i7 4790K ES Review: mildly disappointing?

[TheFamilyMan 280]

Why, did you expect anything different? Did you really think that they would redesign the chip?

 

Disappointing? Only for overclockers, for anyone else it's quite an improvement going from 3.5Ghz to 4, that's about a 15% gain, even better for people like me who have skipped the 4770K.

 

And last.... you're making conclusion on a single site, based on a none-retail version of the CPU.

 

Seems you are responding only to the title of my post.  All your comments are touched upon in my post.

 

As for redesign, BTW I did expect something considering they designated it "Devil's Canyon" rather than keeping the name Haswell.

[ChaoticBeauty 6,692]

But gentlemen... if it's still TIM, think what you can achieve if you delid it!!!!

 

That's the problem; you've still got to delid the CPU to achieve its full potential, taking a huge risk.

[z06z33 763]

The bad overclockability is the fault of the TIM, not the chip itself. If they used fluxless solder (as it was expected), it would overclock fine.

I doubt they want it to be a great overclocker, think about it Haswell is a great CPU but if people could hit 5-6ghz on it no one would buy a new cpu for many many years. Im coming from an AMD chip so if I can run the same 4.5ghz on my 4790k that I hit on my FX8350 I'll still be miles ahead.

[martin-w 3,234]

That's the problem; you've still got to delid the CPU to achieve its full potential, taking a huge risk.

It was tongue in cheek, but...

 

When you consider that the fluxless solder alternative used these days is actually just a conductive glue, and somewhat inferior to the true hot solder used in years gone by... delliding probably achieves better results.

[bic 44]

wow another potential wait?

 

I have an i7-920 which I can barely get to 3.2 ghz.  I was (probably will still) going to upgrade it all in September...

[martin-w 3,234]

That surprises me, my old i7 920 did 4GHz easy, with an NH-D14. I sold it to a friend who is still at 4Ghz with an Archon.

 

I'll gladly ask him for a copy of the settings if you like, as I no longer have them.

 

Of course you may have just been out of luck with the silicone lottery.

[TheFamilyMan 280]

Non-OC temps at 88C???  Holy ****.  :huh: :blink: :wacko:

 

Maybe that was the inspiration for the designation "Devil's Canyon". :lol:

[Tabs 2,790]

wow another potential wait?

 

I have an i7-920 which I can barely get to 3.2 ghz.  I was (probably will still) going to upgrade it all in September...

 

I've been running an 860 at 4GHz for 4 straight years - you must have gotten a really badly binned chip.

[martin-w 3,234]

Agree, he may have come across badly in the silicone lottery.

 

Or... his settings are dodgy.

 

Or cooling.

 

More info required.

[kristolz 26]

Sigh. I guess this isn't it. We need something clockable to 5.0 Ghz.

If you're lucky in the chip lottery, you might already be pretty close! My 3770k is able to crack 5.0 with 1.395v VCore (and a little overvolted PLL). I haven't delidded it, nor lapped it. I fitted it with the biggest Corsair water cooler, and voila. Runs @ +65->70 C in a +22C room max in FSX. Tempratures are a bit alarming in Prime 95, I must admit, but it has yet to CTD FSX (no BSOD either). 

 

That said - my first 3770k, was nowhere near as stable as this one. To punch through the 5ghz barrier, it required almost 1.510 volts Vcore. The Corsair cooler didn't stand a chance at that voltage. Delidded the CPU, and fitted it with some new thermal paste (the old one was rubbish - all hard.), and the temprature dropped somewath, but it was to unstable. It remained stable at around 4.2 ghz, but filled up the event log with "hardware error" messages, and crashed FSX continuously. (It now runs at stock setup, rock stable, in a Ubuntu Server I'm using to run a couple of virtual Windows 7 computers, running ASN, Aivlasoft etc).  

 

I've done all my clocking (although removing turbo boost, HT and everything else smelling of power saving options in BIOS) in the software tweaking stuff bundled with the motherboard. I only apply the overclock when running FSX, P3Dv2.2 or XP10. 

 

Best regards, 

 

Kristoffer

[martin-w 3,234]

 

 

My 3770k is able to crack 5.0 with 1.395v VCore (and a little overvolted PLL).

 

Quite frankly that's an amazing achievement with no deliding.

 

My 3770K is closer to your dodgy old one.

[Christopher Low 5,352]

 

I doubt they want it to be a great overclocker, think about it Haswell is a great CPU but if people could hit 5-6ghz on it no one would buy a new cpu for many many years.

 

You could argue that case for any decent i5k chip upwards. I am not convinced that upgrading from my i5 2500k @ 4.3Ghz would be in any way a smart move at the moment.

[kristolz 26]

Quite frankly that's an amazing achievement with no deliding.

 

My 3770K is closer to your dodgy old one.

I did hold my breath for nearly a couple of months - thought it would go belly up after a few weeks, but forntunately not! 

 

I don't know this for a fact, but I suspect the thermal paste in the old CPU got pretty damaged during some recless clocking attempts, where the temp shot through the 100c barrier a few times. This was prior to de-lidding, and it didn't quite recover after that... Maybe Intel changed the paste for later batches - I don't know. 

 

Cooling, a good case and a lucky chip is alfa and omega (...and zeta).  

[gizmosellsbunnys 106]

Why, did you expect anything different? Did you really think that they would redesign the chip?

 

Disappointing? Only for overclockers, for anyone else it's quite an improvement going from 3.5Ghz to 4, that's about a 15% gain, even better for people like me who have skipped the 4770K.

 

And last.... you're making conclusion on a single site, based on a none-retail version of the CPU.

I am looking into getting it for that reason as overclocking is not my thing anyways and 15% gain helps for XPX and FSX smoothness. This chip will sell well for non oc users that are leery of ocing.

[martin-w 3,234]

I don't know this for a fact, but I suspect the thermal paste in the old CPU got pretty damaged during some recless clocking attempts, where the temp shot through the 100c barrier a few times. This was prior to de-lidding, and it didn't quite recover after that... Maybe Intel changed the paste for later batches - I don't know.

I don't think temperature damages TIM to be honest.

 

It's actually the thickness of TIM due to the epoxy glue used to adhere the IHS to PCB that causes the issue. The TIM itself isn't that bad in terms of quality. Delidding decreases the gap between IHS and die, hence the better temps.

 

It's that variability in IHS/Die gap that accounts for the variable temps with Ivy and Haswell.