Intel Haswell i5-4670K vs. i7-4770K Comparison

[Ikaruga]

Yes. Most bios have the ability to enable/disable the iGPU.

In my experience, there is no difference with it enabled and disabled when overclocking under LN2.

I think Dave wrote it somewhere that disabling the IGP stopped getting BSODs from somebody's OC-ed Haswell .

 

[HeavensRevenge]

Yes. Most bios have the ability to enable/disable the iGPU.

In my experience, there is no difference with it enabled and disabled when overclocking under LN2.

Unfortunate how disabling it doesn't have much effect on the overclockability of the chip. I was thinking that the IGP was getting in the way with more things to go wrong/fail.

Well hopefully they can get a nice breakthrough for Skylake

Thanks for the info!

 

[EarthDog]

I think Dave wrote it somewhere that disabling the IGP stopped getting BSODs from somebody's OC-ed Haswell .

Good to know. I have only really mucked with it a couple times and it didnt do squat, so I just leave it as is.

 

[Raghar]

Wow I thought I'd write rest of the review out of pity when I'd get my i7-4820K. That's quite a difference when I compare it to PSU reviews.

Well I'm not your superior to find all problems in your review and force you to remedy them, it's a sign of quality reviewer to make his own stuff and be creative.

That i7, was that a Intel supplied chip, or a random chip bought from a computer shop? The same question about i5.

I didn't see voltages you used for overclocks. Can you list them?

 

[FireKillerGR]

Wow I thought I'd write rest of the review out of pity when I'd get my i7-4820K. That's quite a difference when I compare it to PSU reviews.

Well I'm not your superior to find all problems in your review and force you to remedy them, it's a sign of quality reviewer to make his own stuff and be creative.

That i7, was that a Intel supplied chip, or a random chip bought from a computer shop? The same question about i5.

I didn't see voltages you used for overclocks. Can you list them?

Welcome to the forum.
Glad you did your first post on my review.
Thanks for your feedback as well. Btw, you shouldnt feel pity as I dont feel so about my reviews..
Also, I would like to make it clear that I prefer showing a review which will include "part" of me than smth which wont.
I dont aim to post 100% professional reviews ( at least not yet), I just want to add my "signature" (make my reviews a bit different) and become better through my work.

As for the chips, both are samples.

For the 4670k I used 1.28v for 4.2 GHz and 1.24v for the 4770k to achieve 4.7 GHz.

 

[Nabarun]

Would be nice to see how the I5 did on a Asus Z87 board, preferably using H100i. Thanks for the review.

 

[HeavensRevenge]

Would be nice to see how the I5 did on a Asus Z87 board, preferably using H100i. Thanks for the review.

For the Haswell CPU's I've tried, it seems that the chipset & controller improvements meant to be paired with a Haswell chip actually influence benchmarks and performance more than the CPU.

The benchmarks I'd love to see since I see this effect myself a bit, are benchmarks with different combination of the previous generation chipset/controllers AND CPU's set in them if possible.

Example:
Z77 + Ivy Bridge CPU
Z87 + Ivy Bridge CPU
Z77 + Haswell CPU
Z87 + Haswell CPU

Since Haswells seem to have such a tiny miniscule improvement, yet many aspects of a system are increased when it's paired with an Z87 chipset. For example, my SSD speeds are way better on a Lynx Point chipset even though it's the same SSD being used each time. It's obvious a new chipset would be better but I'd love to know exactly how much better

Can't wait for the next bench, keep them coming

 

[Nordic]

For the Haswell CPU's I've tried, it seems that the chipset & controller improvements meant to be paired with a Haswell chip actually influence benchmarks and performance more than the CPU when a Z77 based mobo is used.

The benchmarks I'd love to see since I see this effect myself a bit, are benchmarks with different combination of the previous generation chipset/controllers AND CPU's set in them.

Example:
Z77 + Ivy Bridge CPU
Z87 + Ivy Bridge CPU
Z77 + Haswell CPU
Z87 + Haswell CPU

If anyone can do it reasonably simply, that is what I'd love some quick numbers regarding. Since Haswells seem to have such a tiny miniscule improvement when put in a mobo of the previous chipset, yet many aspects of my system are increased monumentously when it's in the Z87 chipset to work with. For example, my SSD speeds are way better on a Lynx Point chipset even though it's the same SSD being used each time. It's obvious a new chipset would be better but I'd love to know exactly how much better

Can't wait for the next bench, keep them coming

z77 and haswell don't go together. Neither do z87 and ivy/sandy.

 

[ViperXTR]

3570K starts at 3.5Ghz?

 

[HeavensRevenge]

z77 and haswell don't go together. Neither do z87 and ivy/sandy.

Sad

 

[Nordic]

Sad

Ya, lga1155 is sandy/ivy. Haswell is lga1150. I want a z87 board but don't want the lga1150 cpu.

 

[Ocssor]

I would like to know why the temperatures are so different under load with the 4770 in this comparison, compared to your actual review of that chip?

@ 4.2 on the review of the chip it's at 78 but on this review it's at 86.

Bit of a discrepancy, is there a reason for this? Apologise if I've misread or made a mistake, just got me a bit confused.

 

[FireKillerGR]

Its a different 4770K.
The 4770K from the previous review was 100% retail, and the best one when it comes to temperatures.

 

[Ocssor]

Ah, that makes sense. I've just ordered a build with a 4770k overclocked @ 4.5 with megahalem rev c and got a bit worried seeing those comparison temperatures...

 

[springs113]

Haswells are temp whores...not so bad in my loop @4.5 stress test goes mid to high 70s now.

 

[Ocssor]

Yeah I'm hoping it won't be too bad. The ambient temp will be ~20c and the case itself comes with 3 fans (2 on top 1 on front) + a noiseblocker pro for rear cooling. Cpu cooler is as I've mentioned megahalems red rev c. Hopefully it will run just fine and not too hot.

 

[Nabarun]

It's usually ~30C here in Kolkata. Gets >40C in summer...

 

[Ocssor]

It's 20c outside just now and that's considered warm here! So inside ambient temperature will actually be lower than that I guess. Might revise my original temp to 13-16c

 

[Nabarun]

 

[Deleted member 24505]

Its a different 4770K.
The 4770K from the previous review was 100% retail, and the best one when it comes to temperatures.

Is that the one you are keeping then

 

[TheGuruStud]

Wow, what a crappy 4670k. It should only take 1.2v to do that (at 1.27 you should be doing 4.5 at least).

I assume you were able to do 1:1 with the uncore clocks?

 

[FireKillerGR]

Yeah, I could do 1:1 uncore but the gain wasnt that high.

Will re-test, it may be the gpu :S

 

[TheGuruStud]

Yeah, I could do 1:1 uncore but the gain wasnt that high.

Will re-test, it may be the gpu :S

Oops, reedited lol (not as relevant to me haha). Yeah, those fps numbers are weiiird.

Odd that the uncore boost didn't give good gains, too. I thought that's what everyone has been harping about doing when OCing.

 

[Nabarun]

FireKillerGR,
If you have access to an Asus Z87 board, could you please try and see if better OC/performance can be achieved with the I5?

 

[Raghar]

FireKillerGR,
If you have access to an Asus Z87 board, could you please try and see if better OC/performance can be achieved with the I5?

I doubt that. That i5 looks like a bad chip, or bad overclocker.

BTW would you run test with something that really taxes the CPU, like with 4 instances of Dwarven fortress? World generation is quite meaty, and well, it took about 20 minutes on my computer, at least that version I tried. I think it would be a fair real world comparison between i5 and I7.



Interesting, Chrome is smart enough to offer correct choice to fill the random question.