borrow the idea and add an extra improvement on new generation roll out? just don't use the washer from stores around the corner, make some proper parts to act the same and flattens the load frame and bam, weeks of R&D saved, temperature improved and ppl are happy.
 
The usual anti intel spill that flies against reality.

Intel = Can casually cool over 320w with entry level cheap as hell air coolers and can cool over 360w with high end coolers

AMD = Can barely cool over 250w with the highest most expensive huge ass radiators.

Internet's conclusion, Intel's cooling solution has a major flaw cause with a custom modded ILM you can drop temperatures by 1C. And then people wonder why im defending intel. Cause of this crap I keep reading on the internet by amd fanatics.
 
  • Like
Reactions: bolweval
The usual anti intel spill that flies against reality.

Intel = Can casually cool over 320w with entry level cheap as hell air coolers and can cool over 360w with high end coolers

AMD = Can barely cool over 250w with the highest most expensive huge ass radiators.

Internet's conclusion, Intel's cooling solution has a major flaw cause with a custom modded ILM you can drop temperatures by 1C. And then people wonder why im defending intel. Cause of this crap I keep reading on the internet by amd fanatics.
I don't know if you're being disingenuous, don't know how current amd cpu works or just hate amd altogether. You do know it sets 95c as its target temperature and boosts the cpu as much as possible while maintaining that temp right? You can't even remove that temp limit in the bios. Also cooling an Intel cpu while it uses cheap air coolers? Yeah good luck with that. You'd probably thermal throttle before even reaching those power usage unless of course you increase your thermal limit and be comfortable with your CPU reaching 100+ degrees. Unless of course you break physics and somehow cool a cpu using 320w using an air cooler (probably rated at 200w or below) below 100c.
 
Last edited:
You do know it sets 95c as its target temperature and boosts the cpu as much as possible while maintaining that temp right?
That's just some excuse amd came up with. Every single CPU in the history of CPUs works exactly as zen 5 does. They keep boosting clocks until they hit a temperature, a voltage or a power limit. Zen 5 is just that bad at handling heat that it reaches the temperature limit first. If it was true that "it kept boosting until it hits 95c" that means that if I put it in the freezer it would boost to 8ghz and still hit 95c, which is just not true.

My numbers came from TPUs cooler reviews.

https://www.techpowerup.com/review/id-cooling-dx-360-max-aio-liquid-cpu-cooler/7.html

Maximum heat capacity, 360w on Intel, 266w on AMD. Yet Intel's solution is considered "a major flaw". But im the hater here. Can you explain to me how a solution that is almost 40% better at dissipating heat than the competition is considered a major flaw? Im all ears. Really.
 
Maximum heat capacity, 360w on Intel, 266w on AMD. Yet Intel's solution is considered "a major flaw". But im the hater here. Can you explain to me how a solution that is almost 40% better at dissipating heat than the competition is considered a major flaw? Im all ears. Really.
Major flaw? Where did I say it's a major flaw? If anything I admire how intel managed to make it so that it doesn't burn itself(not including the ILM and volatge issues of course). Also AMD IHS design was made so it would be compatible with am4 coolers . Even then they managed to achieve better performance compared to intel on what objectively is an inferior IHS design. You don't need 360w design when you can beat the competition at 266w. Look how badly Intel's performance scales with power consumption, it's all diminishing returns in the end. You can have the best cooling solution but if the cpu itself is terrible then it's all moot in the end.
 
I think we need to see updated coolers or mounting brackets that target the new hotspot to see this 1C difference, especially for air coolers.
 
That's just some excuse amd came up with. Every single CPU in the history of CPUs works exactly as zen 5 does. They keep boosting clocks until they hit a temperature, a voltage or a power limit. Zen 5 is just that bad at handling heat that it reaches the temperature limit first. If it was true that "it kept boosting until it hits 95c" that means that if I put it in the freezer it would boost to 8ghz and still hit 95c, which is just not true.

My numbers came from TPUs cooler reviews.

https://www.techpowerup.com/review/id-cooling-dx-360-max-aio-liquid-cpu-cooler/7.html

Maximum heat capacity, 360w on Intel, 266w on AMD. Yet Intel's solution is considered "a major flaw". But im the hater here. Can you explain to me how a solution that is almost 40% better at dissipating heat than the competition is considered a major flaw? Im all ears. Really.
Firstly, Turbo boost is old, but not since CPU in history, historically, they used too boost to a certain guaranteed frequency and not way up like recent ones, hence why back in the days ppl can achieve 1Ghz boost clock

Secondly, as emphasised a dozen times yet you refuse to remember/see, AMD AM5 was being said to have too thick of an IHS so limiting thermal performance for the sake of compatibility with AM4 coolers, which got it's fair share of complaints and subsequently, those modifications of high performance IHS, sanding down the IHS, direct die etc. crop out, just that it was a delibrate choice to keep compatability of coolers and cannot be changed before AM6, they arn't immune to criticism.

Thirdly, in the TPU installation page, be aware that they use the Contact Frames for AM5 also, which suggest they use it also for the Intel platform. So... (I will skip the implication for this one)

And finally, it seems like only you, thus far have said ppl said it is a major flaw, it is a design flaw, albeit not major ones, that creates a problem (1-2C average difference, peaks of single cores in fraction of second during ramp up and creates throttling of the core for a fraction of a second, you could see it even in Hwinfo where it says peak temp: 97C since they probe the temp like 2sec intervals but in the thermal throttle event log it will says yes in default ILM). And a problem/flaw, if solvable easily like the ILM, should be fixed and welcomed to be fixed, not defended.

I think we need to see updated coolers or mounting brackets that target the new hotspot to see this 1C difference, especially for air coolers.
From the early previews of De8auer talking about contact frame compa, it seems that the socket itself have moved southward a bit so the old CF wouldn't fit and will need a new one (Though with RL ILM I myself, if build for others, won't use one), so likely they've already shifted somewhat to cater that issue a bit, but since their IHS isn't as thick as AM5, and that ARL should be running much cooler than ADL and RPL, the need of such solutions would be minimal IMO, but reviews will tell
 
Major flaw? Where did I say it's a major flaw? If anything I admire how intel managed to make it so that it doesn't burn itself(not including the ILM and volatge issues of course). Also AMD IHS design was made so it would be compatible with am4 coolers . Even then they managed to achieve better performance compared to intel on what objectively is an inferior IHS design. You don't need 360w design when you can beat the competition at 266w. Look how badly Intel's performance scales with power consumption, it's all diminishing returns in the end. You can have the best cooling solution but if the cpu itself is terrible then it's all moot in the end.
I didn't say YOU said it. Other people have. Even in this very thread.

You realize that intel (pre zen 5) stands at the top of every performance chart, right? Apparely AMD isn't beating the competition with 266w.

relative-performance-cpu.png
 
they arn't immune to criticism.
Right, cause I see you every other amd thread complaining about their major IHS flaw like you are with Intel. Man, who are you kidding? You've mentioned at least 3 times the last 5 days "intel ihs flaw" and "noctua has to come back with special cooler" while intel's cooling solution is 40% better than AMD's.
 
I think we need to see updated coolers or mounting brackets that target the new hotspot to see this 1C difference, especially for air coolers.
MSI designed an offset mount for their AIOs on LGA 1851 which they claim drops temps by up to 3C. If this is accurate then I assume we'll see more offset mounts hit the market. This and the potential benefit from the RL-ILM are two different things and it makes sense that an offset mount would do more.
 
  • Like
Reactions: bolweval
Right, cause I see you every other amd thread complaining about their major IHS flaw like you are with Intel. Man, who are you kidding? You've mentioned at least 3 times the last 5 days "intel ihs flaw" and "noctua has to come back with special cooler" while intel's cooling solution is 40% better than AMD's.
Coz the IHS can’t be fixed, and I won’t do those thinning, yet the performance is competitive enough with all the offset mount, so why bother? Contrary the ILM is an easy fix and some shill just keep saying it’s not a flaw but a design. If it’s AM6 I sure will complain they don’t come up with a new IHS and cooler mount.
 
MSI designed an offset mount for their AIOs on LGA 1851 which they claim drops temps by up to 3C. If this is accurate then I assume we'll see more offset mounts hit the market. This and the potential benefit from the RL-ILM are two different things and it makes sense that an offset mount would do more.
Well it’s an issue of why not both, offset mounting can be done, but no excessive bending in the first place also
 
You realize that intel (pre zen 5) stands at the top of every performance chart, right? Apparely AMD isn't beating the competition with 266w.
wait a minute, you your self stated this here
TPU has lots of errors in lots of their data. Big errors. Like, as big as 70% errors.
so, going by your own admission, TPU has errors in their data, there fore cant really be trusted, but yet you your self always seems to use them and mostly cherry pick them, to prove your point. if you your self, just stated they have lots of errors in their data, and cant be trusted, then, you cant be trusted at all, as most of your " proof " seems to come from them

bottom line, you can use them as a source, but in the thread where i quoted you saying they have errors, it looks as if once some one used them to prove you wrong, or at least questioned you, all of a sudden TPU cant be trusted ????
 
wait a minute, you your self stated this here

so, going by your own admission, TPU has errors in their data, there fore cant really be trusted, but yet you your self always seems to use them and mostly cherry pick them, to prove your point. if you your self, just stated they have lots of errors in their data, and cant be trusted, then, you cant be trusted at all, as most of your " proof " seems to come from them

bottom line, you can use them as a source, but in the thread where i quoted you saying they have errors, it looks as if once some one used them to prove you wrong, or at least questioned you, all of a sudden TPU cant be trusted ????
When TPU's data can be verified by other reviews they are fine. It's not that complicated man, you just need to use your brain. Tom's hardware has some really good cooling reviews and they pretty much agree with TPU's data, that Intel chips can be cooled a lot easier. The same with hwcanucks, their cooler reviews also verify that.

I never said that the graph you posted from TPU was wrong btw. I said they are obviously testing a different non heavy area, that's how they are getting 200 fps in cyberpunk. You cannot get 200 fps in that game in something like the tom's dinner area. You can barely stay above 100.
 
The joke writes itself post by post.

Also, they are now charging people more for the special ILM in LGA1851. Way to go Intel. Never acknowledge the problem but still sell a non-solution to people that has been crying for you to fix it.

Regards.
 
When TPU's data can be verified by other reviews they are fine. It's not that complicated man, you just need to use your brain. Tom's hardware has some really good cooling reviews and they pretty much agree with TPU's data, that Intel chips can be cooled a lot easier. The same with hwcanucks, their cooler reviews also verify that.

I never said that the graph you posted from TPU was wrong btw. I said they are obviously testing a different non heavy area, that's how they are getting 200 fps in cyberpunk. You cannot get 200 fps in that game in something like the tom's dinner area. You can barely stay above 100.
you never said any thing about verifying TPU, again... you said :
TPU has lots of errors in lots of their data. Big errors. Like, as big as 70% errors.
NO where did you say anything about verifying... you flat out said their data has lots of errors, big errors, so goal post moving when some one questions you...

but now you say if TPU can be verified, they arent wrong, so, that still means if you use TPU to attempt to prove your point, and TPUs data cant be verified, then you are wrong, as TPU is also wrong, as you your self stated.

and um, i never posted any graphs, so putting words in my mouth ?

IMO, if you use TPU to prove any of your points, then you better make sure, you also use a 2nd source to also prove it, cause if TPU data is 70% wrong, then that means you are as well, and that lowers your credability.

i just need to use my brain ? sorry, man, but you should take your own advice.
 
you never said any thing about verifying TPU, again... you said :

NO where did you say anything about verifying... you flat out said their data has lots of errors, big errors, so goal post moving when some one questions you...

but now you say if TPU can be verified, they arent wrong, so, that still means if you use TPU to attempt to prove your point, and TPUs data cant be verified, then you are wrong, as TPU is also wrong, as you your self stated.

and um, i never posted any graphs, so putting words in my mouth ?

IMO, if you use TPU to prove any of your points, then you better make sure, you also use a 2nd source to also prove it, cause if TPU data is 70% wrong, then that means you are as well, and that lowers your credability.

i just need to use my brain ? sorry, man, but you should take your own advice.
Again, it's not hard man. If TPUs data are verified by other reviews (or by you yourself having the hardware and testing it), it's fine. That of course doesn't just apply to TPU, it applies to every review, it just happens that I've noticed TPU (and HUB) having errors in some of their reviews. I don't get why you are making it overly complicated.

How it works is, you post some data. If someone disagrees with that data, you can post more data from other reviews that verify them. If nobody disagrees, then there is no reason to post more data to prove something that the other side has already accepted as correct. Common sense man.
 
it just happens that I've noticed TPU (and HUB) having errors in some of their reviews.
then prove it, show where the errors are, its not that hard.
oh wait, you cant, and more then likely wont, and probably never will
i don't get why you are making it overly complicated.
actually, going by 95% of your posts, you do that all on your own..
How it works is, you post some data. If someone disagrees with that data, you can post more data from other reviews that verify them. If nobody disagrees, then there is no reason to post more data to prove something that the other side has already accepted as correct
yea, to bad thats not what you do in 98% of your posts... all you seem to post, is drivel, that comes across as goal post moving, and argueing in bad faith. looks like there a quite a few on here that prove your are wrong, with sources, but you still insist, you are right, with no proof, or you are just miss interpreting the data, to suit your wrong narrative.

but what ever, you believe what you want.
 
Status
Not open for further replies.