Tick tock. Do you hear that?
Tick tock. Is that what I think it is?
Why yes, it’s Intel’s clock, ticking and tocking like…um…clockwork. Today we have a big fat tock for you in the form of Intel’s new microarchitecture, code-named Haswell. It manifests itself in our hands as the brand new 4th generation Intel Core processor, the i7 4770K.
The Haswell Architecture, 4th Generation i7
Ivy Bridge ushered in the 22nm process using Intel’s tri-gate 3D transistors. Also clocking in at 22nm, Haswell takes that process and gives it the Intel treatment with a new microarchitecture. The enthusiast / desktop models are quad core, both with and without HyperThreading. Haswell weighs in at 1.4 billion transistors fit in a relatively small 177mm² die.
With Haswell, you’re going to see a big push for energy efficiency. The name of the game right now is mobile. While I like to think the desktop (especially the enthusiast desktop) market is still here for showing the best of the best, and what these CPUs are capable of when you build a balls-the-the-wall system around them. As you can see, Intel did tweak quite a bit with Haswell and, while not earth shattering, should still give some noticeable performance benefits. However, while there is increased performance, energy efficiency appears to have been their main focus.
There are some new compute instructions and they should be well received assuming people use software that can take advantage of them. Sandy Bridge’s AVX doubled the flops/core in both single and double precision calculations from Nehalem’s SSE and Haswell is doing the same with AVX2, doubling Sandy Bridge’s flops/core from 16 to 32 flops/core single precision and 8 to 16 flops/core double precision. In addition to that doubling, AVX2 also doubles the L1 and L2 cache bandwidth, up to 96 bytes/clock and 64 bytes/clock, respectively.
Intel is also trying to make bigger strides in the graphics department. Let’s face it, up to Sandy and Ivy Bridge, Intel had been going through the motions in the graphics department. Their iGPUs could get the job done for your typical web browsing / word processing machine, but anything more and you were making some big sacrifices in the gaming department (turning most effects down or off) just to be able to play.
With Haswell, they’re making a stronger push forward. To that end, they now have on package eDRAM cache, which should assist their graphics processor by helping it avoid waiting for the system RAM for every bit of data it needs. There is only so much you can fit into a relatively small on-die cache, but it’s a step. Intel taking steps forward is important when it comes to their on-die GPUs.
Many of our readers will neither care about nor use their iGPU on an i7 4770K or an i5 4670K (which is probably why they didn’t bother putting Iris or Iris Pro graphics on those CPUs), but integrated graphics are huge in the OEM market, and AMD currently has the lead when it comes to integrated graphics. The CPUs that go along with those graphics don’t necessarily hold a candle to Intel’s raw processing power, but in the graphics department, AMD has had the lead for a while now. Obviously, Intel would like to change that.
This next slide is slightly cryptic, but it’s also very important for you to try and understand how pushing Haswell will go. Starting with Haswell, the voltage controller will be completely on-die. You’ll still need a robust power section to deliver power to the CPU, but the control of that power is now within the CPU itself.
One of the biggest practical changes is that LLC is essentially a thing of the past. The voltage you set will pretty much be the voltage you apply. While motherboard vendors may still include the option, in our testing, the voltage holds constant right where you tell it to (with one notable exception where it actually increases using adaptive voltage control, which we go over in detail in our Haswell overclocking guide).
The takeaway from this slide is that you will have a lot of voltages to tweak. Thankfully “Auto” works pretty well, so other than Vcore, the average overclocker can leave well enough alone, but the tweaker has plenty of voltages to adjust if they want. IOA (Input/Output Analog), IOD (Input/Output Digital) and SA (System Agent) can help with both higher RAM and BCLK settings. The Ring bus is something you’ll want to pay attention to and you’ll see how to address it in our overclocking guide.
Ahh, thermals. Unless you were living under a rock the last year and a half, you know Ivy Bridge ran hotter than Sandy Bridge. Despite much “debate” on many forums, it did not put off more heat than Sandy Bridge, but it did run at higher temperatures. Many thought, as we did, that it had to do with Intel switching from soldering the IHS (integrated heat spreader) to the die to using thermal paste as the interface of choice.
Turns out – and I have it on very good authority from one of the biggest motherboard companies in the world – many of you that came up with a different theory are indeed correct – it’s not the TIM at all. Intel uses a very good, quality TIM, probably better than most of the pastes people switched to when they de-lidded their processors. Rather, it’s the gap between the die and the IHS that helps contribute to higher temperatures.
Due to mass manufacturing requirements (you can only spend so much time on one CPU), that black silicon-like material Intel uses to hold the IHS on is just a little too thick. Not too thick for your CPUs to operate perfectly fine, but too thick for getting the best temperatures you might otherwise be able to attain. It’s removing that layer that helped people get five, ten, sometimes even twenty degrees better temperatures.
That’s not to say Intel is doing a poor job at all, but for those willing to attempt de-lidding their processors, there are gains to be had by doing so and there will likely continue to be. However, temperatures will still be tough on Haswell, with or without de-lidding your CPU. As processor nodes get smaller and smaller, and transistor density gets higher and higher, you’re trying to remove a (relatively) large amount of heat from a very small area. It’s just a fact of life that will become more difficult.
For small form factor PCs and ultra-portables (not the desktop CPUs our audience really uses), that’s where SDP – Scenario Design Power comes in. You can read a slide as well as I can, so go to it and discover SDP for that market.
Now the fun part – an added knob! Intel is giving Haswell CPUs an extra level of control – BCLK ratios. Now, just like Sandy Bridge-E before it, you can adjust your CPU’s BCLK strap for extra tweaking. We’re also happy to report that, unlike SNB-E, the 167 MHz strap is perfectly usable on most CPUs.
With Haswell’s launch, Intel is also unveiling Iris and Iris Pro graphics. These iGPUs are not included on the enthusiast -K CPUs and have their own SKUs.
We won’t be testing integrated graphics for this review, but even if we were, we wouldn’t be testing Iris or Iris Pro. They come in their own SKUs. The i7 4770K and i5 4670K come with an HD 4600 iGPU. We will explore how that compares to the latest AMD iGPU in a review coming out soon.
Since we don’t have one of these chips we can test (you’ll probably see them mostly in OEM machines without discrete GPUs), we’ll just post the slides for you to scroll through.
Now we get to where the rubber meets the road – pricing. The i7 4770K’s price-per-thousand has increased a bit from the Ivy Bridge launch, from $313 for an i7 3770K up to $339 for the new i7 4770K.
Likewise, the i5 4670K’s price is being bumped up, from $212 for the i5 3570K to $242 for the i5 4670K.
Now it’s time to meet the i7 4770K itself!
Meet the i7 4770K
Here is the i7 4770K in the flesh. Like all CPUs, it’s what’s inside that counts.
Being a microarchitecture change (Tock), the package changed a little bit. Of course, there are 1150 pads on the under side rather than 1155 like the Sandy/Ivy Bridge chips. There are also changes to the IHS. The cutout that used to exist for small bits of information on the bottom is no more. The IHS itself also seems just a tad thicker. the rows of contacts on the top (never used by consumer boards) were also shrunken and the keys for socket insertion were wisely moved so the chips won’t fit in each others’ sockets.
On the bottom side, the number of small capacitors has shrunk greatly. There appears to be a slight material difference as well – the pins on Haswell don’t look quite as yellow-gold as Sandy Bridge.
Here is the LGA 1150 socket itself, with and without a CPU installed.
Gratuitous CPU photo anybody?
Of course, even with a built-in memory controller and VRM, a CPU can’t do a whole lot without a good chipset by its side, which brings us to the Z87 Chipset.
The Z87 Chipset and the Intel DZ87KLT-75K Motherboard
Thankfully – and finally – Intel has moved to full SATA 6Gb/s support for all six SATA ports. They have also raised the number of native USB 3.0 ports from four to six and USB 2.0 from ten to fourteen. Assuming your chosen motherboard implements it, Smart Response Technology (where you can set up a small SSD to act as a cache to a larger HDD, thus saving money) is still available.
While we’re here, looking at the CPU connectivity, the graphics interface is of course still PCIe 3.0 and you have the same sixteen lanes available. Official RAM support is only up to 1600 MHz, but as you’ll see, this platform has plenty of room to go higher in that department.
Now we’ll look at the Z87 motherboard Intel supplied to test the new CPU in – the DZ87KLT-75K.
Package, First Look and Accessories
This box is similar to what you’ll see from the new CPUs, except it’s black and the CPU boxes will be blue.
As Intel boards go, they did a good job making this one look good, outfitting the heatsinks with top plates that look pretty nice. I could still do without the skulls they insist on using, but they don’t seem to be going anywhere.
As far as accessories, you get the backplate, an SLI bridge, a Bluetooth antenna, the manual and a mouse pad.
There are plenty of headers to hook up most anything you could want on your case as well as LED indicators showing where your board is in the POST process. Interestingly, there are two POST code indicators; both of which are active. You’ll see various motherboards with two indicators in this generation and some that opt to go with just one. There’s even an ASRock board with a graphical POST display.
Very interestingly, in the photo on the right you can see an ASMedia SATA controller, which not only handles a couple extra SATA ports, but that handy mPCIe/mSATA port you see there.
In the upper right, you have two fan headers, a motherboard speaker and convenient onboard power and reset buttons.
There is a removable BIOS chip (though those sometimes are only on the engineering sample boards) above the SATA headers as well as a USB 3.0 header. I wish all manufacturers would move those to the bottom of the board; this location is bad for case wire management.
Here’s a better view of the ASMedia controller, mPCIe/mSATA connector and nuvoTon sensor module.
For storage, in addition the mPCIe/mSATA port, you also get the six native SATA 6Gb/s ports and two extra ASMedia-controlled SATA 6Gb/s ports.
Sound comes to you courtesy a Realtek ALC 898 CODEC.
There is plenty of connectivity on the rear I/O panel. Surprisingly (and happily for those of us that still use it), they brought back a PS/2 port. The two yellow USB ports are fast charging ports to go along with the six USB 3.0 ports. Interestingly, they included a firewire port next to the two ethernet ports; it’s the interface few people use that refuses to die. The single display output is HDMI and you have the standard optical and analog audio outputs. Last up on the right there is a Thunderbolt port to use the Z87’s built-in Thunderbolt controller.
Under the Hood
We’ll have a quick look at the board sans heatsinks. The heatsinks are stout and will do their job, though you’ll want front-to-back airflow in your case to make sure air gets under those ornamental shrouds.
Here’s a closeup of those nice looking heatsinks.
Interestingly, there are no markings on the Z87 chip itself, unlike its predecessors. It’s not just this board either; the ASUS Maximus VI Extreme I have on-hand also has no markings on its Z87 chip.
The power section is an 8-phase design, with two phases for the iGPU and six for the CPU itself. I didn’t run into any power issues when testing and the heatsinks didn’t get too warm at all. It’s not the strongest power section and I wouldn’t want to push it on LN2, but it’s good enough for the majority of average users.
Lastly, we have the LGA 1150 socket itself to house your brand new CPU.
Overall it’s not a bad motherboard, but the UEFI needs work. It threw me completely for a loop for no apparent reason, giving benchmark results and quirky benchmark runs such that I’ve never seen. It was cleared up after repeated loading of defaults, but what a frustrating experience for a day or so. I’m sure that will be improved, but it certainly didn’t endear me to this board, especially after moving to the Maximus VI Extreme, which is just in a totally different class.
Power consumption isn’t something we pay massive attention to, but there are enough green readers that it matters to, so we do grab the numbers while we’re reviewing. Please take note that the idle wattage may not be representative of a system-to-system comparison! This is the first CPU review that used an HD 7970 instead of the older HD 6970, thus power consumption went up a little bit, which is probably why Haswell’s idle power is more than Ivy Bridge.
|Test Setup||Idle (Watts)||CPU Loaded (Watts)|
|i7 4770K||76 W||126 W|
|i7 3770K||70 W||134 W|
|i7 2600K||97 W||158 W|
|FX-8150||121 W||246 W|
|i7-3960X||104 W||244 W|
|FX-8350||177 W||282 W|
When it comes to loaded wattage pull though, Ivy Bridge takes a backseat to Haswell. That’s no small feat considering the RAM is running at a higher frequency, the GPU is pulling more watts at idle and the CPU is running the same frequency. You have to give it to Intel, they certainly accomplished their goal of more efficiency.
Overclocking for Stability
With the new CPU strap option (and the ability to use at least three of them), overclocking the i7 4770K can be as easy or as involved as you want. In the interests of time, we basically went with the voltage + multiplier method (with reduced RAM speed, which is sometimes required with Haswell when overclocking).
We go over that in detail in the overclocking guide that published at the same time as this review, but quickly, depending on your chip as well as your motherboard, you may have to make memory speed compromises when overclocking Haswell. If you get to a stopping point and can’t go any further, try reducing your RAM speed to the DDR3-1600 to DDR3-1866 range. You can then adjust timings down to make up the lost MHz.
Anyway, our stable 24/7 overclock on this Intel board ended up being 4.8 GHz on the CPU and DDR3-1866 on the memory.
The chip itself is capable of more on the memory and overclocked to the same 4.8 GHz on the CPU, but with a successfully stable memory frequency of DDR3-2600 on the Maximus VI Extreme. Thus, while your chip may require compromises in the memory frequency department, a large part of how much of a compromise will rest on what motherboard you have.
Overclocking Haswell is a fun experience, very much so. You’ll need to watch your temperatures and have good cooling. This CPU was on the same cooling we always have for CPU reviews – a dedicated water loop with a Swiftech MCP35x pump, Swiftech MCR-320 radiator and EK Supreme HF Cu CPU block. As you can see, temperatures were hovering around the 80-85 °C level, so you will need to have a solid cooling solution to push Haswell, just like Ivy Bridge.
We explore overclocking this platform fully in our 3-Step Guide to Overclocking Haswell, so please be sure to head over there after reading this review!
Test System, Opponents and Methodology
We have a good lineup for comparison today. You’ll notice this is trimmed a little bit to present comparisons that are more relevant to today’s market. History is nice and all (which is why we have the i7 2600K in here still), but if a CPU isn’t available, is it really doing you a whole lot of good to continue building up graphs with totally irrelevant processors? We think not, and are going with easy real-world comparisons for you.
One thing to note – the RAM on Haswell was operating at the XMP of the G.Skill TridentX DDR3-2600 kit that was installed. Historically, I”ve run the RAM as fast as it would go on the Intel board. After Ivy Bridge stopping at DDR3-2133, I was pleasantly surprised that Haswell, even on the Intel board, was capable of strong memory clocking right out of the box.
|CPU||AMD FX-8150||Intel i7 2600K||AMD FX-8350|
|Stock / Turbo||3.6 / 4.2 GHz||3.4 / 3.8 GHz||4.0/4.2 GHz|
|GPU||n/a||n/a||ASUS HD 7970|
DirectCU II TOP
|CPU||Intel 7 3960X||Intel i7 3770K||Intel i7 4770K|
|Stock / Turbo||3.3 / 3.9 GHz||3.5 / 3.9 GHz||3.5 / 3.9 GHz|
|Motherboard||Intel DX79SI||Intel DZ77GA-70K||Intel DZ87KLT-75K|
|GPU||n/a||ASUS HD 7970|
DirectCU II TOP
|ASUS HD 7970|
DirectCU II TOP
All non-graphics benchmarks at stock are run three times with the results averaged. 3D benchmarks, game tests, and overclocked benchmarks are run once. The operating system is fully updated Windows 7 x64 to give the most accurate representation of real-world performance.
Other than game FPS, the results you see below are graphed relative to the Intel i7 4770K’s stock performance. This means that results by the i7 4770K at stock all equal 100% and the other results are graphed as a percentage relative to the its performance. So, for instance, if the i7 4770K scored 200 points on a benchmark and the i7 3770K scored 160 points, on the graph the i7 4770K would be 100% and the i7 3770K would be 80%, meaning in that benchmark, the i7 3770K is 80% as fast as the i7 4770K. For those that enjoy seeing the actual results, they are in parenthesis under each percentage.
Not much more to say about that, other than enjoy the fireworks!
First up, we have the AIDA64 benchmark results. You’ll probably notice a lot of CPUs from that list above missing. That’s because AIDA completely switched up how things were measured in several of their benchmarks, rendering every historical data point useless.
What you’ll see here was actually not benchmarked on the Intel motherboard. In order to have comparison data, the i7 3770K was benched on a Maximus V Extreme, which has “ASUS stock”, using their “Multi Core Enhancement”, which means all cores were pegged at top turbo (3.9 GHz) 24/7. To make this a fair fight, I’m using results from the Maximus VI Extreme, which has the same feature. The RAM on both systems was run at DDR3-2600 / 10-12-12-31, so you have a true clock-for-clock comparison here.
Haswell comes out swinging, beating Ivy Bridge at everything. Two of the five tests have over 11% improvement and coincidentally the other three all have a 1.2% improvement, which I just noticed as I’m typing this.
With the exception of the FPU SinJulia test, Haswell improved on Ivy’s scores in the FPU department too, but not by any large margin at all.
Memory testing was very interesting. Ivy Bridge can read the memory faster – by over 9% – and can do so with slightly less latency. However, Haswell can write and copy (faster, and much faster, respectively) than Ivy Bridge.
Overall, AIDA64 is a win for Haswell, if not a trouncing.
3D benchmarks were also run on the Maximus VI Extreme. To continue making this a fair fight, Haswell was run at our graphics card testing level (4.0 GHz on the CPU, DDR3-1866 / 9-9-9-24 on the RAM), just like the benches on Ivy Bridge were. Both runs used Catalyst 13.4 (AMD was on a later version of 12.xx, after the big performance gain) on an ASUS HD 7970 DirectCU II TOP, so everything was equal between the systems.
Right off the bat, you can see Haswell will have advantages in 3D benchmarking. We’ll show you results with the CPU overclocked in the coming weeks with the Maximus VI Extreme review, but let’s just say anyone that benched 3D with Ivy Bridge and wants to improve their scores would be well advised to move to Haswell. The results aren’t huge (except Vantage), but there is a definite trend of improvement in all of the benches, even graphics-intensive 3DMark: Fire Strike and HWBot Heaven Xtreme.
Like the benchmarks, games saw an improvement across the board. The improvement can be as little as 0.9 FPS and as high as 22.7 FPS, depending on the game. The Metro 2033 result was most surprising.
Haswell looks great in the 3D department. For the casual gamer, upgrading from an i7 3770K might be a stretch, but if you have anything older it might just be worth your consideration.
Rendering, Video & Compression
Now we’ll have a look at the CPU oriented benchmarks, both with and without overclock and run all on the Intel motherboard.
First up is Cinebench R10 and R11.5. Haswell clearly wins out over Ivy Bridge when it comes to image rendering. What’s crazy to me is that the overclocked i7 4770K actually beat the stock i7 3960X (that’s a kilobuck CPU!) in both Cinebench benchmarks. Very impressive.
PoV Ray continues to impress, with strong gains over the 3770K and continuing to beat the stock 3960X when overclocked.
The first pass of the x264 benchmark is a reading pass; no encoding is done here. Gains are generally smaller here than they are in the encoding pass, but gain Haswell did.
Pass 2 does the actual video encoding and Haswell continues to shine over the 3770K – and overclocked continues to beat out the 3960X.
7zip clearly prefers integer cores – the more the merrier. This is the only benchmark that AMD’s FX-8350 actually holds a candle to Haswell. Haswell still comes out on top (except for the 3960X, with its twelve threads), but the gain isn’t huge.
So far, so good. In real-world representative benchmarks, Haswell comes out looking great, often beating the much more expensive stock i7 3960X.
You saw in the AIDA64 results above that the gains over Ivy Bridge in floating point calculations isn’t all that strong. There were gains, but not tremendous ones. Since WPrime and SuperPi are both floating point calculations, you can expect similar gains. To wit, WPrime looks to gain between two and four percent.
A quick note on these graphs: as they are graphed relatively, a lower time equals a higher percentage, so using the WPrime 32M 4770K results in the graph as an example, the overclocked time of 4.869 seconds is 31.9% better than the stock time of 6.423 seconds.
Likewise, SuperPi shows small to medium gains as well, from as little as 0.7% to nearly 6%.
These gains aren’t a large percentage, but for extreme overclockers that want to squeeze every second (or even tenth/hundredth of a second) out of their system, Haswell will be an obvious move. They can re-sell Ivy Bridge to make up for some of the loss and will enjoy pushing Haswell, especially with the new BCLK overclocking potential.
Like the 3770K review, we’re combining the graphs to do a clock-for-clock comparison at 4.8 GHz. Remember when looking at these that Haswell is taking a slight hit by operating its memory at DDR3-1866 / 9-9-9-24 instead of Ivy’s DDR3-2133 / 9-11-9-28.
Overall for real-world and two 3D benchmarks with independent CPU tests, Haswell comes out looking solid, with an average 9.0% improvement. Floating-point driven timed benchmarks were less impressive but still gained 2.7% average. Compared to the Sandy-to-Ivy Bridge improvements, you have about the same average timed difference (2.8% to 2.7%) but a solidly increased real-world/3D improvement (from 2.7% up to 9.0%)
Pushing the Envelope
For this section, I used the more user friendly ASUS Maximus VI Extreme. So that review wasn’t spoiled, I didn’t push every MHz the CPU would give up, but I did run a few quick benchmarks to show potential at the 5 GHz range. The three main benches are SuperPi 1M, SuperPi 32M and WPrime 32M, all at 5.0 GHz.
5.0 GHz SuperPi and short WPrime at around 1.35V sure sounds good to this overclocker. But what’s even more fun than that? Why SuperPi 1M at 5.0 GHz and, wait for it, 192.5 MHz BCLK of course!
So yea, color me impressed so far! Much more to come when this board gets pushed for its review.
Final Thoughts & Conclusion
Haswell is a solid step forward for Intel’s microarchitecture design. Its improvement over the Sandy-to-Ivy Bridge processor shrink is palpable and impressive.
It overclocks well on ambient cooling – especially our sample – but that overclocking level will depend heavily on your particular CPU. Thanks to speaking with the kind folks at ASUS (whose efforts helped make the on-release overclocking guide with excellent tips and guidance possible), we know that the average CPU overclock on Haswell might go down a little bit. There are dogs (4.3 GHz chips) and there are great chips (4.8 GHz+), all at the same voltage. On average, you can expect general overclocks in the 4.5-4.8 GHz range. The dogs are rare, but they do exist.
Speaking of ambient cooling, you’re going to want a solid cooling setup for Haswell. They don’t run much hotter than Ivy Bridge, but you can expect a 5-10 °C increase due to the integrated voltage regulator. As such, if you’re looking to push these chips hard, a very high end air cooler to custom water loop would be something to consider. The folks at ASUS even think single-stage cooling (think the older Lian Li cases with a weaker built-in single stage) is going to make a comeback, and I’m inclined to agree with them.
Extreme overclockers will be happy to know you can expect good things from good Haswell chips. Even mid-line chips will be worth your investment. What was a 6.7-6.9 GHz Ivy Bridge chip (tough to get, took lots of binning and/or expensive to purchase pre-binned) will now be more like a 6.3-6.5 GHz chip (easier to find, meaning less money involved). You’ll still have to bin to get the 6.9-7.1 GHz Haswell chip of course, but most people can be happy if they get 6.5-6.7 GHz chip, especially if you’re mostly looking at 3D benchmarking.
These chips can be resilient too. Folks saw a CPUz validation of a chip with 2.5V put through it. I can finally spill it – that was not a CPUz glitch. Early ES chips did allow that much voltage and they did put that much through it. The thing lives to this day. You won’t be able to put that much through them any more (I think the cap is ~1.9V now), but you can do so with relative confidence. These chips do have cold bugs though, unlike many Ivy Bridge chips, so expect to monitor your pots and keep them around the -140 °C level (higher for worse-clocking chips).
So, worth it for extreme overclockers? Yes, definitely. How about the regular folks? The ones that don’t want to run hard-core cooling? That depends on where you’re coming from. If you already have an i7 3770K and are an ‘average’ overclocker, then you probably want to stick with what you’ve got. The gains are there, they’re real and they’re not tiny, but the frugal overclocker in me says Ivy Bridge folks with 4.5+ GHz overclocks can leave well enough alone.
Those who would benefit greatly from the upgrade will be the people still on Sandy Bridge and older. Anyone still using an i7 920 or even an i7 980X should very strongly consider the upgrade to Haswell. In this reviewer’s humble opinion, it would absolutely be worth the investment.
Intel continues to increase their lead in the CPU world. AMD’s FX-8350 doesn’t really even compare to the i7 4770K’s raw performance, especially when you take into account how efficient Haswell is. Thankfully, Intel is keeping very near the same price point. NVIDIA should take note. Haswell is really all I expected it to be. There was a solid performance boost, overclocking and tweaking is now endless fun and experimentation with the introduction of several (functional) CPU straps. For those that like toying with high speed memory, on Haswell the world is your oyster. If you’re able to keep temperatures down, Haswell can be a very fun platform to toy with and it has performance to best even Intel’s top of the line CPU. All that leads to the i7 4770K being easily Overclockers Approved.
We’re not done yet folks, we have a ton of Haswell-related coverage coming your way. The first expanded coverage starts right now, with our Haswell overclocking guide, so be sure to head over and see what overclocking Haswell is all about!
For my part, I also have an ASUS Maximus VI Extreme in-house and in-testing.
That’s not all though. I also have some great G.Skill TridentX DDR3-2933 memory on hand and let me tell you, Haswell can handle them well!
But wait, there’s more! Our esteemed editor Joe Shields (EarthDog) has an MSI MPower Max on-hand and Dino Decesari (Lvcoyote) has some ASUS channel boards he’ll be working on. You can bet we’ll be working to get even more boards from the likes of Gigabyte and ASRock once we’re done with these. Haswell is here and in a big way. We look forward to bringing you ongoing coverage of this fun platform!