4790k Delid Summer 2014 Bash!!



On a serious note, I'm hoping de-lidding is not needed with this chip.
 
Not sure if real, but just found this..

zqLhm9E.jpg


From http://www.coolaler.com/showthread.php/315237-Boost-4-4GHz%EF%BC%81Intel-Core-i7-4790K%E3%80%81i5-4690K%E6%AD%A3%E5%BC%8F%E8%A6%8F%E6%A0%BC%E6%9B%9D%E5%85%89 found via http://chinese.vr-zone.com/112241/devils-canyon-with-core-i7-4790k-and-core-i5-4690k-05102014/

4GHz! Really hoping the 4790k will work on my Z87 board (TDP is 4W higher, which I think was one of the concerns behind the rumors of 4690k and 4790k being Z97-only).. otherwise they can keep it.
 
So the 4770 and 4770k both boost to 3.9ghz and the 4790 does 4.0 but the k will boost to 4.5ghz? Sounds weird. And unlikely.

Also pretty sure the rumors that DC will only work on z97 are not rumors.
 
Also pretty sure the rumors that DC will only work on z97 are not rumors.

It's possible, but might pose more of a mess than whatever it is they'd be trying to accomplish by locking out only those 2 chips from series-8 boards. All the manufacturers for these boards state "supports 4th Generation Intel Core processors", and Haswell refresh is still 4th-gen (and is now compatible with series-8 boards).

So are they going to try and convince the consumer that Devil's Canyon is 5th-gen Broadwell, or a 4.1-gen outside of Haswell? Maybe.. but doing this will really piss off many of their customers and cause a loss of sales, all for the sake of helping the mobo manufacturers push Z97 boards that really don't offer much in the way of improvement over Z87.

Seems incredibly dumb on Intel's part.. but we'll find out in a few weeks ;)
 
doing this will really piss off many of their customers and cause a loss of sales, all for the sake of helping the mobo manufacturers push Z97 boards that really don't offer much in the way of improvement over Z87.

Why? If z97 isn't much of an improvement, and Devil's Canyon still needs to be delidded, then why are people going to be any more mad than they usually are at Intel for being greedy? The point is DC has a different pin layout (almost certainly on purpose) and z87 doesn't appear to be compatible with it. Haswell refresh seems like it's still a Haswell, so it is "backwards compatible" in that z97 will accomodate it as well as z87. DC isn't backward's compatible (imo) and if you were Intel would you do it that way? Especially if z97 is hard to sell to begin with? It's hard to argue with this logic.
 
^^ Do Intel actually get a cut of Z97 boards sold?

If so, I can definitely see the logic in them being greedy by having a chip that would help sell Z97.

If not, then they're just throwing away sales from people that are satisfied with their Z87 board, but would still upgrade their CPU.. what would Intel gain by helping ASUS, MSI, GB, et al to push new boards?

I wonder if anyone has compared the pin layout from the new Z97 boards vs Z87 yet, and if you can tell a difference that would actually confirm this (the pin change as seen in the pics of the lady holding the DC chip)?
 
^^ Do Intel actually get a cut of Z97 boards sold?

I wonder if anyone has compared the pin layout from the new Z97 boards vs Z87 yet, and if you can tell a difference that would actually confirm this (the pin change as seen in the pics of the lady holding the DC chip)?

Does Intel make money by selling their motherboard chipsets? :rolleyes: Of course.

Has anyone compared the layout of the motherboard PINS? They look like metal peach fuzz... lol. Comparing the chips would make a lot more sense... but we can't until they are released. We know what DC looks like, and its pins are different from Haswell. I haven't seen any pictures of the Haswell refresh SKU pin layout, but I haven't looked hard because I just assume I'm right. But I might be wrong. Usually when I say "I might be wrong" I'm not, though.
 
Last edited:
Does Intel make money by selling their motherboard chipsets? :rolleyes: Of course.

Lol, nice. Was an honest question.. just something that never once crossed my mind before.

Even if that's true, I'm guessing Intel makes more money selling chips than they do licensing their chipsets, so it would still make sense for them to sell customers multiple chips for a single board purchase, rather than losing sales if an upgrade requires a whole new board and customers say "fuck it, I'll just stick with what I've got or wait for Broadwell to upgrade".

..but I haven't looked hard because I just assume I'm right. But I might be wrong. Usually when I say "I might be wrong" I'm not, though.

And even when your wrong, you're still right? :D

Dunno, but I sure as shit wouldn't put my money on it either way. Intel has stated multiple times that Devil's Canyon is supported by series-9, but after watching the unveil video again, they never once said it wasn't also supported by series-8 (one of the guys in the press during Q&A said it wasn't, though). Anyway, don't worry.. only a few more weeks and you can say "I told you so" :p
 
So the 4770 and 4770k both boost to 3.9ghz and the 4790 does 4.0 but the k will boost to 4.5ghz? Sounds weird. And unlikely.

Also pretty sure the rumors that DC will only work on z97 are not rumors.

I heard haswell refresh cpu's will in fact work with z87, after a bios flash. It's broadwell that will only work with z97, unless I'm mistaken.
 
I heard haswell refresh cpu's will in fact work with z87, after a bios flash. It's broadwell that will only work with z97, unless I'm mistaken.

Yes, you are correct on both counts.

It was only these 2 Devil's Canyon chips (the 4790k & 4690k) that supposedly wouldn't be backward-compatible with Z87.

This conjecture is based on a few details from Intel's DC unveil at GDC in March.. slides stating that unlocked Haswell refresh and Devil's Canyon is "targeting compatibility" or "supported" by series-9 chipsets (though Intel has yet to specifically state that they will be incompatible with Z87); photos of the chip that show a different pin layout as compared to other Haswell chips; and additional speculation of power delivery or microcode changes that would block DC from working with Z87.

For example, this was posted by WCCFTech today..
The Z97 will be the only board, according to Intel, to support the upcoming faster Haswell versions – Devil’s Canyon. Enthusiasts and users alike looking forward towards Devil’s Canyon will therefore have to choose the Z97

But Intel has never once explicitly stated such a thing. Might very well turn out to be true, but they haven't actually said so yet.
 
I have read on various forums that one of the reasons for the higher overclocks from older Haswell K chips is that with Z97, the voltage regulation has been moved back the the chip set from the processor (was moved from the chip set to the die on Haswell vs Ivy Bridge I believe). If Devils Canyon does not have the voltage regulation on die, that in and of itself may allow for much higher overclocks but would make it incompatible with Z87 since it does not handle voltage regulation.. Also Z97 may be disabling on die voltage regulation and is being handled by the chip set again allowing for much higher overclocks from current Current Haswell K chips. If this is true then there a BIG reason to upgrade boards :D Going from 4.5 to 4.9 just by changing boards? :eek: Good times may be ahead.
 
I have read on various forums that one of the reasons for the higher overclocks from older Haswell K chips is that with Z97, the voltage regulation has been moved back the the chip set from the processor (was moved from the chip set to the die on Haswell vs Ivy Bridge I believe). If Devils Canyon does not have the voltage regulation on die, that in and of itself may allow for much higher overclocks but would make it incompatible with Z87 since it does not handle voltage regulation.. Also Z97 may be disabling on die voltage regulation and is being handled by the chip set again allowing for much higher overclocks...


First that I have heard of this. If true then it will be easier to delid as well and it will probably run very cool delidded or not. Very interesting.
 
First that I have heard of this. If true then it will be easier to delid as well and it will probably run very cool delidded or not. Very interesting.

If Devils Canyon has better TIM and a soldered heat spreader ALA Sandy Bridge, then you won't need to delid.
 
If Devils Canyon has better TIM and a soldered heat spreader ALA Sandy Bridge, then you won't need to delid.

Quality of TIM on Haswel and IvyBridge has pretty much been proving to not be the cause of high temps, the problem was the contact quality between die and bottom of IHS. Changing VRM control design from CPU to chipset makes more sense for such a huge performance gain on Devils Canyon and cooler operation.


If Z97 chipset controls voltage in CPU perhaps it can bypass the integrated VRM on Haswel for higher clocks and lower temperature..... hhhhhmmmmm



In either case bare die access for cooling is best situation for direct bare die cooling and I think Devils Canyon will be a go for delidding.
 
If Z97 chipset controls voltage in CPU perhaps it can bypass the integrated VRM on Haswel for higher clocks and lower temperature..... hhhhhmmmmm

This was what I was hoping for. I've ordered a z97 Hero to see if I can get more out of my chip or not in the new platform.
 
This was what I was hoping for. I've ordered a z97 Hero to see if I can get more out of my chip or not in the new platform.

Just speculation. Don't set yourself up for disappointment. Bypassing the Has well integrated VRM should help but don't know the truth of the operation yet.
 
Just speculation. Don't set yourself up for disappointment. Bypassing the Has well integrated VRM should help but don't know the truth of the operation yet.

Either way is ok with me but I'll take some more performance if it's available. I'm more hoping the problems with adaptive voltage will be solved so I don't have to argue with anyone about getting manual to use speedstep and drop voltage at idle anymore. Would be nice if adaptive mode didn't decide to raise your VID above your set voltage. I've seen my chip ask for 1.36v instead of 1.27 like it's supposed to in adaptive and all I was running was Unigine Valley. That's not working as intended.
 
Just speculation. Don't set yourself up for disappointment.

Since some of the reviews for Z97 are already out, wouldn't we have heard more about this by now?

I was checking through some of the ASUS Z97 manuals, and there are still FIVR options, but no mention of disabling it entirely for on-board voltage regulation.
 
Since some of the reviews for Z97 are already out, wouldn't we have heard more about this by now?

I was checking through some of the ASUS Z97 manuals, and there are still FIVR options, but no mention of disabling it entirely for on-board voltage regulation.

No idea. I think you're probably right. Would have had to have the options built into z87Haswell chips, no? I'm not seeing anyone say they got a bigger overclock with z97 yet, but I'm not ruling it out, nor am I ruling out the possibility of disabling the FIVR.

But I did see someone say they had to bump their 4770k voltage higher on the z97 than z87, no clue if that's true, but it sounds more characteristically pessimistic/reality. I'm slightly embarrassed to say I have my hopes up more for the fan-controller software ASUS is bundling. The M6Hero's fan controller crap isn't very good and it doesn't have quite enough fan headers for me. The M7H has one more.
 
Interesting, so this is a K chip that support VT-d.
That will be a first, finally you can have your cake and eat it too.
 
Why Delid? Because it will give you the edge in gaming, number crunching, over clocking, ect.

If you use computers to test your ability to push them to the limits then deliding will be your pass to more performance.
 
Why Delid? Because it will give you the edge in gaming, number crunching, over clocking, ect.

If you use computers to test your ability to push them to the limits then deliding will be your pass to more performance.

What if we find the IHS soldered?
 

really people have spent 2~years crying about TIM on the HS and people are going to make pouty face at the prospect of soldered HS...

sigh... I would more than welcome the return to soldered heatspreaders ..as I'd rather not have to void my warranty to get acceptable heat transfer from my cpu.
 
really people have spent 2~years crying about TIM on the HS and people are going to make pouty face at the prospect of soldered HS...

sigh... I would more than welcome the return to soldered heatspreaders ..as I'd rather not have to void my warranty to get acceptable heat transfer from my cpu.

I would too, but like everyone who learned how to delid we sort of almost lament the idea of the practice dying out, because it's an exciting process and not everyone is up for it. So it adds this like "extreme" side to the enthusiast mindset. It's like the original movie "Crash". Sometimes people get into car accidents and they want more. The 4790k Summer Bash must commence.
 
Too "extreme" for me. I nearly pooped my pants when I saw the picture of them taking a flame to it.
 
the link shows the 4790k having vt-d, if so i'll build a few boxes around that!
 
Too "extreme" for me. I nearly pooped my pants when I saw the picture of them taking a flame to it.

No doubt. Just like a junkie cooking up his drugs in a spoon before his next fix.. that vise & razor shit just wasn't [H] enough :eek: :D
 
I see nothing about board requirements. Only Haswell and 1150 socket. So does this confirm it will work on 87 boards?
 
even if it is soldered these guys will still lap the IHS FOR 2 DEGREES
 
I see nothing about board requirements. Only Haswell and 1150 socket. So does this confirm it will work on 87 boards?

The rumored Z97 requirement is only for the unlocked "K" versions that wont come out until June.
 
really people have spent 2~years crying about TIM on the HS and people are going to make pouty face at the prospect of soldered HS...

sigh... I would more than welcome the return to soldered heatspreaders ..as I'd rather not have to void my warranty to get acceptable heat transfer from my cpu.
They're still not optimal believe it or not. I blowtorched my lapped Q6600 and took that SOB off and it dropped temps even further and allowed for slightly higher clocks with lower vcore and VTT.
 
I thought that the reason why the Haswell refresh CPUs will not work on Z87 motherboards is because the new CPUs (which will eventually include Devil's Canyon as well) have moved the voltage regulator off of the CPU. If the voltage regulator is not on the CPU and Z87 motherboards were designed without their own voltage regulator because they expected the CPU to handle that job, then how will the voltage going to the CPU be controlled?
 
Back
Top