• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Intel Core i7-4770K Haswell

I have L309B318(board testing, 1.040 V) and L311B405(memory testing, 1.020 V).


.

Ok thanks it looks like ones around L310B"488" <- lower better Mala are doing ok getting around 4.6GHz with 1.25v but I've only seen a couple.
 
I guess I'm going to order another one tomorrow, unless the locked 4770 is a better choice. The BCLK straps seem intriguing and it seems like the stock 4770 (if you are able to force all the cores to 39x) would overclock comparatively, would it not? Anyway, time to wait another week and there's no way I am attempting to free it from the IHS either. I will just enjoy it at 4.2GHz or so and reap the IPC benefits with this architecture.

I have L309B318(board testing, 1.040 V) and L311B405(memory testing, 1.020 V).



Might be cache multi since of that label, haven't booted an ASRock Z87 board yet, but soon.

Seeing as the batch code of the one I had was L311B404, does that mean we had chips that were on the same wafer or something? :p
 
I think i want my HTPC to be a Haswell rig, seeing as i already have a SB rig and an IB rig.
 
I think i want my HTPC to be a Haswell rig, seeing as i already have a SB rig and an IB rig.

Might as well, I just buy whatever is out at the moment even if the whole TIM vs. solder issue exists. Supposedly, these should be about 10-20% faster than a 2700K at the same clocks, about 40% more power efficient, and the IGP is almost twice as good. I know everyone wants low temps and 5GHz clocks and whatever, but at least the power consumption decrease and higher per-core performance kind of makes up for it.

Be sure to get a board with the Realtek ALC 1150 if you are using onboard analog sound.
 
The 4770K is an amazing chip no doubt but really not much better than the 3770K! So... for those who of us just need a Gaming/ Streaming rig: AMD FX 8350 vs Intel 3570K vs 3770K vs 3820 - Gami... :slap:

We all know AMD has great pricing, but the performance in some applications can be a crap shoot. Take Civilization V for instance, a game that runs notoriously bad on AMD processors:

civv-lgv.gif


I know many people certainly disagree with my support of Intel mainly due to the cost to relative performance of the processors compared to their AMD counterparts, but there are a few reasons why I prefer them nowadays even if they do cost me a few hundred more dollars overall to obtain. The latest 22nm LGA 1155/1150 i7 processors are more energy efficient, tend to have more predictable performance in games, higher single threaded performance which is crucial in CPU-heavy games that use 4 threads or less, and overclocking that doesn't add 150-350w of power consumption just for the CPU (I'm looking at you, FX-81xx) http://www.bit-tech.net/hardware/2012/11/06/amd-fx-8350-review/7 .

Other notable things I like with the Z87 platform are fresh, new motherboards (AM3+ boards seem pretty stale, and most are from 2011), PCIe 3.0 (which AMD's 2011 essentially rebrand of their 800 series chipsets doesn't have, and FM2 doesn't have it either), native USB3 ports (AM3+ boards rely exclusively on 3rd party controllers), a single PCH over legacy NB/SB that AM3+ has and a large focus for better onboard audio solutions aside from Creative SoundCor3D with Z87 that I have not seen with AM3+/FM2 or even most X79 boards. Granted, I could be quite biased with a lot of these points since most people just care about performance to cost ratio above everything, but I think the extra little features do help warrant the price of modern Intel mainstream solutions.

I'm not completely anti-AMD, as their APUs do make fantastic $600 incl. Windows PCs for home/general users, but IMO their performance platform seems dated (X79 certainly is as well to a lesser extent) and for my preferences, Intel LGA 115x delivers. I play a lot of CPU intensive games and at 120Hz so I need every drop of CPU performance I can get to feed my 7970. Having certain games run a good 10-30% slower with an AMD chip doesn't bode well for my usage scenario. AMD does have a really good price point for people who only invest a few hundred dollars in a PC every couple years, but again my computer investments are quite the opposite.
 
I'm just saddened by the fact that we PC enthusiasts are a dying breed if Intel gets their way and AMD doesn't step up their game.

I mean, don't get me wrong, Haswell is the fastest CPU around when it comes to raw performance, my 4 core 4770K is already faster than my 6 core 3930K in most tasks, even if doesn't OC as much; but the fact that Intel has regressed in terms of OCing potential for the last couple of generations really doesn't bode well for enthusiasts.

It looks like Broadwell is not going to be any better, and the days of socketed CPUs may be over, unless AMD can leverage the advantage of being the chosen architecture for all new consoles and use that as a weapon in the PC front...
 
I'm just saddened by the fact that we PC enthusiasts are a dying breed if Intel gets their way and AMD doesn't step up their game.

I mean, don't get me wrong, Haswell is the fastest CPU around when it comes to raw performance, my 4 core 4770K is already faster than my 6 core 3930K in most tasks, even if doesn't OC as much; but the fact that Intel has regressed in terms of OCing potential for the last couple of generations really doesn't bode well for enthusiasts.

It looks like Broadwell is not going to be any better, and the days of socketed CPUs may be over, unless AMD can leverage the advantage of being the chosen architecture for all new consoles and use that as a weapon in the PC front...

I certainly agree. Intel, as much as they have nicely performing products, has been playing dirty with the whole TIM under the IHS deal among other things. I wouldn't have to get a new 4770K now if they had used fluxless solder, as I cracked my die when I stupidly delidded mine right away as it was hitting 89c at 4.2GHz/1.17v with just normal distributed computing load. The next one I get will be untouched and I'll just deal with the high temperature readings. If it's equivalent to a 4.5GHz IB or 4.8GHz SB (at 4.2GHz) in many cases, then it isn't worth the trouble for me to get a few hundred extra MHz out of mine.
 
I certainly agree. Intel, as much as they have nicely performing products, has been playing dirty with the whole TIM under the IHS deal among other things. I wouldn't have to get a new 4770K now if they had used fluxless solder, as I cracked my die when I stupidly delidded mine right away as it was hitting 89c at 4.2GHz/1.17v with just normal distributed computing load. The next one I get will be untouched and I'll just deal with the high temperature readings. If it's equivalent to a 4.5GHz IB or 4.8GHz SB (at 4.2GHz) in many cases, then it isn't worth the trouble for me to get a few hundred extra MHz out of mine.

So sorry to hear about your proc, do you mind if I ask what method you used for deliding it? I wasn't aware that it was such a dangerous procedure.

The temps you were hitting are just insane, in my particular case, i haven't seen such high temps but I use HWMonitor to check the temps, do you think they are being missreported? I also use the monitoring utility that came with my board, and it seems to coincide in the temps reported by HWM.

I really hope you have better luck with you next Haswell, it seems like there're some procs that are better suited for OCing...
 
So sorry to hear about your proc, do you mind if I ask what method you used for deliding it? I wasn't aware that it was such a dangerous procedure.

The temps you were hitting are just insane, in my particular case, i haven't seen such high temps but I use HWMonitor to check the temps, do you think they are being missreported? I also use the monitoring utility that came with my board, and it seems to coincide in the temps reported by HWM.

I really hope you have better luck with you next Haswell, it seems like there're some procs that are better suited for OCing...

I used the razor blade which is probably the more dangerous method than the hammer/vise/wood method. I had finished my 3770K about an hour before and besides some PCB nicks, it came out great, so I decided to try the Haswell. I started on the side that had no surface traces and got halfway around fine, but I wasn't so gentle with the other half, tried to slightly pry the rest of the IHS off while using the blade which sent too much tension through the PCB and the die hairline cracked at one of the corners and across the short side of the chip. I did do a better job on this chip coincidentally, but when I polished the die I noticed the crack and practically died. If the die were more square instead of a stupidly tall rectangle, I would have probably gotten away fine. Also, Haswell has a row of tiny SMD capacitors on the substrate next to the die now, so even if you delid you'll probably kill the chip with Liquid Pro unless you completely cover the capacitors in epoxy. Anyway, lesson learned, and I hope the people contemplating this hear me out if $349 means a lot to you.
 
I used the razor blade which is probably the more dangerous method than the hammer/vise/wood method. I had finished my 3770K about an hour before and besides some PCB nicks, it came out great, so I decided to try the Haswell. I started on the side that had no surface traces and got halfway around fine, but I wasn't so gentle with the other half, tried to slightly pry the rest of the IHS off while using the blade which sent too much tension through the PCB and the die hairline cracked at one of the corners and across the short side of the chip. I did do a better job on this chip coincidentally, but when I polished the die I noticed the crack and practically died. If the die were more square instead of a stupidly tall rectangle, I would have probably gotten away fine. Also, Haswell has a row of tiny SMD capacitors on the substrate next to the die now, so even if you delid you'll probably kill the chip with Liquid Pro unless you completely cover the capacitors in epoxy. Anyway, lesson learned, and I hope the people contemplating this hear me out if $349 means a lot to you.

That's really sad, using the vise method is probably safer, well, probably your next 4770 will have better thermals :)

You know, I used coolaboratory's liquid pro in the past, and it has excellent thermal properties, but once I almost killed both an Asus MIVE and a 2600K CPU because of it, I was moving these parts to a new case and was about to remove the H100 cooler from the proc but it was bonded to it, I tried gently twisting it but I stopped when I realized I was using too much torque and would probably rip the whole socket from the board.

I finally decided to remove the CPU along with the heatsink, it was almost as if it had been soldered to the cooper plate, I ended up having to wedge a razor blade between the heatsink and the CPU, and almost snapped the blade before it finally gave, the paste had dried to a solid state and effectively bonded my CPU to the H100.

I looked at the syringe filled with the paste, and it had also dried up completely, it had been sitting safely sealed in a cool drawer and was still in its original package and had its cap tightly closed.

I believe the gallium eventually vaporized in less than a year and left nothing but whatever base metal is used to create the alloy, I decided not to use it anymore as this thing was really expensive and had all gone to waste in less than a year! I only used it twice and had to throw almost all of it away, it almost killed my board and CPU too! :(

I realize it may be much harder to bond the naked silicon on a delided CPU to a heatsink, but that, besides the high risk of even the smallest particle causing a short, makes for a deadly combination... :p
 
That's really sad, using the vise method is probably safer, well, probably your next 4770 will have better thermals :)

You know, I used coolaboratory's liquid pro in the past, and it has excellent thermal properties, but once I almost killed both an Asus MIVE and a 2600K CPU because of it, I was moving these parts to a new case and was about to remove the H100 cooler from the proc but it was bonded to it, I tried gently twisting it but I stopped when I realized I was using too much torque and would probably rip the whole socket from the board.

I finally decided to remove the CPU along with the heatsink, it was almost as if it had been soldered to the cooper plate, I ended up having to wedge a razor blade between the heatsink and the CPU, and almost snapped the blade before it finally gave, the paste had dried to a solid state and effectively bonded my CPU to the H100.

I looked at the syringe filled with the paste, and it had also dried up completely, it had been sitting safely sealed in a cool drawer and was still in its original package and had its cap tightly closed.

I believe the gallium eventually vaporized in less than a year and left nothing but whatever base metal is used to create the alloy, I decided not to use it anymore as this thing was really expensive and had all gone to waste in less than a year! I only used it twice and had to throw almost all of it away, it almost killed my board and CPU too! :(

I realize it may be much harder to bond the naked silicon on a delided CPU to a heatsink, but that, besides the high risk of even the smallest particle causing a short, makes for a deadly combination... :p

I hope the next chip is better, unless I have been doing something wrong this whole time, which I doubt. My 2yo 2600K still hums along in the 70c range but sadly it lacks the performance and energy consumption improvements the newer chips have. It would be nice if we were eventually stuck with BGA chips if they can be lidless to make up for the fact that they are permanently attached to your board.
 
Found this good video talks about how Haswell OCing work's, I found it helpful since I will be going from Core 2 Quad to Haswell.

Sounds like most will get 4.5(1.25-1.31v) and thats about it unless they got good cooling.

 
Last edited by a moderator:
(I was thinking about starting a new thread, but I just put it here:)

What do you guys think about the 4770? Did any of you had the chance already to see how far it can be pushed. I know Intel is locked it down like the i3s, but still, I'm curious if any of you experimented with the chip.

I'm asking because I was reading a lot about TSX lately (this article for example), and I kinda like it tbh.
The i7-3770 was no problem to run at 4.3GHz in p67/z68/z77 boards, but I'm skeptical about the Haswell chip tbh.
 
(I was thinking about starting a new thread, but I just put it here:)

What do you guys think about the 4770? Did any of you had the chance already to see how far it can be pushed. I know Intel is locked it down like the i3s, but still, I'm curious if any of you experimented with the chip.

I'm asking because I was reading a lot about TSX lately (this article for example), and I kinda like it tbh.
The i7-3770 was no problem to run at 4.3GHz in p67/z68/z77 boards, but I'm skeptical about the Haswell chip tbh.

I'd imagine it should overclock as well as the 4770Ks, but I could be wrong. We now have 125/166/250MHz BCLK straps with Haswell, so setting the BCLK to 125 and the multi to 34 will get you a solid 4250MHz. If you wanted to go all out and force the maximum turbo multiplier, it would do a whopping 4875MHz at 125 BCLK. I'm almost tempted to buy the locked 4770 and just BCLK overclock the darn thing as it's $30 cheaper and has more features unlocked but has a locked multiplier. Would love to see if they OC as well or not.
 
(I was thinking about starting a new thread, but I just put it here:)

What do you guys think about the 4770? Did any of you had the chance already to see how far it can be pushed. I know Intel is locked it down like the i3s, but still, I'm curious if any of you experimented with the chip.

I'm asking because I was reading a lot about TSX lately (this article for example), and I kinda like it tbh.
The i7-3770 was no problem to run at 4.3GHz in p67/z68/z77 boards, but I'm skeptical about the Haswell chip tbh.

Haswell design is rated to 154 W max. Chips are sold at speeds rated for a max of 84W so far, so that's so big headroom. I estimate that'll be approximately what 5.0 GHz-capable chips are pulling at high loads, based on what few samples I've had. and what OEMs and users are reporting so far.

The question remains if you can cool it effectively to the point that that 154W threshold can be reached without thermal throttle.

4.3 GHz should be no problem, and I believe is that is the actual target for these chips. Some will go higher, but none should do less. They'll get binned to other chips first.

I'd imagine it should overclock as well as the 4770Ks, but I could be wrong. We now have 125/166/250MHz BCLK straps with Haswell, so setting the BCLK to 125 and the multi to 34 will get you a solid 4250MHz. If you wanted to go all out and force the maximum turbo multiplier, it would do a whopping 4875MHz at 125 BCLK. I'm almost tempted to buy the locked 4770 and just BCLK overclock the darn thing as it's $30 cheaper and has more features unlocked but has a locked multiplier. Would love to see if they OC as well or not.

BCLK adjust only on "K" parts. Even the 4-multi boost that was available on past generations is gone.
 
BCLK adjust only on "K" parts.

Oh wow, thank you for the confirmation. 4770K it is (again). Is it just the straps that are unavailable on non-Ks or is the BCLK utterly locked down to 100.0MHz? Sorry if that's a dumb question. :laugh:
 
Oh wow, thank you for the confirmation. 4770K it is (again). Is it just the straps that are unavailable on non-Ks or is the BCLK utterly locked down to 100.0MHz? Sorry if that's a dumb question. :laugh:

According to "info"(not me, in other words), non-K parts are 100% frequency-locked, and maybe even memory clocks, too. Users that normally buy these chips don't normally use those features anyway, so I see it as a non-issue.
 
According to "info"(not me, in other words), non-K parts are 100% frequency-locked, and maybe even memory clocks, too. Users that normally buy these chips don't normally use those features anyway, so I see it as a non-issue.

Dayum, I'm glad I asked. Glad I didn't buy one and figure out that it was locked down like a pre-built computer, and then I wouldn't be able to return it. Thanks for the heads up. :toast:
 
Dayum, I'm glad I asked. Glad I didn't buy one and figure out that it was locked down like a pre-built computer, and then I wouldn't be able to return it. Thanks for the heads up. :toast:

That's the rumor, coming from Anandtech, I think. Personally, I don't believe anything until I try for myself.
 
That's the rumor, coming from Anandtech, I think. Personally, I don't believe anything until I try for myself.

No worries. Even if it is a rumor, I'll just invest in another 4770K and not worry about the problems I could potentially face whether it be with overclocking or difficulty selling a locked processor later. I'll likely order my replacement processor on Monday and then we can start where we left off with the overclocking stuff. ;)
 
Thank you for the reply. I would personaly believe you (because of your great reviews) if you could share why do you think they will do "4.3 GHz...no problem" ?

Common reports from OEMS testing retails, similar results from those that have purchased since launch and my own experience. Understanding the architecture goes a long way here, I think. I don't think everyone fully understands that yet. I know I don't, but I got some tasty documents. :p

Also, I knew what Haswell was and what they would do months and months ago. Ask W1zz if what I told him 6 months ago didn't turn out to be true. Not that I really knew anything secret...but nothing in this launch should have really been a surprise. What does come as a surprise to some out there is the difference between retail and ES parts. I refuse to offer any info in regards to that subject, but I will point out that it's something being discussed already, although I see/have received conflicting reports of what's really different. As much as people want to add mystery to overclocking, it's just math, fundamentally, so I don't get what the big deal is, really.
 
Hey Dave any time frame on the rest of your mobo reviews...really looking forward to the mpower as that is what I currently have.

Did you receive that lambo branded oc formula yet?
 
Common reports from OEMS testing retails, similar results from those that have purchased since launch and my own experience. Understanding the architecture goes a long way here, I think. I don't think everyone fully understands that yet. I know I don't, but I got some tasty documents. :p

Also, I knew what Haswell was and what they would do months and months ago. Ask W1zz if what I told him 6 months ago didn't turn out to be true. Not that I really knew anything secret...but nothing in this launch should have really been a surprise. What does come as a surprise to some out there is the difference between retail and ES parts. I refuse to offer any info in regards to that subject, but I will point out that it's something being discussed already, although I see/have received conflicting reports of what's really different. As much as people want to add mystery to overclocking, it's just math, fundamentally, so I don't get what the big deal is, really.

Thanks.

I actually had a few hours and went to test things.

Max multi was x39 (sync all cores)
Max BCLK freq was 105, and anything higher made the system hang right after the post (interesting enough this option was not locked in the bios).

Tested two chips, same results. 4.3Ghz would be enough for me, but I'm not sure about 4.1 tbh. I know it's not much of a difference, but you know how it is, sometimes you only need that extra 2-3 fps for happyness;)
 
Back
Top