Intel Devil's Canyon: The Good, the Bad, the Ugly @ [H]

He said he was able to get 5 Ghz without a fan. Is it possible he was using a giant block of copper (tons of thermal mass) to get that result? It's not terribly practical for home-use., but if he's working on T&E (testing and evaluation) that might be something that's done. He also didn't say what voltage he was using and to him 1.5V might be normal.

Can you cut to the chase and delid one of the engineering samples and see what it will do with no thermal limit? That's what really matters to me since that's how I run my chips (waterblock directly on the silicon), and the 4770K sucks ass where even with 1.6V it won't do 5 Ghz. I'll still be buying a 4790K just to find out if the limit is any better than a 4770K (I'll know this Saturday if that's the case).
 
OK, did a bit more testing at FULLY STOCK BIOS settings. I had the 4790K running solid under Prime95 for 30 minutes at 4.6/2400. Same voltages as mentioned previously. That was fairly impressive. Bumped it up to 4.7 and it BSODed immediately.
 
Think I'll keep my 920 X58 until Skylake comes out and then just get a non K Skylake
 
How disappointing. I just went and got a new m-itx case and new MSI Z97I Gaming AC board for this. Coming from a 2500k setup I thought it would be dramatic but now I'm considering waiting. This is a large chunk for me to put out for.
 
He also didn't say what voltage he was using and to him 1.5V might be normal.

Can you cut to the chase and delid one of the engineering samples and see what it will do with no thermal limit?

Well, 1.5v might be normal but there is no way to keep the processor from throttling because as soon as it hits 95c per core temperatures it is going to throttle, and we usually see that at around 1.45v.

I only have one right now and I am not taking it apart as of yet.

The temps we are seeing are better than 4770K by ~10%. But when you have to pour more voltage to get the clocks that advantage is erased quickly.
 
Last edited:
hmmm so you say @4.4 highest core temp was 70c, then you say 4.6 was fine (what was the highest core temp?) and 4.7 BSOD immediatly. Looks like BSOD before throttling so it didn't hit 90c I guess...

I think this sample was not cherry picked prior sending it to you.
 
That's not the largest issue (well it is, overhype), but the bragging and 'I'm smarter than you' coming from that rather high up Intel staff member is the part that's really getting this going.
 
Thanks Kyle. I guess my "ancient" Phenom II 955 / 4ghz is going to stick around a while longer. I was really looking for some upgrade excitement but apparently these are not the droids we are looking for.
 
Could AMD pullsomething out there ass like before?

Nope. This ain't the movies...the real worldz the underdog loses, goes bankrupt, gets divorced, dog dies, and then they kill themselves.

It appears the 'New Tim' hasn't done anything for the uneven core temps many have experienced with the 4770K

But if you look at TIM benchmarks, the difference between "the best" and "the worst" when properly applied is only a few degrees C. So the DC TIM being 5-10C better is pretty significant. Yeah not as good as solder or Liquid Ultra, but still for something mass produced...
 
I'm highly curious to see if Intel says anything more substantial than "Everyone is doing it wrong" because this almost doesn't make sense to me.

Why would Intel go through the extra time, effort, money, and resources to delay Devils Canyon after the Haswell refresh if it's a worse chip than the original? Is there something wrong with the engineering samples they gave out?

Also, I'm lost for all these people saying they're sticking with Sandy Bridge. I for one can't wait to dump my 2600k for something new. The clock-for-clock efficiency increase from then till now is worth the price alone. It's not all about the Ghz people. ;)
 
Could AMD pullsomething out there ass like before?

I think this a bit more difficult now since AMD does not own their fabs. And AMD is the one with the inefficient design instead of Intel.
 
Also, I'm lost for all these people saying they're sticking with Sandy Bridge. I for one can't wait to dump my 2600k for something new. The clock-for-clock efficiency increase from then till now is worth the price alone. It's not all about the Ghz people. ;)

No, it's not all about the GHz. However, one has to weigh whether or not the rest is compelling enough, or if waiting for completely new features might be worth it. I'm not even on either "bridge" and still in a "field". I want to upgrade, but when 6 months hit, and a new memory architecture is available, I think that's more enticing. I agree that this would be a huge upgrade alone though. Just waiting for more I guess.
 
Also, I'm lost for all these people saying they're sticking with Sandy Bridge. I for one can't wait to dump my 2600k for something new. The clock-for-clock efficiency increase from then till now is worth the price alone. It's not all about the Ghz people. ;)
A 4.6GHz 2600K ~= 4.47GHz 3770K ~= 4.1GHz 4770K
That's why ;)

If your average 4770K can only attain an OC of 4.4GHz, you have to ask yourself if +300MHz is really worth a new CPU + mobo.
Hell, it took me this long to upgrade from my old i7 920 D0 @ 4.1GHz.
My 4770K @ 4.55GHz is only equivalent to a 5.67GHz Nehalem (not counting new instructions), which is not shabby in absolute terms, but gorram awful considering it's been 5+ years.
 
Well, the original reports did sound a little bit too good to be true.

Hopefully they learn that it's better to under-promise and over deliver, than the opposite.
 
Man so it's sounding like there is nothing better about this 4790 compared to the 4770 at all. In fact it seems almost worse or at best equal. Better cooling but needs more voltage so you're back to square one and you're not getting any higher overclocks out of it. No wonder it's coming out at the same price as the 4770K. It's apparently no better.
 
Man so it's sounding like there is nothing better about this 4790 compared to the 4770 at all. In fact it seems almost worse or at best equal. Better cooling but needs more voltage so you're back to square one and you're not getting any higher overclocks out of it. No wonder it's coming out at the same price as the 4770K. It's apparently no better.

It's not unheard of for a 4770K to not be able to hit 4.4 Ghz on all 4 cores. At least the 4790K minimizes this from happening, plus lower temperatures at the same price as a 4770K.

So it's more like "at worst equal, at best maybe we'll see 5 Ghz with watercooling and a ton of voltage".
 
Thanks for the constant updates, they actually make me feel a little better about this.

Maybe there is some setting getting in the way?

Trust me, that is why I have not just walked away from this with my hands in the air. For Intel to make such specific statements that do not stand up is a bit baffling.

It's not unheard of for a 4770K to not be able to hit 4.4 Ghz on all 4 cores. At least the 4790K minimizes this from happening, plus lower temperatures at the same price as a 4770K.

So it's more like "at worst equal, at best maybe we'll see 5 Ghz with watercooling and a ton of voltage".

EVERY 4770K I have purchased I have been able to get 100% stable at 1.28v.
 
Tried seeing if 4.6/2400 was stable at stock voltage again, but no it is not likely to happen. Three failed tests so far. Back to 4.5GHz again.
 
When @4.6, are temps close to throttling? The fact it BSOD before throttling shows the problem seems not about heat.

Have you tried disabling iGPU? Just curious...

It is not throttling. Temps not even close to throttling.

There is no switch in the BIOS on this board to "disable" iGPU to my knowledge, if I am overlooking it, please let me know what menu it is in. I have been over this BIOS a lot since this board came out and I have not seen it.
 
Sad to hear its not working out, thanks for the confirmation that it is a different ES chip than PCPer, I wonder if this chip is more about the 4.4-4.6 range with lower temps and voltage rather than all out speed we were all hoping for (and Intel lead us to believe...) 3930K at 4.5 vs 4790K at 4.6, power efficiency its worth it, but overall performance is going to be fairly similar trading blows back and forth.

Does anybody know what the wattage delta is between X79/Hex and Z97/Quad overclocked?
 
What bothers me the most about what h is reporting on DC OC is the part about using adaptive voltage. No thanks. Adaptive voltage mode causes stuttering and latency. I prefer manual 24/7. I'm at 3.5 or 4.6 always and I have very smooth web browsing for it. Adaptive doesn't work very well for me on haswell.

Same thing on SB for me ... I really did not like the lag coming out of 1.6GHz mode. So much more snappy pegged above 4GHz.

Man, I was really planning on a DC upgrade from my 2500k. Might still do it, my hardware is abused and getting old, relatively. Those caps don't last forever.
 
So pretty much this chip isnt as good as the 4770K youve got in your rig right now? That is kinda surprising. I didnt in a million years expect that out of DC.


Well, I would not say that.

This DC is currently running stable at 4.5/2400 at 1.26v with 65/70/70/64C per core temps.

We usually run our 4770K CPUs for motherboard bench testing at 4.5/1866 at 1.28v with around 13% higher per core temps.
 
Would there be any difference from an engineering sample to a final product? Just asking.
 
Would there be any difference from an engineering sample to a final product? Just asking.

Depends on where in the life cycle it is. First samples of a new product - hell's yeah there's a difference. Features may not be working at all, cores might not be working right at full-speed, might be down-clocked intentionally to not let on the final product clock speed...etc etc . For DC, which is still based on a mature product, I doubt there's going to be much difference, especially if the part isn't even stamped as an ES.

I don't know if the reviews coming out now we're really written a month ago and embargo-ed since, but if reviewers are getting parts now chances are they're the same stuff going out to retailers as we speak.
 
Seems like 4.5 is much more guaranteed now, is that a fair assessment?
 
still on a bloomfield 2.6 running at 3 and happy enough. but then I stopped buying so many game that I cant find time to play.
 
Also, I'm lost for all these people saying they're sticking with Sandy Bridge. I for one can't wait to dump my 2600k for something new. The clock-for-clock efficiency increase from then till now is worth the price alone. It's not all about the Ghz people. ;)

You think we want to wait?
DC could have been the chip to upgrade to, but it isnt.
Approx 20% performance increase avg doesnt do it for me when its going to cost over £400.
I'd much rather have other toys.

Then theres the rewarding Intel for making things up angle.
It looks like this is what has happened, but maybe, just maybe, EVERYONE is doing it wrong :D
Hard to believe given they needed LN2 with an air cooler for the "air cooled" demo.
I credit Kyle for his patience.
 
It is a bit strange that the wall doesn't appear temp related with DC (same for haswell really) they just get hot as anything else. They simply had a crazy good example on their hands I guess. There will be folks that luck out on these and hit 5ghz I'm sure. But if Kyle puts it at 5%. Well I've played enough games with bad odds to know not to play that one. Took me 6 4770ks to hit 4.6ghz prime stable.
 
Yeah I dont think Intel deserves my money with this Devils CanYawn stopgap processor. Until performance actually jumps (IPC wise) or clocks get a fair bit higher, 5GHz+ im sticking with my X79...Sorry Intel no money from me this time 'round, I was ready too. Bad PR stunt...
 
Couldn't agree more to stay where you are at if you aren't just about always looking for an excuse to do a build but it is a bit disappointing in general because I think we all just want to see progress since the recession. And this isn't real progress but then again I have to stress I had 3 4770k's in a row that would not budge past 4.2ghz. So if they can guarantee you 4.4+ I guess that's something. They didn't exactly say: hey you'll be 5ghz 24/7 p95 stable on air on stock manual voltage. That was what I wanted but doesn't that sound likd too much to ask for considering were still trying to get them to solder the chip together? :rolleyes:
 
Seems like 4.5 is much more guaranteed now, is that a fair assessment?


I feel like that would be a good "rule of thumb" to roll with. I know there are some more samples out with ODMs that required a little voltage bump to get to 4.5 though.

So, if you are buying, there is no reason not to buy a 4790K if it is in your price range. No reason to go 4770K.
 
From the five websites I have seen so far with 4790K overclocking results..

Kyle - ASUS Z97-Deluxe and got 4.7GHz at 1.36v core using a Koolance EX2-755 with Koolance 380 model waterblock

PCPer.com - ASUS Z97-Deluxe and got 4.7 GHz at 1.36v with a Corsair H80

Lanoc.org - ASUS Z97-A and got 4.7GHz at 1.288v core using a 120mm Noctua NH-U12S heatsink, although later tweaking had results of 4.84GHz with a bclk of 103 and a multiplier of 47 (no core voltage info given)
http://lanoc.org/review/cpus/6877-intel-i7-4790k-devil-s-canyon?showall=&start=4

Vortez.net - MSI Z97 XPOWER AC and got 4.8GHz at 1.38v core using a Corsair Hydro Series H105 (done "After many hours of adjustments and tweaking")
http://www.vortez.net/articles_pages/intel_devils_canyon_core_i7_4790k_review,5.html

Tomshardware.com - MSI Z97 Gaming 7 and got 4.6GHz at 1.31v core using a Noctua NH-U12S (fan set at 100%) (this review doesn't give any indication about how hard it tried to achieve the maximum stable overclock it could; actually gave me the impression it didn't try very hard)
http://www.tomshardware.com/reviews/core-i7-4790k-devils-canyon-overclock-performance,3845-2.html

Just wondering if..
a) the ASUS Z97-Deluxe isn't the best motherboard to use for overclocking.. and
b) air or water don't seem to make much difference in 4790K overclocking
 
From the five websites I have seen so far with 4790K overclocking results..

Kyle - ASUS Z97-Deluxe and got 4.7GHz at 1.36v core using a Koolance EX2-755 with Koolance 380 model waterblock

PCPer.com - ASUS Z97-Deluxe and got 4.7 GHz at 1.36v with a Corsair H80

Lanoc.org - ASUS Z97-A and got 4.7GHz at 1.288v core using a 120mm Noctua NH-U12S heatsink, although later tweaking had results of 4.84GHz with a bclk of 103 and a multiplier of 47 (no core voltage info given)
http://lanoc.org/review/cpus/6877-intel-i7-4790k-devil-s-canyon?showall=&start=4

Vortez.net - MSI Z97 XPOWER AC and got 4.8GHz at 1.38v core using a Corsair Hydro Series H105 (done "After many hours of adjustments and tweaking")
http://www.vortez.net/articles_pages/intel_devils_canyon_core_i7_4790k_review,5.html

Tomshardware.com - MSI Z97 Gaming 7 and got 4.6GHz at 1.31v core using a Noctua NH-U12S (fan set at 100%) (this review doesn't give any indication about how hard it tried to achieve the maximum stable overclock it could; actually gave me the impression it didn't try very hard)
http://www.tomshardware.com/reviews/core-i7-4790k-devils-canyon-overclock-performance,3845-2.html

Just wondering if..
a) the ASUS Z97-Deluxe isn't the best motherboard to use for overclocking.. and
b) air or water don't seem to make much difference in 4790K overclocking

I wouldnt think Asus isnt the best cause they usually are or right near the top. With all those reviews coming within a couple hundred MHz of each other, Id say all those high end motherboards are good enough to overclock. A difference of 100 MHz is more silicone lottery than it is motherboard probably.
 
Thanks for the review Kyle.

Judging by what we are seeing, it is looking like these chips are not much better overclocking than many 4770Ks.

I wonder what's causing it?

- This process doesn't seem to overclock well at all - maybe FinFET vs planar just doesn't scale that well
- The FIVR is making things hotter for sure
- Even though they narrowed the gap, the thermals aren't as good without a fluxless solder as on the 32nm parts

We just aren't seeing the performance gains that we once did. That and Intel really messed up their marketing message this time.
 
From the five websites I have seen so far with 4790K overclocking results..

Kyle - ASUS Z97-Deluxe and got 4.7GHz at 1.36v core using a Koolance EX2-755 with Koolance 380 model waterblock

PCPer.com - ASUS Z97-Deluxe and got 4.7 GHz at 1.36v with a Corsair H80

Lanoc.org - ASUS Z97-A and got 4.7GHz at 1.288v core using a 120mm Noctua NH-U12S heatsink, although later tweaking had results of 4.84GHz with a bclk of 103 and a multiplier of 47 (no core voltage info given)
http://lanoc.org/review/cpus/6877-intel-i7-4790k-devil-s-canyon?showall=&start=4

Vortez.net - MSI Z97 XPOWER AC and got 4.8GHz at 1.38v core using a Corsair Hydro Series H105 (done "After many hours of adjustments and tweaking")
http://www.vortez.net/articles_pages/intel_devils_canyon_core_i7_4790k_review,5.html

Tomshardware.com - MSI Z97 Gaming 7 and got 4.6GHz at 1.31v core using a Noctua NH-U12S (fan set at 100%) (this review doesn't give any indication about how hard it tried to achieve the maximum stable overclock it could; actually gave me the impression it didn't try very hard)
http://www.tomshardware.com/reviews/core-i7-4790k-devils-canyon-overclock-performance,3845-2.html

Just wondering if..
a) the ASUS Z97-Deluxe isn't the best motherboard to use for overclocking.. and
b) air or water don't seem to make much difference in 4790K overclocking




Here's another to add to the list:

Legitreviews.com - 4.9GHz w/1.4VCore. Kept getting BSOD so they dropped back to 4.7GHz @ 1.351 to get a stable OC. Temps around 76C. And that's all with Corsair H105 Extreme water cooler kit.

Legit Reviews 4790K CPU Review
 
Back
Top