News Arrow Lake CPUs can still get bendy despite revamped socket — specialized contact frame flattens chip, claims to lower temperatures by up to 6C

[Admin]

Thermal Grizzly has debuted an all-new LGA1851 contact frame designed with Arrow Lake CPUs in mind. The new contact frame purportedly reduces CPU temps by up to six degrees Celsius.

Arrow Lake CPUs can still get bendy despite revamped socket — specialized contact frame flattens chip, claims to lower temperatures by up to 6C : Read more

 

[TheHerald]

Here we go again 😱

 

[thestryker]

These are very good for flat and lapped heatsinks/blocks, but otherwise doubtful to have any notable cooling advantages (certainly not the up to 4C/6C mentioned). This is the same reason they sell AM5 contact frames Intel's socket is just longer so there's slightly more bending.

Looks like this one is designed to be just screwed in snug unlike the original LGA 1700 one which required tension adjustment.

 

[TheHerald]

These are very good for flat and lapped heatsinks/blocks, but otherwise doubtful to have any notable cooling advantages (certainly not the up to 4C/6C mentioned). This is the same reason they sell AM5 contact frames Intel's socket is just longer so there's slightly more bending.

Looks like this one is designed to be just screwed in snug unlike the original LGA 1700 one which required tension adjustment.

To my understanding, the more layers your mobo has, the more worthless this is, since additional layers will prevent the entire socket from bending.

From my personal experiences with 3 different CPUs on 2 different mobos (same cooler though) these things are just trouble. They reduce peak temps by literally 1-2C but you sacrifice IMC stability - at least when you are trying to run on the edge of your IMC capability.

 

[Sluggotg]

I used these on 5 LGA 1700 builds. They work fine and were very easy to install. I do wish Intel would design the sockets better so we don't need things like this.

 

[rluker5]

These are very good for flat and lapped heatsinks/blocks, but otherwise doubtful to have any notable cooling advantages (certainly not the up to 4C/6C mentioned). This is the same reason they sell AM5 contact frames Intel's socket is just longer so there's slightly more bending.

Looks like this one is designed to be just screwed in snug unlike the original LGA 1700 one which required tension adjustment.

I've got 2 - an Aliexpress noname and an Amazon bought Thermalright and both came with plastic spacer stickers. They both worked fine by snugging to about 1 ftlb (it really just has to hold the CPU in place until the cooler is mounted). But more than once I took one off and some thermal paste had gotten between the IHS and the contact frame and when I pulled up the contact frame the CPU came too, ready to drop at any second and bend some pins. That's a hazard that is pretty rare with a normal ILM.

I do like contact frames though and I would probably get one if I was picking up an Arrow Lake, even if I didn't need to save 4c.

 

[thestryker]

I've got 2 - an Aliexpress noname and an Amazon bought Thermalright and both came with plastic spacer stickers. They both worked fine by snugging to about 1 ftlb (it really just has to hold the CPU in place until the cooler is mounted).

Yeah I was referring to the original TG one which was designed solely with overclockers in mind who might need adjustable tension. If they weren't secured just right for the cooler it was really easy to mess with IMC operation. The noname/Thermalright didn't have this issue as they weren't designed to be adjustable at all just set it and forget it.

I've got one on my server box and will probably continue to get them so long as Intel is using the current style ILM. My primary two systems are LGA 2011 and those along with LGA 2066 just have much better mechanisms, but they also have much higher mounting pressure and cost more to manufacture.

 

[YSCCC]

TBF if it is using the RL ILM in the first place and the bending was akin to the pre 12th gen level I think keeping the stock RL ILM is a better choice as even AIO now are more or less convex to compensate for it as well as the like of Noctua cooler.

The 1700 one became a problem firstly as it is so hot to boot, then the bending even bends the PCB after like 6-12 months as some online users showed, which the CPU still works, just to some extend might affect the long term stability or with cooling performance decreased, plus it made more budget motherboards bendy too.

Now as ARL have the heat center shifted, keeping it dead flat with a convex cooler base might not be a cool idea to boot, and that ARL is finally being reasonable in heat and power draw, it may not worth the hassel to use a contact frame this gen. Of course, for the non RL style ILM, the force needed to lock it from the 12700KF era is like much larger than what I am personally comfortable to use in a CPU, if new board come with that and the thermalright fool proof frame is available at $10 as it is, I will keep going with that with ARL.

All these are pure discussion as myself and those around will definitely skip intel for near future, just see how it goes, at this point of time it looks just like in the "delidding class" to ARL, which is only for extreme overclockers but not for real cooling/anti-(severe)bending for normal users

 

[wingfinger]

If you have a budget where you can upgrade only very infrequently, it is quite surprising how disconcerting the current environment is.

There is the 12VHPWR/12V-2x6 melting issue. There is the 3.5cm no bend zone at the connector and the relevant case issues.

There is power supply, ATX version, model number, churn (that affects the ability to find good reviews for non 1000W, non platinum, ... models)

There are these bending cpu ILMs. The aftermarket frames are an additional expense to increasingly expensive motherboards. There is the issue of the warranty being voided. A warranty is nice is you don't want to plunk down for a new motherboard if anything goes wrong with it.

And, there is the possibility of damage during removal of the old ilm and installation of the new ilm. You could damage your new MB, due to an issue that you shouldn't have. Can this not be solved via slightly thicker and or harder steel?

Would there be a reasonably priced MB with one of these aftermarket ilms pre-installed?

It seems everyone in manufacturing is pushing too hard and it results in issues being pushed off onto consumers. We have had a lot of enthusiasm for tech products but now we have to wonder about how long products will last.

As a side note, the current lack of performance progress isn't quite as concerning to us that are not running that last couple of generations. It is a problem if the situation continues, but everything is still better than what we have now. You might want to consider that when writing reviews. Is there anything to be concerned about? is the last gen, really a good option?

Have the intel CPU damage issues been fixed? Is it too soon to tell?

Is it partially consumers fault for always favoring slightly better performing products instead of solid, reliable and error-free ones? (with, of course, marketing depts multiplying that x1000)

 

[YSCCC]

Is it partially consumers fault for always favoring slightly better performing products instead of solid, reliable and error-free ones? (with, of course, marketing depts multiplying that x1000)

The issue is that reliability only creeps in for considerable time, for CPU to say, normally each generation is one year, and a socket in the not extended tick tok cycle is 2 years, with RPL refresh for example being the third year. The bending issue was discovered and documented/spread in the internet after like half a year or so since release, and it is a minor issue to boot with (may have performance hinder due to the contact issue and thermal throttling, but to make a larger number, it is still below 5% of a non-bending CPU using a good air cooler), and degradation was discovered after some 6-9 months, and since everyone could have overclocked, misused or simply being unlucky, it took it till almost end of generation for it to be finally admitted and tries to fix. So in that sense, we cannot opt for the less performant but reliable part.

Ironically, I usesd to opt for Adler lake and not Zen 3 was like AM3 was at EOL, AM4 wasn't out yet, and traditionally, Intel is more reliable and more optimized (understandably) by software developers, as they are the market dominating brand, so if one do production also and not chase for the last drop of performance, Intel was always the safer choice, then it comes the RPL disaster......

 

[newtechldtech]

The CPU Cooler should be screwed into the case and not the motherboard. and the socket should be simple one (like notebook sockets).

it is about time they change the old way of screwing everything on the motherboard.

The standard screws layout should be on the case same like motherboard mounting holes.

 

[rluker5]

This article reads like sponsored content.

 

[YSCCC]

View: https://www.youtube.com/watch?v=GOTpkf4rVPQ


A bit more through data on the Standard ILM vs the RL ILM thing, it seems if you don't have a cooler exerting enought pressure on the CPU, with the RL ILM there might be contact issues, but with a cooler having enough pressure, the RL-ILM is better in thermals and substrate bending as the 1700.

It sounds risky for the RL-ILM at first but with boxed cooler being a historical relict and modern-ish coolers all have noctua style spring loading to sustain the heavy tower, or to securely mount the AIO coldplate, it sounds like basically all coolers shouldn't get into problem into the mounting pressure. If anything those contact frames should be an easy job too.

 

[TheHerald]

Playing with contact frames for 3 years to get a 2C reductions in temps, meanwhile LM will you get 10C. Yeah, I'll pass.