First complete review of Haswell i7-4770K

[Cloudfire777]

Here it is guys, the first complete review of i7-4770K.
Its in chinese but it have tons of graphs and pictures. I`ll upload some of them, so go to the source to find the rest. It involves CPU performance, IGP performance, CPU and IGP overclock, temperature measurements and much more.

CPU performance, i7-4770K vs i7-3770K

HD4600 (4770K) vs HD4000 (3770K)

Temperature measurements. First one is idle, second is running 0rthos paging monitor software.

Power draw. First one is idle, second is running Intel Burn Test Max mode.

The review: http://translate.google.com/translate?hl=en&sl=zh-CN&tl=en&u=http%3A%2F%2Fwww.chinadiy.com.cn%2Fhtml%2F24%2Fn-9024-3.html

 

[SiliconWars]

It's hard to believe it's that bad - think I'll wait on some mainstream reviews still.

 

[Vesku]

Seems like I should grab a 2500K/2600K before they completely disappear. Looking like that will be the last true enthusiast CPU from Intel.

 

[LagunaX]

No surprise, once there are more samples out, someone needs to delid one and use cool laboratory to show what it is capable of for the mainstream overclocker on high end air or water.

 

[RaistlinZ]

::yawn::

Guess my i7-930 will trod on for another generation refresh.

 

[podspi]

It's hard to believe it's that bad - think I'll wait on some mainstream reviews still.

Yep. Even if this is right on the mark, I guess we can be happy about the performance per watt improvements... and GPU performance looks great!

If the final benchmarks look like this I might have to march myself down to Microcenter and pick myself up a 3770k.

 

[Hulk]

It's hard to believe it's that bad - think I'll wait on some mainstream reviews still.

This.

 

[tracerbullet]

Thanks for the link. Not super impressed, but glad to see it for sure.

Agreed, would like to see some more reviews. The slim chance I was going to swap to one of these probably just went to nil.

 

[frozentundra123456]

Somewhat skeptical of this. Not expecting anything great, but real reviews will be out soon on production samples. Who knows where this came from.

 

[buklau]

I guess I will keep my 2500k for a while or maybe buy a used 2600/2700k later on...

 

[Charles Kozierok]

I will never understand why, on one of the premiere review sites on the Internet, people are constantly posting random reviews from random sites done by random people with random track records. Or worse, giving them any credence.

 

[Maximilian]

I will never understand why, on one of the premiere review sites on the Internet, people are constantly posting random reviews from random sites done by random people with random track records. Or worse, giving them any credence.

They were right about BD.

 

[Hulk]

I will never understand why, on one of the premiere review sites on the Internet, people are constantly posting random reviews from random sites done by random people with random track records. Or worse, giving them any credence.

We're all patiently waiting for real info on Haswell and any "new" info, even if we know it's fake helps bide the time until the real (read Anandtech) review is available.

I wouldn't search out this type of non-info but I will click the link in the forum just to be sure I'm not missing anything

 

[StrangerGuy]

They were right about BD.

The PRCs got real Galaxy S4 leaked before the official unveil, too.

 

[blackened23]

Haswell is supposed to have 10-15% improved IPC over IB. These results look in line with that, and should not be surprising.

Some of the other anomolous results are completely within the margin of error - anyone expecting amazing increases with Haswell should have known the anticipated IPC increase. 10-15%, same as SB to IB.

Efficiency and graphics are the important metrics today, and Haswell should absolutely deliver on those fronts. Anyway, for desktop users the real wildcard will be overclocking, and we just don't have good data on that. I'd suggest waiting on reviews from reputable websites for that, the CPU reviews here at AT are actually always good.

 

[meloz]

To get any significant improvement on CPU side, you will have to use software specifically compiled for Haswell (AVX2, TSX etc).

Or compile software yourself, which is easier for us open source folks than those stuck with Windows and other closed source stuff. Without re-compiling, CPU performance won't increase dramatically.

So Haswell will exhibit dual personality: those who compile software for it will get a monster, others will get a modest upgrade, at best.

 

[BallaTheFeared]

Am I seeing something different?

I see a processor that is slower while using more power!


After Bulldozer and watching Intel flip their CEO, I wouldn't be surprised

 

[frozentundra123456]

Haswell is supposed to have 10-15% improved IPC over IB. These results look in line with that, and should not be surprising.

Some of the other anomolous results are completely within the margin of error - anyone expecting amazing increases with Haswell should have known the anticipated IPC increase. 10-15%, same as SB to IB.

Efficiency and graphics are the important metrics today, and Haswell should absolutely deliver on those fronts. Anyway, for desktop users the real wildcard will be overclocking, and we just don't have good data on that. I'd suggest waiting on reviews from reputable websites for that, the CPU reviews here at AT are actually always good.

Those results are a lot less than 10% to 15% except in a couple of apps that must use AVX2 or open CL. There are even some where it is slower than ivy, which is what I find hardest to believe to be correct.

 

[ShintaiDK]

The obvious flaws they make have already been shown in another thread. They even write it themselves in the review too. Like the 3770K running 4 bins turbo under any load and so on.

They only test same frequency two places, and the graphics one is severely GPU limited with the GTX660:

 

[Carfax83]

Heck, I may as well sidegrade from my overclocked 920 to a Core i7 970 or 980x LOL! :biggrin:

 

[Haserath]

I do believe that 3770k is cheating.

Aida actually proves it because they would have the same read speed per clock!

 

[thilanliyan]

I am disappointed in the power consumption

 

[AtenRa]

Funny how GT2(4600) is faster in Synthetics (Vantage, 3D Mark 11) than GT630 but is almost produces half the frames in real games.

It seems that GT3 will have hard time beating GT630, and people expecting to be close to GT650 because of Intel’s 3D Mark 11 slides.

 

[Vesku]

Is 3DMark not as sensitive to memory bandwidth as actual games?

 

[georgec84]

The numbers don't look great, but I'll wait for more reviews to arrive.