4790k Delid Summer 2014 Bash!!

Zangmonkey said:
It's always been the case that *some* Z87 boards will be able to run devil's canyon
Click to expand...

almost makes you wish intel would just change sockets again... what a confusing mess :D


so some Z87's and some Non K devil's canyon cpu's ... ok my head hurts now...
 
atarione said:
almost makes you wish intel would just change sockets again... what a confusing mess :D


so some Z87's and some Non K devil's canyon cpu's ... ok my head hurts now...
Click to expand...

It's no different than amd, when they released bulldozer there were boards not up to snuff that couldn't handle it.
 
Does anyone know if the non K refresh chips have the VRM on or off dye? I just skimmed over the thread so my apologies if I missed it. If they are still on chip that would explain the compatibility with z87. Possible only the refresh DC Ks have the VRM moved off which is why they will only work on z97.
 
atarione said:
almost makes you wish intel would just change sockets again... what a confusing mess :D


so some Z87's and some Non K devil's canyon cpu's ... ok my head hurts now...
Click to expand...

Intel doesn't claim Z87 supports Devil's Canyon.

Motherboard manufacturers are on the hook for that one.
 
Goomz said:
Does anyone know if the non K refresh chips have the VRM on or off dye? I just skimmed over the thread so my apologies if I missed it. If they are still on chip that would explain the compatibility with z87. Possible only the refresh DC Ks have the VRM moved off which is why they will only work on z97.
Click to expand...

The voltage regulator is still on die for both Haswell Refresh and Devils Canyon.
Moving the voltage regulator back onto the board would require a significant redesign of the chip.
Haswell refresh is using exactly the same die, package and TIM as original Haswell, just clocked 100MHz higher.

Devil's Canyon is also using exactly the same die (maybe a newer stepping, we don't know) just binned for much higher clocks.
The difference is that DC has a redesigned package (the PCB the die sits on, includes components on the underside of the CPU) and better TIM.

I'm no expert but from the information currently available, the DC incompatibility with z87 (if it is indeed incompatible) appears to be entirely artificial.
 
Something about this whole "we're asking you to get another socket just because" to run this refresh seems like foul play to me (along with the better OCing and whatnot). ie both intel and the MB manufacturers colluding and getting seriously money hungry. I mean it's probably been the case before as well, but this just seems to go past what they've done before.
 
This is a REFRESH, not a REDESIGN. The VRMs will still be onboard the chip, it's still the same architecture.

The move of the VRMs back to the motherboard will happen with Broadwell, that'll be its selling point for enthusiasts.

Also, the rumours about 4.4ghz boost on the K part are most likely bullshit. At BEST maybe it can run a single core at 4.4ghz dynamically under certain conditions. But given how much of a hit AVX-2 has, and that intel has specific code to boost voltage under AVX-2 usage, I seriously doubt it.

Intel are keeping tight lipped about it all and are just letting the rumour mill do all their PR for them.

I haven't seen a real trusted source confirm any of the latest rumours, and I've NEVER known an Intel K part to have different base clocks to it's non K counter part.

Never.
 
Last edited:
spine said:
I haven't seen a real trusted source confirm any of the latest rumours, and I've NEVER known an Intel K part to have different base clocks to it's non K counter part.
Click to expand...

Yeah, the new 4790k info could very well be fake, but even outlets like Forbes have been reporting it now. So trusted news source reporting rumor = still rumor :p

Though, with all the hype and extra wait since GDC, a 4790k at 3.6GHz will be incredibly disappointing.
 
d50man said:
even if it is soldered these guys will still lap the IHS FOR 2 DEGREES
Click to expand...

It isn't just about dropping a couple degrees. It is also about evening out the temps between cores.

Very uneven IHSes can cause hot spots.

Just because only a couple degrees is dropped from the temp sensor doesn't mean that a huge temp drop wasn't gained in a small place where the IHS was not getting good contact with the heatsink.
 
spine said:
This is a REFRESH, not a REDESIGN. The VRMs will still be onboard the chip, it's still the same architecture.

The move of the VRMs back to the motherboard will happen with Broadwell, that'll be its selling point for enthusiasts.

Also, the rumours about 4.4ghz boost on the K part are most likely bullshit. At BEST maybe it can run a single core at 4.4ghz dynamically under certain conditions. But given how much of a hit AVX-2 has, and that intel has specific code to boost voltage under AVX-2 usage, I seriously doubt it.

Intel are keeping tight lipped about it all and are just letting the rumour mill do all their PR for them.

I haven't seen a real trusted source confirm any of the latest rumours, and I've NEVER known an Intel K part to have different base clocks to it's non K counter part.

Never.
Click to expand...

Never before, but Intel seems pretty clear they have made changes to provide more value to enthusiasts, so a 4.0ghz stock/4.4 turbo would fit in with Intel's mention of a 'nice performance pop' for devils canyon.
Graff has adjusted Intel's silicon roadmap in order to address that problem, and today, she revealed some key aspects of Intel's desktop product plans for 2014. Much of the plan seems like it could have been ripped from the pages of our CPU reviews; it addresses a number of our beefs with Intel's product decisions of late.

The most exciting of those products may be the one based on current Haswell silicon. Intel desktop processors since Ivy Bridge have had limited overclocking headroom due to excessive heat, and over time, enthusiasts have pinned much of the blame on the combination of packaging and thermal interface material (TIM) used in newer CPUs. Folks have even taken to de-lidding their brand-new processors in order to recover some of the clock speed headroom. Intel will address that problem head-on with a new unlocked Haswell part code-named "Devil's Canyon," coming in mid-2014. Devil's Canyon will have redesigned packaging and an improved TIM meant to increase overclocking potential.

Graff said this product should offer a "very nice performance pop," so I'm hoping its default clock speeds will be quite a bit higher than existing chips, too.
Click to expand...
http://techreport.com/review/26189/intel-to-renew-commitment-to-desktop-pcs-with-a-slew-of-new-cpus
 
Yeah, I wanna believe it, but when I read 'improved packaging and thermal TIM' I just think of the lowest that that could imply, and basically it's addressing the 2 faults of haswell:

1. Crappy TIM, but actually more critically...
2. Crappy hardened IHS glue that impedes CPU <-> IHS contact.

So 'improved packaging and thermal tim' could simply mean they're using a better TIM and soft glue. Or, at best in this scenario, soldered IHS.

I very much doubt it means they've changed the actual cpu silicon to improve the on-die VRMs. That would be very impressive, but I can't see them doing that for 1 or 2 cpu's, in the midrange, during a refresh. Yes, Z97 is 'midrange' in Intel's eyes.


4.4ghz boost as standard just seems unlikely given the overclocking results of the wider community, unless they're cherry picking. Which would be awesome, and I'm hopeful, but I know what I'm actually expecting...
 
In preparation for the *Bash* I took out my z87 board and cleaned up my 4770k. New z97 Hero works great, it's a "stable" version of the m6H, if you follow me. A lot of the little quirks seem gone, like the triple/double boot problem. It will still double boot if it needs to verify voltages from a UEFI change, but you'll know what I mean if you have had a triple boot on the M6hero it's just not right, this guy double boots slower and if it gets into a boot loop with fast boot enabled it will give you the chance to hit the bios button after 2 cold boots. Very helpful. I have literally not had to reset the bios yet and I've been messing about with voltages plenty and causing windows to not be able to load etc. It's just a nice stable seeming board, imo.

Anyway, on to the pertinent subject:
I completely cleaned the Liquid Pro off of my 4770K. It got a little dicey at one point and I became worried that I was scratching the die. Actually, I did slightly scratch the die, with a micro fiber cloth that must have had a piece of rough dust on it or something, but it's a superficial and tiny scratch and apparently did no damage. I was pretty rough with the thing to be honest, because I couldn't figure out how to get the CLP off at first, then I tried some of that ArtiClean Thermal Material remover and it worked like a charm--the CLP just turned into dirt and you could wipe it off then.

I actually got some of the CLP on the liquid tape I had put on the VRMs or whatever they are on the side, so I pulled all the tape off and reapplied it. Still ended up with CLP on the liquid tape I reapplied so it looked like crap but then I put the stupid heatspreader on top anyway. Was too tired to redo the whole process over and it fired up just fine anyway.

Then, as a final scare, I somehow managed to plug in the pump of my cooler incorrectly and it didn't function so I was trying to figure out how I was getting a CPU overheating warning from the UEFI almost immediately. All is well, the 4770k seems pretty robust. I've smacked it with a piece of wood in the wrong place accidentally, I've rubbed liquid metal all over it, I've tried to melt it. Still running strong.

I haven't had a chance to test a higher overclock on the z97 board with my 4770k yet, but I'll get to it, but I can tell you that it's perfectly stable at exact same voltages for 4.6Ghz.

BTW: I think I have seen the first symptom of "chip degradation" on my 4770k, which is not very old. 2 or 3 days ago my system started doing a triple cold boot thing on wakeup from deep sleep mode. Id get the Asus logo after 3 cold boots upon attempting to wakeup from deep sleep and then straight back to desktop with everything like it was (applications still up and working) with no errors in log. It had always done hybrid hibernate perfectly until this and it was pretty random how it came up. I decided to do a windows overhaul to test and have now, with the addition of the new board narrowed the cause down to (duh) VCCIN or CPU input voltage. I always ran it on auto, and messing with it never helped me solve the double/triple cold boot issue with the M6H, which at least using my power supply, the M6H will do if mains are cut to the board and all lights go off. Next time you fire it up, it will triple cold boot. I think after a recent BIOS update mine had gone to double cold boot instead. Still weird. Well, the M7H doesn't do this, but it will still triple cold boot out of deep sleep if eventual CPU voltage is left on AUTO (with myOverclock engaged) just like the M6H did. I haven't figured out how low I can go yet, but to get it to stop doing it, I changed the eventual CPU input voltage to 1.9v which is a big bump from where it wants to run on auto. It worked before though. And now it doesn't. So is that the chip dying? Or did a windows 8.1 update cause this? That was the ONLY change to my system when this started happening, and obviously I'm all up to date and it's still doing it unless I up that voltage setting. Not like 1.9v is that high really but I'm just curious. Sorry, end /wallotext
 
Last edited:
^ Good news, I'd like to hear back from you if you decide to try to push it higher. Also about your 4770k.... will it blend? ;) heh heh heh... .
 
Wendig0 said:
will it blend? ;) heh heh heh... .
Click to expand...

Indeed ^_^
I doubt it is prime95 28.5 stable at 4.7Ghz because of change in platform, though. I'd be surprised. I tried to get 4.7 stable on the M6H and it simply wasn't. 4.6 is solid, the only funny thing I've seen running it at 4.6 for a few months is this thing with the eventual Cpu input voltage and wake from sleep. Other than that no issues. It boots and runs prime95 on auto eventual cpu input voltage. It also couldn't do 2666 properly although it would post, the one time I tried a set of modules rated for that speed, I had to run them at 2400 to get a decent benchmark out of them. But I really don't see myself trying any other RAM unless they get cheaper.

Edit: It looks to me like z97 runs chips hotter at idle by about 5 degrees (not so much at full load) or this board just take temps differently, or maybe I just need to reseat. But the board runs cooler than the M6Hero, and the VRMs run cool (and are monitored). I'm still waiting to up the voltage. I'd use an air cooler ideally, so I'm not going to do any long term stability testing. When I got this chip I was using an NH-D14 and it was sketchy because I didn't have it delidded, but I got it validated using prime95 28.1. Now, even with the delid the H50s are sketchy with prime95 even though they don't get anywhere near as hot. I'm not worried for the chip, I'm worried I'll cause a blowout of the cooler, because I've done it. One oddity is that, although I did a really careful job lapping my surfaces for this seating, I didn't expect to get my cores all so close together in temp, like 3-4 degreees between them. So maybe the temps sensors on this board is different? Wouldn't the socket be exactly the same?
 
Last edited:
I have to admit, it's very much looking like 4.4ghz is for real. Awesome.

Here's to hoping it'll work on Z87. I'm not prepared to buy an almost identical mobo just for this chip...
 
spine said:
I have to admit, it's very much looking like 4.4ghz is for real. Awesome.

Here's to hoping it'll work on Z87. I'm not prepared to buy an almost identical mobo just for this chip...
Click to expand...

I wanted to get a z97 board anyway and I'd be happy if both your hopes come true but I really don't think either of them will. I'm also not sure I'm sold on this board yet.
 
Bluesun311 said:
I wanted to get a z97 board anyway and I'd be happy if both your hopes come true but I really don't think either of them will. I'm also not sure I'm sold on this board yet.
Click to expand...

I think you mentioned wanting to get a new Z97 board, to have the new ASUS fan controls? I was considering the same, until I found out that their stupid 60% minimum duty cycle restrictions for chassis fans are still in place in the BIOS. They did add a new "fan stop" feature, though.

No way in hell I'm paying them more money, when the only additional feature I really want still hasn't been implemented correctly (and should actually be corrected on their Z87 boards as a BIOS revision).

If the 4790k actually is 4.0/4.4GHz, and Intel makes it Z87 compatible at a reasonable price, I'll probably upgrade my 4770k at some point.. but I'm definitely not "upgrading'' to a Z97 board, though.
 
AFD said:
I think you mentioned wanting to get a new Z97 board, to have the new ASUS fan controls? I was considering the same, until I found out that their stupid 60% minimum duty cycle restrictions for chassis fans are still in place in the BIOS.
Click to expand...
This is not the case. The z87 hero BIOS fan functions had the limitation. The z97 Hero has a fan auto tune function that is insanely good. It goes WAY below %60. It even has a Super quiet version. The custom fan ramps can go as low as the fans will allow. Thus, the stop function. My cpu actually gets kinda hot it wants to run my fans almost off. It also has a probe header so you can ramp the fans off the GPU temp.

This is all really sweet and everything, but I think I'm gonna drop my z87 board back in my machine. I've reseated the cpu twice now and it seems to me the z97hero just plain runs my chip hotter than the M6H. That's at both idle and full bore. Coupled with the fact the z87 trick for manual voltage mode with undervolting doesn't work with z97 I'm just kinda underwhelmed by the board. If they decide to drop a 5Ghz turbo 4790k (they wont) I'll get a z97 board again but unfortunately this seems like 2 steps forward and 1 step back. Lost 2 USB ports I was using, too.
 
Last edited:
Bluesun311 said:
This is not the case. The z87 hero BIOS fan functions had the limitation. The z97 Hero has a fan auto tune function that is insanely good. It goes WAY below %60. It even has a Super quiet version. The custom fan ramps can go as low as the fans will allow. Thus, the stop function.
Click to expand...

Hmm, weird. I was comparing the BIOS fan controls of their Z87 ITX Deluxe/Pro & ROG vs their new Z97 ITX Plus (no better version available yet), and 60% was still the minimum duty on chassis fans (according to the manual), other than fan stop.

Then again, that board doesn't feature a super quiet profile.. so maybe just a feature difference between their low-end and high-end boards?

In any case, this is still something that customers have asked for, and should have been fixed on Z87 as a BIOS update.
 
Excellent news!

#1 the manual mode voltage drop "trick" works on z97. Also I've figured out how to make it work. I always used this setting out of habit but forgot the habit recently:

C-states need to be enabled not auto.
Package c-state support needs to be enabled not auto.

Pretty sure that's correct. Should allow manual voltage mode to drop down nicely at idle. On z97 it's hard to tell because a lot of monitoring software isn't updated and z97 has a metric shit ton of voltage meters. In ai suite 3 and other software it's clear it is working and what's more it runs cooler too. But once you have a full load on the system it will hit your manual voltage and stay there.

Doesn't really seem unintended to me at all. Raja says it's an exploit but I think intel meant for that to be there. To me it just confirms that adaptive and auto settings are useless when you know the z87 platform and it's uefi well.

#2 losing half your RAM channels after a delid isn't necessarily permanent. I just recovered mine after I switched motherboards and redid my CLP and the board would not post (error 55) if a stick of ram was in either of the first 2 slots. I was pretty bummed but I stayed calm and decided to take a good look at the chip because I did not know how it had been harmed. I hadn't had any awkward moments. By cleaning the bottom of my chip with thermal material remover I was able to remove some specs of either CLP or NT-H1 TIM that had somehow gotten on the chip's underside in several spots. Popped it back in after wiping it down with a good micro fiiber cloth with that ArctiClean TIM remover and fired it up--both channels back in action. Huge sigh of relief. Many problems solved this week. Most of them stupid mistakes.
 
Last edited:
22 of May and I think Intel is behaving way to slow about its new 4970k which looks like a 5ghz beast for those that do exotic water cooling.
 
hendermd said:
Did a google on BX80646I74790K and found some sites taking preorders for CORE I7-4790K FC-LGA12C 4.1GHZ 8MB BOX price range 353-410 dollars.
Click to expand...

Hmm... could be one of those "up to 4.1GHz" deals?

From ShopBLT (comparing 4770k and 4790k SKUs)..

(CM8064601464206) CORE I7-4770K FC-LGA12C 3.9GHZ 8MB 22NM 84W TRAY
(BX80646I74770K) CORE I7-4770K LGA1150 3.5G 8MB BOX
(CM8064601560016) CORE I7-4790K FC-LGA12C 4.1G 8MB TRAY
(BX80646I74790K) CORE I7-4790K FC-LGA12C 4.1GHZ 8MB BOX
Click to expand...

Really don't see it being a 4.1/4.5GHz chip for that price.. probably 3.7/4.1GHz. If so, they can keep it.. that and their super-special non-conflict TIM sauce :D
 
Oddly, after checking out all the Devil's Canyon SKUs that are now up for pre-order, I stumbled upon an article written this morning saying that DC at Computex is now rumored to be a paper launch only, with market availability as late as September.

http://www.expreview.com/33570.html

Hopefully complete BS, with Broadwell expected soon after.

Or, maybe Intel's plans for DC all along were just 4.1GHz (turbo) (and special TIM & magic Z97-only thermal wizardry), and then that recent 4.0GHz (base) rumor from last week actually took off and kicked the wind from their sails?
 
AFD said:
Or, maybe Intel's plans for DC all along were just 4.1GHz (turbo) (and special TIM & magic Z97-only thermal wizardry), and then that recent 4.0GHz (base) rumor from last week actually took off and kicked the wind from their sails?
Click to expand...
Of course it's 4.1. How that rumor made OCP front page news is between steve and his maker.
 
Well unless there is some huge hold up in the production facilities there would be no reason to hold back if they want to make a lot of profit. If I had to wait as late as September I would just wait another 4 months for Broadwell.
 
So has anyone been able to confirm or deny that the the 4790k will support VT-D as shown on the link from the forums?

It would be the first K processor's that support this on the non serverish (2011) socket.
Then the only point of going 2011 would be more memory bandwith/capacity and more pcie lanes, if you are not buying a hex/octo etc core.
 
man if AMD were more competitive we'd have 8-core desktop/server esque chips with higher overclockability. Oh well.
 
Bluesun311 said:
Of course it's 4.1. How that rumor made OCP front page news is between steve and his maker.
Click to expand...

Looks like the original 4.0 / 4.4GHz info might have been correct..

http://chinese.vr-zone.com/114262/intel-devils-canyon-core-i7-4970k-and-core-i5-4690k-cpu-z-show-05272014/

kbtHOp1.jpg


:D
 
If the rumored clock speeds are true, I just wonder how much headroom is left with these chips. Hoping for the best.
 
You must log in or register to reply here.
Back
Top