*** MSI 16L13 (Eurocom Tornado F5)/EVOC 16L-G-1080 15.6" Owner's Lounge ***

I've done some additional diagnostics on my stability problem and those pesky WHEA-Logger "correctable" PCIe errors.

It occurred to me to review the really old Windows event logs to establish exactly when those began occurring, and then to cross-correlate with the setup logs (from another part of the event viewer) to see what changed at that time. The results surprised me.

It turns out, the correctable PCI errors go back as far as the logs do, way back in May. The first entry is on May 15, and there is nothing even close to that in the setup logs. This means it's unlikely the errors relate to any software changes. They may relate to my overclock settings, but I think it's also possible this error may just be something that happens from time to time.

The main point is, this exact same error started back in May, and there have been over 10,000 occurrences since then, even during a time when I considered the machine to be absolutely rock solid. Conclusion: The WHEA-Logger correctable PCI error is, surprisingly, unrelated to my recent stability issues.

 

Played BF1 @4.6GHz and 1110Mv, crashed, put it on 1130Mv and it is stable but CPU gets to 93c and core 1 thermal throttle, I need to try it on 1120Mv and use an air pad cooler to see how the heat is going to be on that CPU, GPU was a cozy 85c max. 0.9v.

 

I took the screws back out of my heatsink, as you suggested, and went a step further. If the point is to insulate one from the other, then @Mr. Fox (re: this post) has the right idea with the cardboard. I used a section of paper napkin to achieve the same thing:



I was uncomfortable with the idea that sometimes it might make contact more than others, so for consistency I wanted it either fully in contact, or fully out of contact.

My curiosity is around the whole notion of whether the coupling is good or bad. If either the CPU or GPU is significantly hotter than the other, then one might suppose the coupling would, in effect, allow the other heatsink section to take up some slack. If both devices are approximately equal temperature, then the coupling would not matter because there is no delta-T to drive heat transfer in either direction. This would behave like a load aggregation (thermal, not workload). But if one device is transferring heat to the other, and it puts the second device into thermal throttling, then that may be a net performance loss for the system as a whole.

One question that arises for me, is that I will often have three very distinct categories of workload:

• Compiles, databases, and lightmap bakes (and similar batch jobs) load my CPU at nearly 100% on all four cores, but almost ignore the GPU.
• Games that I play tend to load the GPU heavily and the CPU less so.
• My Unity visualizations/simulations tend to behave differently from traditional FPS games, in that the simulation code loads the CPU more than in many games. The GPU, however, still is loaded more heavily than the CPU.

I wonder if the unified heatsink would help for #1 and #2, but then would be a drawback for #3? It's going to take a long time to figure this out. @Donald@HIDevolution, would it be possible for you to send me (at your convenience, no hurry) a set of the missing screws, so that when/if I do want it coupled, I can have it firmly coupled? I've got a support ticket open with @Zoltan@HIDevolution, but obviously he's working remotely and isn't going to be the one to pack and ship them.

I'd love to hear from @Mr. Fox and other folks who've been testing this aspect, to better understand your reasoning as to why you feel the coupling pipe is overall a bad idea. I'm not intending this to disagree with you, just to better understand the rationale.

 

The issue was fixed awhile ago actually. All new Eurocom systems ship with the latest system BIOS which includes the brightness controls for the 120Hz panels. Buyers that got the 120Hz panel before the BIOS update was available to fix the issue should be able to schedule a remote flash session with one of @Eurocom Support tech support agents to complete the flash to fix the issue. At least that is what they did with me and I have brightness controls with my 120Hz panel and Kaby Lake support.

 

Yup, just create a ticket with support on EUROCOM website.

 

I can half picture a meme/gif where the tail wags the dog now, lmao

 

Undervolting isn't de-tuning, you don't loose performance you loose excess heat production.

Why do you all not get this? All the people with laptops that undervolt immediately see the benefits, yet you all are curiously blind to the benefit's to this simple tuning technique.

Conveniently so, I might add

 

Actually I agree this is "tuning" not de-tuning, as far as terminology. Why care so much (about what people call it) though ?

 

You are joking right?

De-tuning is reducing performance, tuning the CPU voltage to meet particular CPU need doesn't reduce performance, it reduces heat, which can actually increase performance by staving off thermal throttling.

People can benefit from undervolting, my caring about them getting the false impression that it's "bad" for performance is understandable.

What I don't understand is why you all are doing this, what's the goal of providing disinformation about undervolting, giving the impression it's a de-tuning activity?

Really wanting LGA to replace BGA is fun, telling people BS to get them to drop BGA through providing false information isn't.

That's why I care

 

NOOOOOO

Do not put paper, especially thin tissue/napkin paper on hot surface. While it is true that the auto-ignition point of paper is ~218 C, prolonged exposure will degrade the wood cellulose and glue turning it to crumbly dusty mess. Tissue, napkin, and cardboard are heavily recycled; there is no guarantee any of those particles are not plastics or foams that will smolder at much lower temperatures.

 

With respect to both sides of this discussion, as a neutral party it seems to me there may be some semantic disconnect.

Fundamentally, undervolting, overvolting, and overclocking are all exercises in pushing a specific set of components (as part of a complete system) closer to their limits. Engineering is the science of compromise among competing input parameters in order to optimize a set of output metrics. What "optimize" means is different for each person. Every parameter of every component has nominal values, functional limits, and absolute limits. We can turn a few dozen knobs, which internally impact thousands of component parameters. Which limits are you willing to approach, and which limits are you willing to exceed under carefully-controlled conditions? How close to the edge of the slope, or the cliff, can we safely dance?

Reasonable people could even disagree about what the list of output metrics includes. Frame rate? Frame rate predictability? Attainable pixel resolution at a given frame rate? Visual quality at a given resolution and acceptable minimum frame rate? Peak temperature of components? Average temperature of components? Benchmark score? Component longevity? Stability? (And for stability, how do you quantify that?)

This is an intrinsically multi-dimensional problem, and there is no single closed-form solution, even if every one of these computers and their environments were absolutely identical. People with equal skill, and equally valid motivations, can still disagree on what the word "optimize" means in the output metrics. It's not even a simple case of deciding which metric is "the most important", but rather a complex decision in which they are all important, but to varying levels according to your own goals.

I respect all of your opinions, and I'm sorry if I've offended any of the fine folks here from whom I am learning so much.

 

That sounds like pretty solid advice, which I will heed. Thanks.

What alternative insulating material would you recommend I consider? I thought about plastic but was concerned about melting and/or embrittlement.

 

Meanwhile, back at the ranch...

I reverted my Nvidia driver to the last version I had that was stable, and also reverted my MSI Afterburner settings to that same point in time. Most importantly, I visited the Nvidia control panel to see what might be going on there, and found that G-Sync had been spontaneously enabled -- presumably by one of the Nvidia upgrades. Sneaky little devils! I know I didn't turn that on myself, because don't use it.

My CPU voltage is now 1.150 static, up from 1.140, and that change seems to have made things a lot more stable for me, overall.

Anyway, with these changes, here's my latest FireStrike score -- a personal record! Also, my temps for CPU stayed around 70C, and the GPU around 80C, maximum, and this was the latest of several FS runs with only a brief interlude between to make the next tweak. I credit the bottom panel mod for the lower temps.

 

The recent move to de-tune GPU's using Max-Q Design to fit full performance GPU's into small frame laptops is a performance losing method of de-tuning.

There is more than reducing voltage in Max-Q, there is also frequency reduction and power throttling.

That is a great example of de-tuning.

Another example of de-tuning is when a laptop has a broken cooling system and you are trying to make it through a few hours, days, weeks to get use out of it until you can get it fixed.

You do this by reducing the maximum frequency, reducing the available cores (disable Hyperthreading), and disabling Turboboost scaling. This reduces performance to reduce thermal generation.

That's another example of de-tuning to reduce performance to reduce thermal output.

The key connection between those two examples is they both reduce performance to reduce thermal output to keep from overheating.

Undervolting is tuning the CPU / GPU voltage to minimize voltage used to what is required, and to reduce thermal output while getting the *the same or better performance*.

That's the difference, and why it's important to distinguish the difference.

I've already had one random person ask about performance loss when undervolting. That's why I am pushing back so hard on this right now.

Now is the time to stop the disinformation before it ruins people's lives.

 

I respect you and see your point and can agree with your sentiment. It's still semantics. When I run my book at 4.9 GHz I get much better performance than I do running my book at 4.5 GHz with voltage reduced to the limits. My temps are better at 4.5, but I'm not experiencing any throttling at either frequency. It can be argued that keeping the system at 4.5 with lower voltage is a "detuned" state for this book even if it is tuned at 4.5 to run as efficiently as possible. The Toyota Supra came fully tuned at about 320 hp from the factory. It runs efficiently at that state of tune. Those in the field consider that detuned because they know with stock components the Supra can easily handle 500 + hp. You consider it tuned because the thermals are great and the performance is good. @Mr. Fox considers it detuned because he gets better performance at a higher frequency without sacrificing too much in terms of thermals and almost nothing in terms of reliability.

Sent from my SM-G935T using Tapatalk

 

This is why I made the point about semantics, as distinct from any question of technical merit of one side or the other.

Another way to look at the preceding example is this: "Because of a cooling system failure, I temporarily choose to tune for reduced power consumption and am willing to compromise system performance toward that end." Really, all you have done in this hypothetical case is to change which variable you think is most important. You are still trying to achieve the "best" result, just changing what you mean by "best". Semantics.

I do hear what you're saying about undervolting while constraining constant performance. Again, though, your use of the term implies constrained constants elsewhere. Someone else might more broadly define undervolting as "setting the CPU voltage to a value lower than the manufacturer's published nominal", without regard to why you are doing so or what else is or is not changing.

I'm not trying to take sides, just pointing out that the words themselves can be interpreted more than one way, and I think that may be part of the reason for the miscommunication.

 

The pointed use of "detuning" to describe undervolting is the issue, not any other nebulus semantic issue.

"Detuning" carries with it the implied baggage of reducing performance.

Using "detuning" has picked up common use here recently, and like I already mentioned, someone asked me about the performance loss by undervolting that happened in parallel with the "detuning" comment.

There's a clear difference in the definition of the terminology, there's no semantic confusion involved.

de·tune (google)
present participle: detuning

1. cause (a musical instrument) to become out of tune.
2. reduce the performance or efficiency of (a motor vehicle or engine) by adjustment.

http://www.thefreedictionary.com/tuning
c. To adjust (an engine, for example) for maximum performance.

Tuning means getting the best performance, detuning means reducing performance.

Please stop using "detuning" when talking about undervolting as there is no loss of performance when undervolting.

 

You have sabotaged your own argument. By limiting voltage you limit frequency which in turn limits performance - that is one of the exact definitions of detuning as you have posted. It should be clear now how some can refer to performance detuning and how you may see it as temperature tuning. This book has been proven to handle the 7700k at above stock frequencies with better than stock performance. To those that understand this character of the book, they see undervolting to maintain 4.5 GHz as a lower state of tune or detuned. It is semantics and even moreso it is perspective. If your perspective is temperature is performance then you will consider undervolting as a state of tune, and if you consider ability as performance than you will consider highest safe frequency as tuned. When you are running your highest possible frequency, your temps aren't going to be at their absolute best even if they are safe and stable. Those that consider temps as the priority state of tune may see this as detuned. When you are running your lowest possible voltage, you will not have absolute capability, and those that prefer to squeeze every drop of blood from their machine safely will consider the book detuned. For most situations there isn't going to be only one viable solution...

Sent from my SM-G935T using Tapatalk

 

You are trying to make undervolting far more complex than the simple thing that it is.

I've helped hundreds (thousands?) of non-technical and technical people undervolt easily.

The benefits of undervolting are immediately verifiable as a drop in CPU load temperature that is usually enough to stop thermal throttling without repasting.

Undervolting is the right first thing to do (after backing up your recovery partition) when getting a new laptop, It's easy to do with immediate positive benefits.

Again, there is no performance loss or frequency loss involved with undervolting via a CPU voltage offset

 

All of what you said is true and still does not negate that you are potentially running at a lower performance state than what you could be achieving with higher voltages. This scenario can be viewed as a lower state of tune or detuned.

Let's simplify it. If you are achieving < 85 C at 4.9 GHz (which I can) and you do not experience throttling (which I don't) would you say the performance is going to be lower than < 78 C temps at 4.5 GHz? Keep in mind IPC doesn't care about temps. Throttling does. I'm not experiencing throttling yet I am safely running more clocks giving me more performance per second increasing my capable performance. 4.5 is a decrease in performance compared to 4.9 and provides little to no stability or reliability benefit. Your definition specifically called a reduction in performance detuning. Is there any point of contention?

Sent from my SM-G935T using Tapatalk

 

I work remotely too...don't worry though, opening the ticket with Zoltan is the correct thing to do. He will take care of you

Fortunately in this wonderful digital world, we no longer have to congregate in a single location to get things done.

 

Read everything in blue As I have pointed out several times.... 10mv higher voltage can/will give you some headroom against risking stability problems.

 

Locked CPU's that can't OC aren't missing out on higher frequencies because they can't reach them by design. There is no advantage to running higher than needed voltages on a locked CPU.

For those locked CPU's the only tuning available is undervolting to reduce thermal output, reducing the temperature by tuning for the lowest stable negative voltage offset.

There is zero advantage in running those locked CPU's at a higher than *needed* voltage.

The too high BIOS CPU voltage setting is a detriment for these CPU's, undervolting fixes that problem, reduces heat - stops thermal throttling for most that are experiencing it.

Your simple example is inadequate even for a desktop CPU, you don't mention tuning the CPU voltage. Are you running with the default BIOS CPU voltage setting?

When I discuss undervolt tuning for unlocked CPU's it's a different discussion, which can be confusing to people with locked CPU's or unlocked HK/K CPU's they aren't comfortable overclocking so I keep it simple and leave it out.

The 16L13 K CPU can also be simply undervolted, and if you look back in the thread you will see people mentioning this. It's something to do early on before getting into the more involved process of finding maximum OC and tuning the CPU voltage up and then back down when optimizing for maximum OC vs reducing heat output.

Undervolting is typically used to describe the simple method of changing the voltage offset available for laptop CPU's, so perhaps that's what is confusing you?

 

Why do you keep saying that? Of course if you are unstable after lowering the CPU voltage you would increase it again.

Once you are stable there is no advantage to increasing CPU voltage another 10mV if you are already stable, right?

Or if you are now *more stable* at that higher 10mV setting, then *that* setting would be your stable setting.

You wouldn't continually increase your voltage setting by 10mV randomly assuming another 10mV would make you "more" stable.

 

Then, why are you hijacking this Owner's Lounge to discuss BGA CPUs? There are no BGAs in this model. Perhaps...this is what is confusing you.

Please take your discussion to an appropriate thread.

 

You can simply undervolt an LGA CPU laptop, it was discussed and done here in this thread many times, so it's not off topic.

 

We were on the conversation of undervolt so it is implied. An undervolted 4.5 GHz 7700k is not going to compete with a 4.9 GHz properly volted counter part. We are in an owers group that doesn't deal with cpus that have no OC ability.

Sure if you cannot overclock, tinker where you can and this conversation doesn't apply to you or your situation.

The question needs to be asked, is your end goal achieving understanding or is it proving your point regardless of how far outside of the boundaries of the original conversation you need to travel to do so?

We already know how certain members feel about cpus that can't be overclocked and are soldered onto the MB. This isn't that conversation. This is about having the ability to push clocks up, but deciding to undervolt instead. In this particular situation - choosing undervolting can be legitimately viewed as detuning. We don't need to talk about male penguin fatherly responsibilities to prove or disprove this point.

Sent from my SM-G935T using Tapatalk

 

I've kept on topic, and I've done my best to clarify the reason that undervolting isn't detuning, and it's not going to reduce performance.

Of course you are going to set the voltage you need to get the OC you want, it's the same for an HK or K CPU.

*After* you get a general voltage setting that lets you get the highest OC you want, then you undervolt it to get it down to the minimum voltage required for stability at that highest OC.

You aren't going to reduce the voltage further than needed to be stable at the highest multiplier or OC frequency setting you are able to attain, but you'd be incorrect to insist on keeping a higher voltage than necessary to attain that OC setting.

So, locked or unlocked, undervolting to the highest stable setting isn't detuning, as you are still running at the highest attainable OC and the highest performance.

It's that simple.

 

Frankly, I find your conversation to be neither polite, nor winning anything. Rather it tends to be downright rude and disrespectful, as well as off topic.

To say that "I haven't been talking about the more complex method for OC'ing + Voltage tuning for a desktop CPU. I've mainly been talking about undervolting laptop CPU's." is not off topic boggles the mind. This is an Owner's Lounge for LGA based laptops having nothing to do with BGA based laptops, and therefore your stated topic of conversation is indeed hijacking...and as you can see from the responses...not appreciated here.

I repeat my request. Please take your conversation to an appropriate thread...or if you want to stay here, keep your conversation on topic, polite, and respectful.

 

It's not that simple. At 4.9 GHz I'm running above stock voltages so I'm not undervolting at all. I'm overvolting.

Intel created these chips to handle frequencies above 4.5 as denoted with a "K" in the alphanumeric model number. That means at factory release these are already detuned down to 4.5 with an over safe voltage. Just because you tune the voltage to maximize that specific frequency doesn't mean you have even moved out of the detuned state the factory shipped the chip in. You have merely moved into a better state of tune than what was given. An example of this - you buy a guitar from the store. It is out of tune but all of the strings are equally out of tune so it sounds okay. You tune the low "E" string by ear and come out with a "D" instead and continue to tune the rest of the strings to that note. The guitar now sounds much better than before, but it still isn't living up to its potential. Hitting that optimal sound means tuning it to what it was really designed for which is "E." You may have tuned the guitar to "D" and it may sound great, but it is still not tuned to what it could be. You are still detuned. Let's not get into the argument that you prefer "D" tuning, because at that point there is no reference point of tune it all becomes subjective opinion.

Sent from my SM-G935T using Tapatalk

 

We are having a civil discussion between members, and it's been far more on topic than deserves your comments.

My single sentence out of a dozen or more posts: "I've mainly been talking about undervolting laptop CPU's"

Was added to help them understand the scope of the discussion doesn't just include LGA CPU's, while not excluding them.

Most of my undervolting discussions are with people that don't have an unlocked LGA CPU, which is relevant to the discussion in this thread because the confusion started in this thread by claiming undervolting was giving up performance was spreading outward and confusing people.

One of them asked today about the loss of performance from undervolting, which was started by discussions in this thread and elsewhere.

So I came here to ask politely to please stop using the term "detuning" with undervolting because it's not true and it's confusing people.

The long discussion came up from that simple request.

 

Yeah, and I also recommend overvolting when raising the OC past the stable point for the BIOS CPU voltage setting.

But, when you are overvolting you are doing it in increments that are larger than needed, at least I do, to more rapidly get at the highest stable clock rate.

Whether you are undervolting from the fixed CPU voltage that came in the BIOS, or the higher than needed voltage setting you came up with to reach the highest OC, you eventually "undervolt" that setting to arrive at the final stable voltage for that multiplier and frequency, without losing performance.

After I've verified a stable OC, then I start reducing the voltage again by smaller increments until unstable, then bump it up that increment again.

I might go by 25mV jumps when getting to an overvolt I need, and after finding the stable OC, drop down 5mV at a time until unstable, then jump up 2mV, drop down 1mV.

Always tuning for highest performance, and lowest voltage for best efficiency and lowest heat output.

The tuning for a stable "undervolt" happens when you've picked a frequency desired - either highest OC, stock settings, or at various points for specific use profiles.

I have a profile for long term running at 36x and a much higher undervolt, while at maximum 4.2ghz I have an overvolt, and at 4.0ghz I run with a slight undervolt.

All different multiplier frequency settings, all with their own unique voltage offset setting that was found by "undervolting" to tune for the lowest possible stable CPU voltage.

The goal of undervolting isn't increasing performance or reducing performance, it's a performance neutral tuning for minimum voltage required at a particular CPU multiplier or frequency to reduce thermal output from the CPU.

Does that make sense?

 

I'm 110% sure yoo understood what I meant

I fix it for yooo

 

Rubber would be good. Neoprene is very good insulator, good to temps around 260 C. More importantly its decomposition temperature is 200 C

https://www.amazon.com/6x6-Inch-Neo...ie=UTF8&qid=1504353323&sr=8-5&keywords=rubber

 

@ThatOldGuy 's suggestion helped me to recall what I've used in the past for long term thermal insulation between metal to metal or metal to circuit boards, and that is Ceramic Insulator Plate material.

Usually they come in sheets or precut squares for a particular application. You can find them at electronic component supply houses - cheaper because you can pull one or two from a bin, otherwise you need to buy a quantity of them online.

Here are a couple of examples from amazon:

https://www.amazon.com/uxcell-Ceram...54976&sr=1-2&keywords=Ceramic+Insulator+Plate

https://www.amazon.com/uxcell-Ceram...54976&sr=1-2&keywords=Ceramic+Insulator+Plate

And from Alibaba:
https://www.alibaba.com/showroom/ceramic-insulator-plate.html

You could also use a section of thin PCB blank. A high temp reinforced ceramic tile thin enough would also work. If you don't have an electronics components depot nearby, you could try a hardware store, Home Depot, etc - maybe call around and ask. They might have flexible "cloth" ceramic weave that could also work.

Orchard Supply usually has a seasoned expert on site that can help with obscure requests like this

 

It makes sense. I've already stated that I agree with many of your assessments. The problem that we are having is you are being very selective on how you want to define things and very rigid with trying to understand how others mean them. You are calling something undervolting even in the condition that someone is above stock volts but by the lowest possible margin for stability and thermal efficiency. I can agree with that interpretation even if it isn't accurate because I understand what you mean. That's the whole point of having conversations - understanding the other perspective. So, once again, are you trying to understand? If so allowing people to use terms in a way you may not see fit shouldn't be an issue because you are looking at the broader scope of their intentions. On the other hand, you can be unyielding in allowing others their own interpretation of words just to be right.

Nobody here is arguing that using the lowest most stable voltage at any given frequency is bad practice. The argument is - get your thermals in check by making sure all surfaces are mated properly, proper TIM is correctly applied, make sure the system has optimal airflow and heat dissipation, then use what voltage you can to maximize your clocks without running extreme temps, back your voltage down at your comfortable highest OC to get the best thermals and stability. If you can't OC or have hit your thermal limit than there's no need to OC higher just focus on your voltage at that point.

Sent from my SM-G935T using Tapatalk

 

Crashed again at 1130Mv. 4.6GHz.
Increased to 1150Mv, 4.6GHz, CPU hits 97c.

Now reduced to 1120Mv and down to 4.5Ghz, time to test again. I'm ordering some kapton tape, some phobya, kraunaut and conductnaut grease tubes and a SF-17 cooler pad via HIDevolution, time to get serious with these temps.

@Mr. Fox where did you get that foam stuff from to make the dam?

EDIT:
Crash Dump Analysis
Crash dump directory: C:\WINDOWS\Minidump

Crash dumps are enabled on your computer.

On Sat 02-Sep-17 18:03:28 your computer crashed
crash dump file: C:\WINDOWS\Minidump\090217-6687-01.dmp
This was probably caused by the following module: ntoskrnl.exe (nt+0x16C560)
Bugcheck code: 0x50 (0xFFFFF8027E0F3340, 0x0, 0xFFFFF8023548A7F9, 0x2)
Error: PAGE_FAULT_IN_NONPAGED_AREA
file path: C:\WINDOWS\system32\ntoskrnl.exe
product: Microsoft® Windows® Operating System
company: Microsoft Corporation
description: NT Kernel & System
Bug check description: This indicates that invalid system memory has been referenced.
This appears to be a typical software driver bug and is not likely to be caused by a hardware problem.
The crash took place in the Windows kernel. Possibly this problem is caused by another driver that cannot be identified at this time.



On Sat 02-Sep-17 18:03:28 your computer crashed
crash dump file: C:\WINDOWS\memory.dmp
This was probably caused by the following module: ntkrnlmp.exe (nt!KeBugCheckEx+0x0)
Bugcheck code: 0x50 (0xFFFFF8027E0F3340, 0x0, 0xFFFFF8023548A7F9, 0x2)
Error: PAGE_FAULT_IN_NONPAGED_AREA
Bug check description: This indicates that invalid system memory has been referenced.
This appears to be a typical software driver bug and is not likely to be caused by a hardware problem.
The crash took place in the Windows kernel. Possibly this problem is caused by another driver that cannot be identified at this time.

 

Unless an unstable overclock has slowly created corruption in your operating system over time, and now issues are surfacing? (sfc /scannow)?

 

That foam is just regular packing foam. Usually used for small parts. Any hardware (not computer hardware) store should have it, or should know where to get it. A hobby crafts store should also have materials like that. You want the non dense one that is easily compressible to literally nothing.

I believe it's something like this:

https://www.amazon.com/EcoBox-Polyu...4366948&sr=8-5&keywords=foam+packing+material

That looks exactly like the piece I have, except it's thicker. But if you trim it down it should function the same.

Worried about the prices in that link though and how much you get..it's important for the foam to be highly compressible (you should be able to compress it with your fingers to literally a micrometer, so it doesn't interfere with heatsink pressure. You may want to send them an email and ask exactly before spending good cash for that, if you can't ride down to a local store and buy it.

 

All this "tuning" & "de-tuning" arguments are pointless. Call overclocking overclocking, call undervolting undervolting, call over-volting over-volting, call under-clocking under-clocking, you don't even need to use the words tuning & detuning because they're too vague anyway. Describing undervolting as "detuning" is blatantly inaccurate & detrimental to the newbie understanding what undervolting is, but really there's no need to even use words like tune & detune, not really specific enough.

 

Happy Saturday! I see we are burning a lot of calories pole vaulting over mouse turds again. LOL.

I was clear that undervolting is necessary for most notebook owners to run stock clocks without overheating due to half-assed engineering and inexcusably poor thermal management on most laptops. This is status quo. Even with a wimpy BGA piece of crap with half the TDP of a good CPU requires undervolting, and even with undervolting most of the sorry notebook trash being sold is still running too hot to be nice. It's a necessary evil due to sloppy engineering and wanton disregard for quality. And it is complicated by anorexic form factors that are too small to accommodate anything awesome. While one still could undervolt and get away with it, they wouldn't have to if cooling systems were properly engineered. Since I don't run anything stock and only use unlocked CPUs, it's really nice just being able to flick the multis up another 300 to 500MHz without having to jump through a bunch of hoops farting around with a pathetic abortion that can't even run stock without danger of a thermal meltdown.

I think this bears repeating...

...the right amount will be case-by-case and vary with CPU manufacturing variance, thermal paste, ambient temperatures, and the degree of apathy involved with the engineering of each product. We're fortunate most of the sorry notebook manufacturers out there only went about 90% Nazi on us and left the voltage adjustable, unlike many other things. Had they locked that down as well, we would not have any mouse droppings available as props for our exercise program.

 

The difference is small with it disconnected, but an accumulation of small improvements can add up to a lot. Unified heat sinks in general is an inferior design concept. I haven't seen an example of one that works well. It also increases the possibility of improper fit due to misalignment. The improper fit can occur on the assembly line or later on when the machine is disassembled for service and something happens that throws things off in terms of fit. I'm guessing it must be cheaper to make them and may use less space in the chassis, and since most manufacturers seem to not care whether the products they sell function correctly, anything goes.

What makes me question the intelligence of the "finger" pipe are the following observations:

• Overclocked CPU runs hotter than the GPU
• The GPU already runs hotter than we would like it to
• The heat pipe bridging them leads to a dead end both directions

By dead end, I mean either direction leads to heat. Heat will flow toward the colder part until they equalize and then it will just start storing heat. There is no way for the CPU or the GPU to effectively shed heat by them puking on one another. The heat needs to flow in the direction of a radiator, and there is no radiator.

 

Oh, you suspect people think it's bad for performance ? Our audiences here on NBR are quite capable of understanding it, IMO.

I do not think "Tuning" it fits, it sounds or feels like we're speaking of a car... but guess in lack of anything better or more appropriate it's fine.

It's weird how INTEL has these CPU's overvolted to begin with (or is it the chipsets/motherboard manufacturers?) , I don't quite understand why or how it happened anyways.

It's like they wanted to pretend LGA caused more heat, so desperate infact, they did a fairly poor job of cooling the Skylakes under the IHS on purpose. I don;t believe conspiracy's so I think it's more likely some budget cut somewhere.

 

Yeah and I personally always hated the way they combined heatsinks in recent laptops. Like the BGA's and shiz, I can understand, they are aiming for mediocre, so be it. But MSI ? in a LGA - why ?

 

LOL... if there is one thing we can count of for consistency, all of them are stupid in one way or another. It would be nice if at least one or two notebook manufacturers did most things right most of the time. We have to carefully and methodically pick our poison. That's why I like the P870 product family better than anything else. It's far from perfect, but it has the fewest number of stupid mistakes as far as I can tell. The Tornado F5/16L-G-1080 is a close second place. Both are a good place to start with fewer mods and tweaks needed to end up having what you expected to get for your money. I wouldn't touch a P7 with a 10-foot pole because of the unified heat sink. Otherwise, it's a nice product.

Add the stupidity of Micro$loth and their Windows 10 feces smartphone OS and it's a small miracle that there is any such thing as a PC any more.

 

A barebone using the GT73VR would be awesome.

 

The main problem today is the massive thin, thinner thinnest mentality who flowing out everywhere. I'm sure the Engineering teams who work in the notebook industry is forced to make inadequate mediocre cooling due cost and budget... Or, there is a shortage of people in the notebook industri today who really can Thermal Engineering. Pick what suits. But the main problem is still... The final product must be able to be produced at the absolutely lowest possible cost!! Quality will as always suffer and come as number two or last. Yeah, last There is no reason to not be able to create cooling in laptops that work excellent with a OC'd 91/95W mainstream LGA CPU or a single 1080 graphics cards today!! 1080 Sli can be more challenging, but fully possible. Just avoid making and use AW, Razer, Aorus or an Apple designed chassis

 

I shall look for it locally, wish me luck as I know sometimes even simple stuff like this can't be found in a third world country! Maybe if I go to Dubai....

 

Thanks, that's all I was asking to have recognized as it was affecting newbies that all of a sudden were worried about losing performance by undervolting, and I tracked it down to using the word detuning in reference to undervolting.

As @Papusan said this all falls under the umbrella of Overclocking.

Back in the day before undervolting everyone would find the highest voltage that was safe and rode that voltage to get the highest OC, assuming more voltage was better.

As an engineer, best circuit part tuning isn't necessarily at the highest voltage, so I experimented with dropping the voltage, then trying higher OC, and it worked. The side effect was often less heat - voiced by the PC as lower fan speeds. At that point I perfected my tuning for least voltage at highest performance.

I used the word undervolting to describe it as the opposite to overvolting - running higher than stock CPU voltage.

Really it's the process of tuning for lowest voltage at highest performance.

The BIOS CPU voltage setting is as arbitrary as the first Overvolt choosen - one is vendor recommended operating voltage that is stable for all CPU production variation, and the other is highest "safe" voltage.

These days most people share their highest OC voltage, but it's not necessarily the lowest voltage possible to get that OC, and of course it varies across individual samples.

So from any high voltage, undervolting is done to get the most efficient stable cool running operation.

The problem comes in when strictly assuming the BIOS voltage is the marker between undervolting and overvolting.

To get highest OC you are often not going to use the BIOS voltage, you are going to want to set the highest safe voltage (modulo pucker factor - how close do you want to get to max safe voltage) and OC from there, and undervolt from that voltage setting.

Anyway, enough of that, I'll drop replying to the other responses, I think we've reached an understanding - please don't frighten the newbies. Thanks

 

I suspect something is wrong with your heatsink seating or current paste job. I am running my i7-7700K @ 4.6 wth -0.100V. I barely break 72 C on a stress test.

When you repaste, make sure to screw each screw in gradually in order