4790k OC is at 100 degrees!!

[done12many2]

7 minutes ago, Imakuni said:

Reminder that AVX-512 is (supposedly) coming with Cannon/Coffee.

 

Guess I'll sell my oven and use one of those processors instead...

 

Spot on bud!  512 is in Skylake X as well. 

 

AVX 512 is going to be fast though! 

 

People lack the understanding of things like AVX, Speed Shift and other enhancements that improve performance at the expense of additional power draw in some cases and heat. 

[done12many2]

3 minutes ago, CostcoSamples said:

Thanks, I appreciate the voice of experience.

 

You've got a badass setup and a nice OC man.  Enjoy it and let that Intel 750 SSD rip!

[Majestic]

40 minutes ago, Imakuni said:

It's a Hyper 212-esque type cooler... c`mon, you can't expect too much out of it. People overhype the power of those air towers.

LOL, a NH-U12S is nothing like that thing. People overhype the 212.

[PCGuy_5960]

1 hour ago, tom_w141 said:

That Intel TIM tho...

For the millionth fucking time. The thermal paste that Intel uses is called Dow Corning TC-1996, EVGA uses the same paste and Intel uses the same paste on Skylake and Devils Canynon and most of those CPUs don't have thermal issues. Which shows you that the thermal paste itself is fine, but the application method sucks. You always keep saying that Intel should solder their CPUs, but for the millionth time I am telling you that Intel can't do that:

http://overclocking.guide/the-truth-about-cpu-soldering/

If you think that the article is false, send an application form to Intel. They could definitely use your "expertise".

Now can you please stop shitting on Intel?

[done12many2]

23 minutes ago, Majestic said:

LOL, a NH-U12S is nothing like that thing. People overhype the 212.

 

That Noctua is pretty sweat for an air cooler. 

[CostcoSamples]

30 minutes ago, Majestic said:

LOL, a NH-U12S is nothing like that thing. People overhype the 212.

agree.

6 minutes ago, done12many2 said:

 

That Noctua is pretty sweat for an air cooler. 

Yes, I agree it's pretty awesome.  I've had this cooler since 2008 and it's still running as new!

 

But of course single tower with single fan can only do so much.  I can't buy a new cooler right now, so I was thinking of adding another fan (would have to be a slim fan due to RAM clearance).  Hopefully that buys me another 5 to 10 degrees off the top!

[Guest Darth Revan]

That processor is a 90 watt tdp while the cooler is a 95watt tdp.

If you overclock the processor it's going to go beyond those 90 watts by a lot.

So the cooler is the problem. It's not capable of cooling the processor.

You need a better one.

[tom_w141]

58 minutes ago, done12many2 said:

 

Why don't you just add this Intel slam to your signature so you don't have to post it as much?

 

Could it just be something as simple as Prime95 at 1.3v on a Haswell architecture chip is a bit much to ask of a Noctua NH-U12S cooler?

 

Every single temp problem with an Intel chip is not related to TIM.  

 

For a guy who proclaims to not be an AMD or Intel biased individual, you are definitely not acting like it.  

 

 

This is about what you can't expect out of your cooler bud, given your voltage and clockspeed.  Prime95 with AVX is a beast on any CPU.  

I'd like to see the gains post delid.

[tom_w141]

31 minutes ago, PCGuy_5960 said:

For the millionth fucking time. The thermal paste that Intel uses is called Dow Corning TC-1996, EVGA uses the same paste and Intel uses the same paste on Skylake and Devils Canynon and most of those CPUs don't have thermal issues. Which shows you that the thermal paste itself is fine, but the application method sucks. You always keep saying that Intel should solder their CPUs, but for the millionth time I am telling you that Intel can't do that:

http://overclocking.guide/the-truth-about-cpu-soldering/

If you think that the article is false, send an application form to Intel. They could definitely use your "expertise".

Now can you please stop shitting on Intel?

I didn't say they should solder it, I said delid + liquid metal compound would yield better temps. Read what I said.

 

EDIT: I'm also aware application is the issue but saying TIM is shorter and more people know what you are on about, the application method is consistently poor and therefore it basically makes the TIM poor (because the way it is applied makes it perform poorly so good TIM becomes poor in terms of performance)

[PCGuy_5960]

14 minutes ago, tom_w141 said:

therefore it basically makes the TIM poor (because the way it is applied makes it perform poorly so good TIM becomes poor in terms of performance)

No.... Let's say that you use liquid metal, if you don't apply it properly, you will run into issues with thermals. Does this mean that liquid metal is a bad TIM, or does this mean that the application method you used is bad?

[done12many2]

3 minutes ago, PCGuy_5960 said:

No.... Let's say that you use liquid metal, if you don't apply it properly, you will run into issues with thermals. Does this mean that liquid metal is a bad TIM, or does this mean that the application method you used is bad?

 

We don't even know if the application method resulting in higher temps was intentional or not.  

 

With Kaby Lake, the die is more delicate and the PCB is thinner then ever.  We don't know that the use of additional sealant wasn't an intentional measure in protecting the die or both.  The additional sealant definitely increases the gap between the die and the IHS so it wouldn't be far-fetched to believe that this was done intentionally by design.

 

After all, Intel has been making CPUs for a long time.  They may actually know how to make one or two properly.  

 

I guess if you say the same exact shit over and over again, more people will believe your version of the story based on your perceived expertise in the field.  

 

So at first Intel used poor quality (cheap) TIM to save money and increase profits.  Then it turns out that the TIM isn't actually that cheap at all and is of at least decent quality.  What is it now?  Are they intentionally overusing adhesive in excess to save money.  Nope, that logic doesn't work.  

 

Could it simply be as simple as their designers and engineers determined that it was necessary?

 

The great thing about the internet is everyone, to include myself is allowed to be a self-proclaimed expert.  We have so many experts now we don't know what to fucking believe anymore.  

[JoostinOnline]

I have a Cryorig R1 Ultimate and I can only get 4.6GHz out of it without reaching those temps.  It's a hot CPU.

[Majestic]

1 hour ago, Darth Revan said:

That processor is a 90 watt tdp while the cooler is a 95watt tdp.

If you overclock the processor it's going to go beyond those 90 watts by a lot.

So the cooler is the problem. It's not capable of cooling the processor.

You need a better one.

Depends on how fast it reaches that temperature. If it immediately spikes to 100C, it's not the cooler. It's the thermal resistance.

[stealth80]

What version of Prime95 are you using? There's a well known bug with Prime95 and Haswell, think it's version 28.5 + and is related to AVX

 

Run some other benchmarks

 

It used to get some pretty insane temps out of mine and I was on a custom loop

[tom_w141]

1 hour ago, PCGuy_5960 said:

No.... Let's say that you use liquid metal, if you don't apply it properly, you will run into issues with thermals. Does this mean that liquid metal is a bad TIM, or does this mean that the application method you used is bad?

If you take something good and apply it so badly that its ineffective then it might as well be bad because the result is the same as if it were bad. That is the point I was getting at.

[PCGuy_5960]

Just now, tom_w141 said:

If you take something good and apply it so badly that its ineffective then it might as well be bad because the result is the same as if it were bad. That is the point I was getting at.

This is what you said.....

1 hour ago, tom_w141 said:

EDIT: I'm also aware application is the issue but saying TIM is shorter and more people know what you are on about, the application method is consistently poor and therefore it basically makes the TIM poor (because the way it is applied makes it perform poorly so good TIM becomes poor in terms of performance)

I really don't get how a bad application method would make the TIM worse..... 

[tom_w141]

52 minutes ago, done12many2 said:

 

We don't even know if the application method resulting in higher temps was intentional or not.  

 

With Kaby Lake, the die is more delicate and the PCB is thinner then ever.  We don't know that the use of additional sealant wasn't an intentional measure in protecting the die or both.  The additional sealant definitely increases the gap between the die and the IHS so it wouldn't be far-fetched to believe that this was done intentionally by design.

 

After all, Intel has been making CPUs for a long time.  They may actually know how to make one or two properly.  

 

I guess if you say the same exact shit over and over again, more people will believe your version of the story based on your perceived expertise in the field.  

 

So at first Intel used poor quality (cheap) TIM to save money and increase profits.  Then it turns out that the TIM isn't actually that cheap at all and is of at least decent quality.  What is it now?  Are they intentionally overusing adhesive in excess to save money.  Nope, that logic doesn't work.  

 

Could it simply be as simple as their designers and engineers determined that it was necessary?

 

The great thing about the internet is everyone, to include myself is allowed to be a self-proclaimed expert.  We have so many experts now we don't know what to fucking believe anymore.  

I read somewhere they use this method for its longevity as the average mainstream user doesn't oc, which i'm fine with. This is not true for the kind of people who purchase X299 cpus though thats where the logic is a bit messed.

 

The fact oc'ing a modern Intel CPU to its full potential always involves a delid does kind of speak for itself.

 

Ofc those at Intel know how to make a processor but they also know how to make a processor within acceptable tolerances as profitably as possible. Just like any good business does. I doubt its a design necessity more like a cost calculation.

[tom_w141]

2 minutes ago, PCGuy_5960 said:

This is what you said.....

I really don't get how a bad application method would make the TIM worse..... 

You aren't getting it. i'm saying if you mess up the application then it is the same as if it were bad not that the substance actually is bad. Not applying it properly causes a dip in performance hence why people say the TIM is bad, the compound itself is good but if you use it poorly then its poor.

[done12many2]

4 minutes ago, stealth80 said:

What version of Prime95 are you using? There's a well known bug with Prime95 and Haswell, think it's version 28.5 + and is related to AVX

 

Run some other benchmarks

 

It used to get some pretty insane temps out of mine and I was on a custom loop

 

You can actually run any version of Prime95 up to the latest without AVX if you want.  All you have to do is find the local.txt file within the Prime95 folder and add "CpuSupportsAVX=0" without quotation marks to it.  Save, restart Prime95 and let her rip AVX free.  

[PCGuy_5960]

Just now, tom_w141 said:

Ofc those at Intel know how to make a processor but they also know how to make a processor within acceptable tolerances as profitably as possible. Just like any good business does. I doubt its a design necessity more like a cost calculation.

http://www.intel.com/content/www/us/en/jobs/hiring/apply-online.html

[tom_w141]

1 minute ago, done12many2 said:

 

You can actually run any version of Prime95 up to the latest without AVX if you want.  All you have to do is find the local.txt file within the Prime95 folder and add "CpuSupportsAVX=0" without quotation marks to it.  Save, restart Prime95 and let her rip AVX free.  

Off topic sorry OP but genuine question here, can you explain why prime95 is still so commonly used when it is known thats its overkill and a real world load won't ever cause this effect?

[PCGuy_5960]

2 minutes ago, tom_w141 said:

You aren't getting it. i'm saying if you mess up the application then it is the same as if it were bad not that the substance actually is bad. Not applying it properly causes a dip in performance hence why people say the TIM is bad, the compound itself is good but if you use it poorly then it performs poorly.

FIFY

[Ordinarily_Greater]

2 hours ago, Majestic said:

NH-U12S is not a weak cooler...

I realized, "4790K OC". It's overclocked

[tom_w141]

3 minutes ago, PCGuy_5960 said:

http://www.intel.com/content/www/us/en/jobs/hiring/apply-online.html

hilarious haha - this has nothing to do with cpus its business. You make a product then you find out how to achieve the same result as cheaply as possible while achieving the original spec. Its my job in fact (not CPUs just the business aspect)

[done12many2]

6 minutes ago, tom_w141 said:

I read somewhere they use this method for its longevity as the average mainstream user doesn't oc, which i'm fine with. This is not true for the kind of people who purchase X299 cpus though thats where the logic is a bit messed.

 

Yet you continue to post the same repetitive shit in mainstream Intel CPU related threads.  

 

Quote

The fact oc'ing a modern Intel CPU to its full potential always involves a delid does kind of speak for itself.

 

Ofc those at Intel know how to make a processor but they also know how to make a processor within acceptable tolerances as profitably as possible. Just like any good business does. I doubt its a design necessity more like a cost calculation.

 

As a stated in the post that you quoted, we're all experts.  One thing is for certain, you're great at entering conversations under the guise of help, but we can expect the the derogatory Intel comment will follow shortly after.