Intel's responses to AMD Ryzen with i7 7740K and i5 7640K

[DELTAprime]

Well I'd be interested in the 7740k if the thermal issues of the 7700k get fixed in the new chip, otherwise I'll hold out for either a Devil's Canyon style refresh or Coffee Lake since it's supposed to use the same socket. While I am CPU limited on my 6600k in two games, I can live with that till Intel fixes their damn thermals.

[JoostinOnline]

2 minutes ago, DELTAprime said:

Well I'd be interested in the 7740k if the thermal issues of the 7700k get fixed in the new chip, otherwise I'll hold out for either a Devil's Canyon style refresh or Coffee Lake since it's supposed to use the same socket. While I am CPU limited on my 6600k in two games, I can live with that till Intel fixes their damn thermals.

Since they're just using a shitty TIN (I think), shouldn't just using better quality fix the issue?  Sure, it sucks for people who ordered early, but better to fix it for most people than for nobody.

[iamdarkyoshi]

11 hours ago, Deli said:

For the 1001 th time. The TIM that Intel uses is good stuff. If you can find some better thermal paste that can withstand as much heat cycles and years of use, you win.

http://a.co/8OpgR8p

[JoostinOnline]

11 hours ago, Deli said:

For the 1001 th time. The TIM that Intel uses is good stuff. If you can find some better thermal paste that can withstand as much heat cycles and years of use, you win.

It appears that they went the cheaper route on Kaby Lake though.  Even if it's not the TIM, CPU's with completely identical specs to Skylake run hotter on Kaby Lake.  Something is affecting the temperature.

[TheRandomness]

35 minutes ago, deviant88 said:

Just FYI, all intel cpu's that are from same slice of wafer or binned, i7 non extreme, i5 ,i3 , pentiums and maybe even celerons, can run hyperthreading if intel decided to switch it on inside the cpu. The only reason they dont is either the chip is broken in that regard or most commonly intel has no competition so they turn it off and sell as dual/quad core without HT for less performance to convince people to buy i7's with HT, just because they can.

I knew that, it's just Intel must actually be worrying about AMD and Ryzen to enable it on an i5

[Agosto]

3 minutes ago, JoostinOnline said:

It appears that they went the cheaper route on Kaby Lake though.  Even if it's not the TIM, CPU's with completely identical specs to Skylake run hotter on Kaby Lake.  Something is affecting the temperature.

Higher stock frequencies mostly.

The issue is die-IHS distance, not the thermal conductivity of the TIM itself

[Lays]

2 minutes ago, JoostinOnline said:

It appears that they went the cheaper route on Kaby Lake though.  Even if it's not the TIM, CPU's with completely identical specs to Skylake run hotter on Kaby Lake.  Something is affecting the temperature.

The same damn difference is there whenever you delid any chip from ivy forward. 20-25c drops are NOTHING NEW from delid + CLU.

[DELTAprime]

1 minute ago, JoostinOnline said:

It appears that they went the cheaper route on Kaby Lake though.  Even if it's not the TIM, CPU's with completely identical specs to Skylake run hotter on Kaby Lake.  Something is affecting the temperature.

I haven't gotten time to read it myself but I heard that HardOCP's article the other week on deliding Kaby Lake suggested there may simply be too large of a gap between the CPU die and the IHS.

[JoostinOnline]

1 minute ago, Agost said:

Higher stock frequencies mostly.

The issue is die.IHS distance, not the thermal conductivity of the TIM itself

 

1 minute ago, Lays said:

The same damn difference is there whenever you delid any chip from ivy forward. 20-25c drops are NOTHING NEW from delid + CLU.

When you compare the i3-6100 and G4620, which are completely identical except for architecture, the Kaby Lake G4620 runs hotter.

[Agosto]

4 minutes ago, JoostinOnline said:

 

When you compare the i3-6100 and G4620, which are completely identical except for architecture, the Kaby Lake G4620 runs hotter.

Can you link me a temperature test? Same ambient temperature and cooler?

However, again, the distance between the die and the IHS makes a lot of difference. Ask the good old monsieur Fourier

[JoostinOnline]

9 minutes ago, Agost said:

Can you link me a temperature test? Same ambient temperature and cooler?

However, again, the distance between the die and the IHS makes a lot of difference. Ask the good old monsieur Fourier

Well this is f*cking annoying.  I can't find the video that shows the temperatures between the two.  I am 100% sure I've seen one, because I immediately noticed the difference. Both framerate and temperatures were slightly higher on the G4620.  I do understand you wanting more than my word on it.  If I find it I'll make sure to post it.

[Agosto]

2 minutes ago, JoostinOnline said:

Well this is f*cking annoying.  I can't find the video that shows the temperatures between the two.  I am 100% sure I've seen one, because I immediately noticed the difference. Both framerate and temperatures were slightly higher on the G4620.  I do understand you wanting more than my word on it.  If I find it I'll make sure to post it.

I'm not accusing you of lying, but the variables are a lot in these cases

[JoostinOnline]

1 minute ago, Agost said:

I'm not accusing you of lying, but the variables are a lot in these cases

I know, I wasn't offended or anything.  I'd want to see it for myself if I were you though, so I understand. 

[Teddy07]

12 hours ago, HKZeroFive said:

I'm all for some competition but introducing new CPUs such as the i7 7740K and the i5 7640K would...

 

a) Undermine the purchases of recent customers (especially the people who just bought Kaby Lake).

Yeah i agree. I would totally be ouraged.

[MrDynamicMan]

9 hours ago, Carclis said:

By not cracking the silicon die? GPU manufacturers don't demand any proof and still ship their products giving full access to the bare die.

Yes, but when was the last time everyone who was building a PC had to strip the gpu and put new fins/water block on? 

[Carclis]

13 minutes ago, MrDynamicMan said:

Yes, but when was the last time everyone who was building a PC had to strip the gpu and put new fins/water block on? 

It's not just for people who plan to water cool. I took apart my 290x previously because it had a decent build-up of dust which wasn't accessible while the cooler was left on. I also noticed that the application of TIM wasn't the best. All up the process brought down temperatures by about 5C.

[The Benjamins]

8 minutes ago, Carclis said:

It's not just for people who plan to water cool. I took apart my 290x previously because it had a decent build-up of dust which wasn't accessible while the cooler was left on. I also noticed that the application of TIM wasn't the best. All up the process brought down temperatures by about 5C.

What he is saying is during a normal initial build of a PC the GPU die is NOT exposed when a CPU would be so the heat spreader on a CPU is their to protect the CPU from damage during installation. 

[Carclis]

23 minutes ago, The Benjamins said:

What he is saying is during a normal initial build of a PC the GPU die is NOT exposed when a CPU would be so the heat spreader on a CPU is their to protect the CPU from damage during installation. 

And I never once said that there should not be an IHS. Just that it shouldn't be installed in such a way that makes it so easy to damage the die in the removal process in addition to completely voiding the warranty.

[vorticalbox]

3 hours ago, deviant88 said:

Just FYI, all intel cpu's that are from same slice of wafer or binned, i7 non extreme, i5 ,i3 , pentiums and maybe even celerons, can run hyperthreading if intel decided to switch it on inside the cpu. The only reason they dont is either the chip is broken in that regard or most commonly intel has no competition so they turn it off and sell as dual/quad core without HT for less performance to convince people to buy i7's with HT, just because they can.

not true the features were deemed unstable or out right broken so then disabled them and drop them as a lower chip.

 

I remember and used to have an "unlock cores" feature that would allow you to reenable the cores at the risk of them being unstable.

[NumLock21]

9 hours ago, leadeater said:

I think his point was more around this won't happen now days.

 

 

Edit: Video removed, Google AMD Duron Explode to see it.

No, he thinks there is no tim on the die, so his idea was to delid the ihs and apply tim by himself and intel should not avoid his warranty for doing it.

 

 

 

 

[Ashiella]

4c 8t isn't enough for me, I said I won't upgrade until their consumer i7 has 6 cores. If the 8700k has 6c 12t I'll upgrade. I look at the gpu market and everything is getting upgraded so fast and by so much. And then Intel releases either 5% higher IPC or 5% higher clock.

[Carclis]

2 hours ago, NumLock21 said:

No, he thinks there is no tim on the die, so his idea was to delid the ihs and apply tim by himself and intel should not avoid his warranty for doing it.

I made it pretty clear that I knew that there was paste between the IHS and die when I said that the stock compound was low performance with a super long lifespan, obviously so that the consumer would not have to touch it (funny how it works the other way around huh?). If simply changing the paste results in anything over a 10C drop, something is very wrong and this is the case with Kaby Lake where 20C is quite typical.

[michaelocarroll007]

17 hours ago, Scruffy90 said:

Than what would the atoms and celerons be?

also have core m- 3 5 7 & Y lol

[Scruffy90]

11 minutes ago, michaelocarroll007 said:

also have core m- 3 5 7 & Y lol

Completely forgot they existed

[MrDynamicMan]

5 hours ago, Carclis said:

It's not just for people who plan to water cool. I took apart my 290x previously because it had a decent build-up of dust which wasn't accessible while the cooler was left on. I also noticed that the application of TIM wasn't the best. All up the process brought down temperatures by about 5C.

And what happens to your warranty when you remove the cooler?