Testing Ivy Bridge: Intel Core i7-3770K Reviewed

By on April 23, 2012, 11:00 AM

Since late last year Ivy Bridge seems to be the architecture everyone is waiting for. Although Intel is only anticipating a 10–15% processing performance bump when compared to Sandy Bridge, the big news come in the form of improved graphics and efficiency, thanks to the move to a 22nm design process using new 3D transistor technology we recently explained in detail here.

Today the company is unveiling its full new line of Core i7 and Core i5 processors, accompanying chipsets and Centrino wireless options. Ivy Bridge is a 'tick' release, but Intel is calling it a tick+ due to the more significant overhaul the graphics side of things is getting. The new chips are set to provide 20–50% better GPU performance over Sandy Bridge, the kind of jump we'd normally expect from a tock release.

Having already discussed the new Tri-Gate transistors in great detail, the new 7-series chipsets, and some of the motherboards that use them, we are going to focus primarily on the Core i7-3770K processor in this review.

Read the complete review.




User Comments: 48

Got something to say? Post a comment
slh28 slh28, TechSpot Paladin, said:

Holy crap... 1.52V on air! What were the temps like?

I think overall the performance and power consumption are pretty much as expected... nothing to get too excited about and looking at the gaming performance I still don't have an urge to upgrade from Bloomfield.

EEatGDL said:

No comments needed -the review stands by itself. It is a little bit more than I expected in some scenarios, anyway, I was about to upgrade in june from my Allendale (C2D) to a 2500K, so now I'm going for the IB processor I can buy with the money formerly destined to a 2500K. Only in mobo, case, processor, cooling, and RAM I'll invest $700 keeping my HDDs, 9800 GT, and optical drive for a while; I think on december I'm going for new graphics card. It's settled.

Guest said:

Many applications are only 5% faster. the GPU is better but still too bad for gaming, Quicksync is a lot better and power consumption under load is better. More of a minor update really, not worth upgrading from Sandy Bridge if you're already there.

abysal abysal said:

Temperatures anyone? wtf?

Guest said:

LOL!!

1.52V is a little scary.

I'm sure techspot might be aware of crazy temps, and they choose not to disclose it so they wont deter any customers from buying it.

I think I'll keep my 2600K at 4.9Ghz with 1.355V

Thanks though!

Vrmithrax Vrmithrax, TechSpot Paladin, said:

I'd have been interested to see a power consumption comparison between Ivy Bridge and the A8 in gaming with integrated graphics. More out of curiosity, just wondering how well that power efficiency scales when the IGP is in heavy use as well.

---agissi--- ---agissi---, TechSpot Paladin, said:

Im most interested in how IB can OC in comparison to SB, since its typical performance improvements are minor. If the chip can OC to 5GHz on average, with manageable temps on air, then thats news I want to hear! /off to google

captainawesome captainawesome said:

Can't wait to see how Trinity compares in a few weeks time.

Guest said:

The 3570T looks nice. I'm using an Athlon X3 455 in my file server, but it's got like a 65-90 Watt TDP... going to a 3570T should reduce power/heat in my system... might have to look into this.

Guest said:

This review is useless without telling us what temperatures you got with your overclock. I'm guessing they don't show them because the don't want to scare potential costumers when they see the cpu at over 100 degrees celsius!

Guest said:

A6 and A4 were not included, I'm pretty sure that the on die gpu graphics beat the intel integrated graphics, also the temperatures... you just hide it because of the fact that overclock becomes quite hot in this processors.

EEatGDL said:

This review is useless without telling us what temperatures you got with your overclock. I'm guessing they don't show them because the don't want to scare potential costumers when they see the cpu at over 100 degrees celsius!

The only ones hiding here are both of you "guests", it looks like you haven't seen any other review here, as you can see, in the last Bulldozer review, there was no temperatures chart neither in stock frequency or overclocked, so your point's not valid about "don't wanting to scare potential consumers". By fact, the review won't be extended just because you ask, and has nothing to do with Intel biasing.

Guest said:

The only ones hiding here are both of you "guests", it looks like you haven't seen any other review here, as you can see, in the last Bulldozer review, there was no temperatures chart neither in stock frequency or overclocked, so your point's not valid about "don't wanting to scare potential consumers". By fact, the review won't be extended just because you ask, and has nothing to do with Intel biasing.

Why does the fact that I don't have a user here change the fact that the review is not complete in it's current state? - no matter what a Techspot review usually contains. Every other professional site, and even some lousy sites, have temperatures in their review because it's as much a part of the review as anything else, except Techspot who, for some unknown reason, apparently thinks it doesn't matter. I did not in any way mean that Techspot was biased towards Intel. But TTL said on the TTLcostums Youtube channel that Intel tried to stop, or at least delay, him making his video about IVB because he was giving actual data on the insanely high temps, which was what I implied also happend here on Techspot.

1 person liked this | dividebyzero dividebyzero, trainee n00b, said:

Every other professional site, and even some lousy sites, have temperatures in their review because it's as much a part of the review as anything else.

Nope. Anandtech, Guru of 3D, HardOCP, Bit-tech, Tech Report for example....and the reason is fairly obvious. Motherboard, setup (chassis or open air test rig), and most importantly, cooler choice play too much a part in the variation between CPU's and CPU architecture. The "your mileage may vary" scenario comes fully into play - some motherboard manufacturers set up their BIOS's to gain from "stock/turbo" settings from the get-go, some boards don't clock without adding extra voltage for stability, and unless you're using the stock heatsink fans, how do you evaluate CPU temps across different platforms, using different settings and different cooling? [link] ...unless you're of the school that believes that Bulldozer runs 7

C over ambient (stock clock) or 12C over ambient when OC'ed to 4.76

The take home message for Ivy Bridge is [link] , where it falls away is when the OC is ~4.7G and above (surmised as a trait of the tri-gate transistor)-unsurprising that IB's introduction mirrors pretty much every other CPU, in that as the process matures the average OC increases. Early release Sandy Bridge was no different -and sites are now comparing early revision IB with late revision SB. Most seem to forget that many early SB samples had trouble reaching 4.7, whilst samples produced from the latter half of 2011 onwards are generally 4.9-5.0 capable.

A6 and A4 were not included, I'm pretty sure that the on die gpu graphics beat the intel integrated graphics,

Probably offset by the fact that Intel's QuickSync is currently the best graphics switching solution out. I guess Intel can take consolation that while they don't have the graphics horsepower (and I'm pretty sure Haswell will make up for that deficiency), they have solid wins in CPU performance, performance/watt, performance/$ and performance/mm. I don't think anyone is under the impression that Intel's HD 4000 is anything more than a means to get an image onto a screen. Likewise I don't see Llano in any serious gaming, business or workstation machines

But TTL said on the TTLcostums Youtube channel that Intel tried to stop, or at least delay, him making his video about IVB because he was giving actual data on the insanely high temps.

It's TTL (Time To Live) Customs. And the guys name is (Tiny) Tom Logan. His site also (OC3D) [link] , so maybe the issue isn't as all-important as you seem to think..

1 person liked this |
Staff
Steve Steve said:

Holy crap... 1.52V on air! What were the temps like?

I think overall the performance and power consumption are pretty much as expected... nothing to get too excited about and looking at the gaming performance I still don't have an urge to upgrade from Bloomfield.

When overclocking I only took a quick look at the temps and they were around 85 degrees using the Prolimatech Super Mega. I didn't bother reporting these temps because I had nothing to compare them with and I honestly didn't know how accurate the reads from AIDA were.

Temperatures anyone? wtf?

The title is "Testing Ivy Bridge: Intel Core i7-3770K Reviewed" not, "Ivy Bridge Overclocking Explored" so I have no idea why you guys are expecting in-depth overclocking results. When testing new CPU's we simply include some overclocking results if we have time, it's hardly in-depth.

I'm sure techspot might be aware of crazy temps, and they choose not to disclose it so they wont deter any customers from buying it.

Yes you have it all worked out :S

- Finally thanks once again dividebyzero for using your skills as a trainee n00b to bring some common sense to the thread

DanUK DanUK said:

Great review, thanks for the info!

This looks like a solid CPU, impressive to see that level of performace with such a small footprint.

For me however I will hold off because my i7-920 still suits all my needs just fine.

Sarcasm Sarcasm said:

All in all great processor, but not an upgrade for current Sandy Bridge owners.

LinkedKube LinkedKube, TechSpot Project Baby, said:

Here, I'll help you. The temps were somewhere in between room temp and the surface of the sun! But it still works...so

cliffordcooley cliffordcooley, TechSpot Paladin, said:

The 3770K would actually be a downgrade from a Sandybridge...
I watched the video to see where you were coming from. I'm not sure how you can make that statement, when the video was not a comparison video. They were generalizing the characteristics between Ivy Bridge and Sandy Bridge but never once compared their performance. You come here making a claim and the guy in the video during the conclusion made it quite clear he wasn't willing to make.

amstech amstech, TechSpot Enthusiast, said:

I would love to see the i7 920/930/950 @ 3.5Ghz against the i7-3770k.

I have a i7 930 @ 4.0Ghz and results like these make me happy I haven't upgraded yet.

With Intel firmly in control of the market and no one pushing thier architecture but themselves, they are free to release newer CPU's at a slow, unmotivated money eating pace.

dividebyzero dividebyzero, trainee n00b, said:

Looks like we have a "Guest" spamming the board for OC3D- thought they wouldn't need the traffic after their last giveaway competition.

I would love to see the i7 920/930/950 @ 3.5Ghz against the i7-3770k.

I have a i7 930 @ 4.0Ghz and results like these make me happy I haven't upgraded yet.

[link]

@ Linked Kube: Your avatar looks like prOn for Q*Bert

/shows age

//shows self out

abysal abysal said:

^ This.

It is assumed that temperature results are part of a CPU review. Especially with the temperature controversy surrounding Ivy bridge. Please do not act ignorant. And please understand I appreciate the time spent to do the review, it was still somewhat helpful, although omitting the temperature results was disappointing.

dividebyzero dividebyzero, trainee n00b, said:

It is assumed that temperature results are part of a CPU review.

Maybe they should be...if the review used the stock heatsink fan that comes packaged with the retail CPU. Know how many reviews used the stock HSF? Two. Hardware Secrets and Hot Hardware out of ~40 reviews that took an in depth look at the 3770K.

You could also argue, I suppose, that stock clocked + stock heatsink fan temps should be included. Now, I'd take a couple of points into consideration:

1. What workload do you use to load the CPU that provides equal loading across cores (including turbo), is indicative of real world usage, takes into account the differing Vcore of individual CPU's, the differing board requirements for stable voltage, and represents a standard for chassis cooling?

1a. How do you compare systems that don't utilise the same HSF ? (AMD systems, SB-E ). [link] with the same CPU.

2. How many 3770K users are going to be using the CPU at stock clocks and be using the useless chunk of aluminium that takes up most of the box.

So, how do you make any relevant comparision when you have a raft of differing OC's, voltages and cooling? Just as a cross section of the reviews I can see:

5.0-5.05GHz (voltages range from 1.35-1.50v) Lab501 (NH-D14 cooling), Hardware Heaven (H100), Bit and Chips (unknown)

4.9GHz (voltages range from 1.35-1.52v) TechSpot (Super Mega), HardOCP and BSN (watercooled), Guru3D (Noctua ?), OCLab.pl (NH-D14), KitGuru (Phanteks TC14PE), Tech Report and Hardwareluxx (unknown)

4.8-4.85GHz (voltages range from 1.26 -1.50v) Two reviews used watercooling, one used the Intel RTS2011LC, 2 were unspecified, the others used air coolers ranging from entry level Cooler Master to the not-available-everywhere Alpenfohn K2.

You have temps ranging from 75C for a Corsair A70 air cooler (4.8 @ 1.35v at Hexus) to 88C for a better air cooler in the Havik 140 running slower with less voltage at X-bit (4.6 @ 1.2v)

1 person liked this |
Staff
Steve Steve said:

Again dividebyzero proves that he knows what he is talking about. Ohh and for the record we were not supplied with an Intel heatsink which is why I tested with the Prolimatech cooler.

@abysal - I am not acting ignorant, this is the same format we have used to test new CPU's on launch day for a long time now and I see no reason to change this. If we post an Ivy Bridge overclocking article then yes you can expect to see detailed info on cooling setups and temperatures.

1 person liked this | dividebyzero dividebyzero, trainee n00b, said:

Thanks Steve. I keep this up I'm giving myself a promotion...the days of trainee n00b will be behind me- long live Complete n00b !!!

Prior to the official launch I was reading the review that awardfabrik put up. They compared sixteen IB's:

For overclockability, voltage, temp etc. (using some very reasonable kit):

And found what most people expected: Namely, that the variance in a launch CPU step/revision is all over the place -since backed up by the mainstream reviews. Amazing how short some people's memories are- I have distinct recollections of exactly the same lack of standardization for Sandy Bridge, Bloomfield and Bulldozer.

For sure IB runs hotter once you start getting above 1.25-1.3v, but from what I understand that is a product of transistor density (so people should get used to it unless they plan on being on 45/32nm forever). Intel have obviously taken some of this into account since IB's Tjuncture max (throttling) is now 105C -up from Sandy's 98C.

Interested to know what Intel's "safe 24/7" core temp is, although with thermal throttling being so pervasive it probably makes it largely a moot point

2 people like this | abysal abysal said:

Again dividebyzero proves that he knows what he is talking about. Ohh and for the record we were not supplied with an Intel heatsink which is why I tested with the Prolimatech cooler.

@abysal - I am not acting ignorant, this is the same format we have used to test new CPU's on launch day for a long time now and I see no reason to change this. If we post an Ivy Bridge overclocking article then yes you can expect to see detailed info on cooling setups and temperatures.

Perhaps ignorant was too harsh a term, my apologies, as I had not considered that you did not receive a stock heat-sink for a base line temperature reading. I appreciate the fact that a temperature reading from a 3rd party heat-sink fan combo could be misleading, so it makes sense to do a separate section regarding overclocking and temperatures. It's just you guys had some of the first reviews on Ivy Bridge and I found it odd that temperatures were not mentioned. Either way I'm looking forward to a more in depth review of the new chip.

Carmaine said:

Again dividebyzero proves that he knows what he is talking about. Ohh and for the record we were not supplied with an Intel heatsink which is why I tested with the Prolimatech cooler.

Still a good review.

Given that we now know the heatsink used, can you give us a peek at the temps you guys encountered at 4.92 GHZ on 1.52V?

I can't speak for the majority here, but I'm one of the few that uses the intel stock heatsink. I'm sure the one used in this review is alot better than what I have at the moment.

Sarcasm Sarcasm said:

Again dividebyzero proves that he knows what he is talking about. Ohh and for the record we were not supplied with an Intel heatsink which is why I tested with the Prolimatech cooler.

Still a good review.

Given that we now know the heatsink used, can you give us a peek at the temps you guys encountered at 4.92 GHZ on 1.52V?

I can't speak for the majority here, but I'm one of the few that uses the intel stock heatsink. I'm sure the one used in this review is alot better than what I have at the moment.

Yes it's rarely that people will use the stock heatsink, but there are people who do if they don't overclock. But considering how insanely easy it is to overclock a K-series cpu, it's almost a crime if you don't do it.

And really, a $25 hyper 212 plus is all that most people will need for a decent overclock which yields a pretty good performance gain.

Sarcasm Sarcasm said:

^Strange, my "quote" didn't come up correctly.

dividebyzero dividebyzero, trainee n00b, said:

Yes it's rarely that people will use the stock heatsink, but there are people who do if they don't overclock.

Probably better off saving a few bucks and opting for a "non K" CPU in that instance. The stock heatsink is still obviously rated to 3.9GHz (the 3770K's turbo clock) so even non-overclocked has a few horses under the hood

And really, a $25 hyper 212 plus is all that most people will need for a decent overclock which yields a pretty good performance gain.

True. But for comparative testing-which is where the discussion seems centred, you'd need to ensure that all review samples were using the same cooling method and that won't happen. Reviewers use the kit that vendors send out to them, and those components differ from reviewer to reviewer depending on target audience, geographic distribution and perceived standing in the tech community. You won't find Tweakers.net or Madshrimps testing a CPU with a $25 cooler for instance.

The temperature discussion seems to lie in the fact that Sandy Bridge (on a mature 32nm process- remember that Clarkdale launched Jan.2010) is such a stellar performer - and IB doesn't stack up well in comparison in temps once you breach 1.3v - (somewhat of a straw man argument when most users won't use that kind of voltage in attaining a decent overclock) Moreover most reviews pretty much neglected every other CPU family except Sandy Bridge in comparitive reviews.

To me the whole thing seems a non-issue, since IB ,as has been stated numerous times by numerous sources, is not an upgrade for Sandy Bridge users- the target market is those who held off upgrading over the last 2-3 architecture generations

hahahanoobs hahahanoobs said:

Still a good review.

Given that we now know the heatsink used, can you give us a peek at the temps you guys encountered at 4.92 GHZ on 1.52V?

He said he "took a quick glance" at temps and it read 85c. He said it in the comments, don't know why he didn't add it to the review.

Their are numerous sites out there that have overclocked and shown their temps using 3rd party coolers. Give thetechreport.com a look. i love those guys over there... and they have an answer from Intel about the high temperatures.

pgianni8 pgianni8 said:

Question for the Tech Geeks:-); i7 3930K or 3770K if you plan on doing mild overclocking, and running multiple applications and charts, Im questioning whether the $599 price is worth it, given IvyBridge smaller die shrink and similar numbers to 3960.

cliffordcooley cliffordcooley, TechSpot Paladin, said:

If you are in question as to whether you need a 3930K, then chances are you don't need it.

dividebyzero dividebyzero, trainee n00b, said:

Question for the Tech Geeks ; i7 3930K or 3770K if you plan on doing mild overclocking, and running multiple applications and charts, Im questioning whether the $599 price is worth it, given IvyBridge smaller die shrink and similar numbers to 3960.

This doesn't make a whole lot of sense to me. If you're looking at a straight-up comparison and perf/price is the metric for comparison, then the natural choice is the 3820 not 3930K, since the 3830 is already fairly close to the six core in performance

3820.....4C/8T...10MB L3...3.6GHz base freq...3.9GHz turbo....$305 retail ($310 at the Egg atm)...$294 (OEM)

3770K..4C/8T......8MB L3...3.5GHz base freq.. 3.9GHz turbo....$313 retail (MSRP)

Guest said:

holy scomoly! that dude will melt my copper heatsink!

pgianni8 pgianni8 said:

If you are in question as to whether you need a 3930K, then chances are you don't need it.
Hi Clifford, not asking about need. The answer has always been NO for 95% of users for the past 6 yrs. lol! Indulge me for a second as far as overclocking, and purely a performance standpoint would you spend the extra for the 3930K vs 3820, do you think the benchmarks would be very similar or would the extra two cores and extra threads be a big advantage.

captaincranky captaincranky, TechSpot Addict, said:

I can't speak for the majority here, but I'm one of the few that uses the intel stock heatsink. I'm sure the one used in this review is alot better than what I have at the moment.
Yes, you can certainly state that with conviction.

I'm not sure which model Intel CPU you're running, but there are a few low priced avenues to upgrade, a couple using stock parts.

If you have a 65 watt TDP dual core CPU, (socket 775), latching on to an Intel stock cooler from an earlier Prescott P-4, will net you a bunch more mass in the assembly, Some early coolers, at least I've been told, have copper cores. Plan "B" might be the Xigmatec "Apache" coolers. (15 bucks tops). These are similar to Zalman "Orb" coolers, and have 92 mm PCM controlled fans, much bigger than stock Intel @ 82 mm (?) . So, they're a bit quieter.

The best current route for you to cooling bliss, would be one of the Cooler Master "Hyper 212" series. One of these should run you no more than $25.00.

I can attest to their efficacy with the Core i3 530 (stock clock) machine I'm using (lightly) ATM.

TDP peaked out at 73 watts, and core 0 ran all the way up to 29 degrees C.... The CPU idles around 27 watts. For all intents and purposes, the Hyper 212 keeps this CPU cooler than the HDDs in the same case.

dividebyzero dividebyzero, trainee n00b, said:

Well, it looks as though the issue with Ivy Bridge temperature scaling with voltage has been isolated. Intel, in their wisdom used TIM to attach the heatspreader to the die package -as opposed to the solder usually used....and the TIM used doesn't seem of the best quality judging by [link] . (Use your favourite translator if not adept in Japanese).

Quick summary (in English) at TPU

cliffordcooley cliffordcooley, TechSpot Paladin, said:

In other words the temperature barrier was purposely engineered by Intel to prevent a huge over-clock.

captaincranky captaincranky, TechSpot Addict, said:

Well, it looks as though the issue with Ivy Bridge temperature scaling with voltage has been isolated. Intel, in their wisdom used TIM to attach the heatspreader to the die package -as opposed to the solder usually used....and the TIM used doesn't seem of the best quality judging by [link] . (Use your favourite translator if not adept in Japanese).

Quick summary (in English) at TPU

So does this make just grabbing a i7 2600K instead of of the 3770K a really good idea?

Guest said:

Fortunately I just learned Japanese last week, so the real question is can we just remove the heat spreader and pit the heatsink copper baseplate directly on the chip to achieve maximum heat transfer?? Can someone whos tried it post a step by step

cliffordcooley cliffordcooley, TechSpot Paladin, said:

Fortunately I just learned Japanese last week, so the real question is can we just remove the heat spreader and pit the heatsink copper baseplate directly on the chip to achieve maximum heat transfer?? Can someone whos tried it post a step by step
Lets not forget about the thickness of the heat-spreader. If removed there would have to be compensation somewhere to make up for the removal of the heat-spreader. Definitely a delicate process I wouldn't be comfortable doing with one of my CPU's.

captaincranky captaincranky, TechSpot Addict, said:

Wouldn't you have to put TIM on the heat sink and die anyway? It's not like you can just slap a heatsink on without it.

dividebyzero dividebyzero, trainee n00b, said:

Wouldn't you have to put TIM on the heat sink and die anyway? It's not like you can just slap a heatsink on without it.

True enough. You would also need an adequate mounting system that applied a reasonable anount of downforce without crushing the die- Ivy Bridge doesn't present a lot of surface area.

Guest said:

the only thing I wanted to ask is if you maximized the amount of shared memory from your ram to the intel HD 4000 graphics. I've know from using intel's igpu's in the past with maximum DVMT that it definitely shows a big differance. I thought you probably already did that, but I couldn't be sure cause I don't remember you specifying it.

Guest said:

Thank you for writing this Review. It was sufficiently great, I will return to your Site.

Every bit of information gained from various Reviews adds to our knowledge and allows us to make informed decisions - well at least for those with good judgement.

  • If the issue of temperature (versus previous generation) were briefly addressed in the Review and if those requesting the info be provided approached the matter with a different tact then this Thread might have been shorter.
  • The voltage used to OC was too high (out of spec) and thus some were keen to know the temps since they had heard that the new Chips are hot.

Perhaps a one-liner like this might have sufficed:

"The Intel Core i7 2600K was more than 20 degrees Celsius cooler than the Core i7 3770K 'Ivy Bridge' Processor under the same circumstances even though we were running more voltage through it!." (Source: http://www.legitreviews.com/article/1924/8/ ).

A Graphic is helpful for some.

[image link]

Load all comments...

Add New Comment

TechSpot Members
Login or sign up for free,
it takes about 30 seconds.
You may also...
Get complete access to the TechSpot community. Join thousands of technology enthusiasts that contribute and share knowledge in our forum. Get a private inbox, upload your own photo gallery and more.