The update to Sandy Bridge brings 22nm fabrication technology, a souped up integrated graphics chip, and a host of minor improvements to LGA1155. Best of all, Ivy Bridge is backwards compatible with series 6 motherboards and doesn’t cost an arm and a leg.
April 23, 2012 by Lawrence Lee
Intel Core i7-3770K
For the past few years, Intel has been switching sockets and updating their
CPU architecture at a frantic pace compared to the days of Socket 478 and LGA775.
Ivy Bridge is a welcome change, an updated CPU architecture that is actually
widely compatible with current hardware, in this case LGA1155/Sandy Bridge motherboards.
Their last major processor launch, Sandy
Bridge Extreme for the newly minted LGA2011 socket, left many regular
users in the cold with its costly chipsets and processors aimed at professionals
and enthusiasts who could afford such extravagance.
LGA1155 was and continues to be extremely successful despite being relegated to being Intel’s mainstream platform. Sandy Bridge processors exhibit very strong performance and excellent energy efficient and are reasonably priced. The integrated graphics in each chip is also a nice bonus too for users who have no need for discrete graphics cards. It’s nice to see Intel push out an update for a popular existing product line rather than jumping straight into their next big thing.
Ivy Bridge quad core die layout.
Being pin-compatible with current motherboards, it’s logical to deduce that Ivy Bridge is not fundamentally different from Sandy Bridge. The die is arranged the same way but the manufacturing process has been shrunk from 32nm to 22nm utilizing Intel’s much lauded Tri-gate transistor technology (it’s explained nicely by AnandTech), the integrated graphics chips has been beefed up with more horsepower, the memory controller has been updated to officially support DDR3-1333, and there’s support for the new PCI Express 3.0 standard. Other minor updates include an improved version of Turbo Boost, updated instruction sets, and a couple of new security features (SecureKey, OS Guard) aimed at mobile/enterprise users. New series 7 chipsets have been released incorporating a native USB 3.0 and PCI-E 3.0, but they’re not required to work with Ivy Bridge. Most Sandy Bridge motherboards will be compatible with a BIOS/UEFI update.
Sandy Bridge vs. Ivy Bridge integrated graphics comparison.
One of the defining features from the Sandy Bridge launch were the integrated
graphics chips. The on-die GPUs offered the usual upgrades like updates to DirectX
10, HDMI 1.4, and improved video playback, but Intel also included their hardware-accelerated
QuickSync video encoding feature and boosted the 3D performance considerably
compared to their previous onboard graphics solutions. The HD
3000 chip on the higher end SKUs was the first Intel integrated graphics
processor we encountered that rivaled entry level discrete video cards.
Ivy Bridge’s HD 2500/4000 graphics is more of an incremental upgrade, offering more four more execution units, DirectX 11, OpenGL 3.1, and QuickSync 2.0 support. The clock speeds remain the same, but multiplier overclocking is supported on all motherboards. Also notable is the ability to use three displays at once, a feature that previously required a discrete graphics card to accomplish.
Our Core i7-3770K sample.
Like Sandy Bridge, the stars of Intel’s opening Ivy Bridge salvo are the enthusiast-friendly
"K" series chips that allow almost unbound multiplier overclocking
but are lacking in enterprise/management features that typically aren’t utilized
by overclockers anyway. The lineup mirrors the Sandy Bridge launch 15 months
ago, with the faster IGP on the i7 and "K" models, and no Hyper-threading
and less cache on the i5’s. The various SKUs also hit the price-points occupied
by current Intel chips. It’s not clear what the fate of Sandy Bridge will be
— either its price will be cut or it will quickly be kicked out of the nest
by Ivy Bridge. Our first Ivy Bridge sample is a Core i7-3770K, which
happens to match the current flagship Sandy Bridge i7-2700K in clock speed (3.5
GHz, 3.9 GHz maximum on Turbo Boost speed) and price.
Though there are no dual core chips being released this time around, those
craving extreme energy efficiency may be sated by a series of low power quad
core models. The familiar "T" and "S" suffixes live on,
branded on CPUs with 45W and 65W TDPs respectively. Similar to previous incarnations,
they employ a clever trick to maximize performance within a low thermal envelope.
The base frequencies are fairly low, but the maximum Turbo frequency is 3 to
8 ranks higher than normal. When fewer cores are active, the clock speed ramps
up considerably more than the standard Ivy Bridge models, making the system
Common CPU Test Configuration:
Common IGP Test Configuration:
Measurement and Analysis Tools
Timed Benchmark Test Details
x264 1080p: Spaceship is a 1080p x264 clip encoded from the
Our main test procedure is a series of benchmarks, timed tests of real-world applications. System power consumption (AC) is measured with a Seasonic Power Angel during these tests (an average of the first 10~15 seconds) as well as at idle, during playback of a 1080p H.264 encoded clip, and during full CPU load. To stress the CPU we use either Prime95 (large FFTs setting) or CPUBurn depending on which
produces higher system power consumption. The AC system power is then later converted to DC.
Certain services and features like Superfetch and System Restore are disabled
to prevent them from affecting our results. Aero glass is left enabled if supported.
We also make note if energy saving features like Cool’n’Quiet and SpeedStep
do not function properly.
Estimating DC Power
The following power efficiency figures were obtained for the
Seasonic SS-400ET used in our test system:
Seasonic SS-400ET Test Results
DC Output (W)
AC Input (W)
This data is enough to give us a very good estimate of DC demand in our
test system. We extrapolate the DC power output from the measured AC power
input based on this data. We won’t go through the math; it’s easy enough
to figure out for yourself if you really want to.
Before we jump into our test results, please note the operating voltages
of the processors tested today as sort of a disclaimer. Different samples of
the same processor run at slightly varying operating voltages which
can affect energy efficiency. Different motherboard models do not apply the exact same core voltage either.
In the last three versions of 3DMark using the default settings, Intel’s HD 4000 graphics scored in the same range as the Radeon HD 6550D, the better half of AMD’s A8-3850 APU. We tested the HD 4000 with both DDR3-1333 and DDR3-1600 memory, but the faster RAM didn’t make as much of an impact as it did with the HD 6550D.
At 1366×768 resolution and details set to low, the HD 4000 slaughtered its predecessor, the HD 3000. It delivered an impressive 65% increase in frame rate in Lost Planet 2 and a 53% boost in H.A.W.X 2. Using DDR3-1333, the HD 4000 was more or less a match for the HD 6550D, but faster DDR3-1600 memory favored AMD’s chip greatly, giving it a substantial lead in two of the three games.
We encountered new lows in CPU usage during Flash and VC-1 Blu-ray playback, but disappointingly high results with an H.264-encoded Blu-ray and Quicktime trailer. The solution, at least for the Quicktime clip, was to switch software. We tried a few different playback apps (we use PowerDVD by default), but only XBMC helped, bringing CPU utilization down to 3%. Unfortunately this didn’t help with Blu-ray playback as it’s not supported by XBMC.
Our H.264 difficulties caused a 7W increase in power consumption by our estimation,
and the energy efficiency during playback was higher than the Core i5-2500K
Sandy Bridge, closer to that of AMD’s A8-3850
GPU/Platform Power Consumption
Running on integrated graphics, the power consumption of our i7-3770K / DZ77GA-70K
combination was rather poor at idle. Most H67 motherboards we’ve tested idle
at ~20W when paired with an i5-2500K
(the i7-2600K is similar),
but the Ivy Bridge platform consumed 7W more. With the CPU on full load, stressing
the GPU increased power consumption by 15W, about 5W more than the HD 3000.
It’s not a lot in absolute terms, but in the world of integrated graphics, it’s
CPU Power Consumption
The energy inefficiency noted earlier appear to be caused by the integrated
graphics. Switching to our CPU test platform with a discrete graphics card,
the i7-3770K used a mere 40W DC, closely resembling Sandy Bridge. H.264 video
playback was more frugal by 5W. The GeForce 9400GT used is an older card and
whatever workload is leftover for the CPU seems to be more efficiently handled
by the i7-3770K and i7-3960X.
Running our synthetic CPUBurn/Prime95 stress test, the i7-3770K actually used less than the i7-2600K but in a real life stress test, video encoding with HandBrake, the Ivy Bridge processor used 8W more. With all the hubbub about its improved transistor technology and lower TDP, our sample’s thermal envelope seems fairly similar to Sandy Bridge.
According to RealTemp, our Ivy Bridge sample hit an average core temperature of 45°C above ambient using a Scythe Kabuto cooler with its fan running at 800 RPM (note the i7-3960X was tested with an Intel liquid cooler with the same fan/pump speed). With load power consumption more or less in line with the i7-2600K, the temperature difference seems a little high.
In our Photoshop test, the i7-3770K came right out of the gate with a win, edging out the i7-2600K by two seconds, corresponding closely to the increase in clock speed.
The i7-3770K took our NOD32 test as well, slipping by the Sandy Bridge Extreme i7-3960 and using much less power doing so. It was on par with the Sandy Bridge chips in energy efficiency.
CPU Performance (Continued)
Ivy Bridge was less impressive in our WinRAR test, barely finishing ahead of the i7-2600K.
The i7-3770K was the king of our iTunes encoding test, completing the task ~7% faster than the i7-3960X and ~9% faster than the i7-2600K, while using the least amount of power.
A 12% performance improvement over the i7-2600K was noted in our TMPGEnc video encoding test, as well as a savings of 2W. It just lost out by a hair to the i7-3960X, but it also used about two thirds the energy.
Our HandBrake encoding test was one of the few areas that AMD’s Bulldozer FX-8150 processor shined, leveraging its eight core design to a rare victory over Sandy Bridge. Sadly it was not to last as Ivy Bridge beat it handily and with a savings of almost 80W.
We arrived at our overall performance score by weighing each test equally (each composing 1/6 of the total). Mathematically, a processor that finishes first in every single test would receive a score of 100.
The i7-3770K ran neck-and-neck with the i7-3960X in most of our tests, but the Sandy Bridge Extreme CPU’s superb HandBrake result enabled it to retain its crown with a total score of 97.4. The Ivy Bridge chip did very well though, garnering a ~9.5% performance bump over the i7-2600K. If the i7-2600K were clocked 100 MHz faster to match, at best the performance difference would still be ~6%.
CPU Energy Efficiency
The i7-3770K used slightly more power than the i7-2600K in a few benchmarks, but as it finished faster in all of the tests, the total power consumed while running our test suite was lower by an impressive 11%.
Looking at energy use only during testing isn’t representative of common use
scenario, so we also determined "average power consumption" by assuming
the system is used half the time for light load activities (an average of idle
and H.264 playback) and the remaining half for heavy load (an average of the
power used running our benchmarks). In such balanced scenario, the i7-3770K
uses just a scant more juice than its Sandy Bridge counterparts, and substantially
less than AMD’s current high-end offerings.
CPU Performance Analysis
To determine performance per watt, we divided the overall performance score by the average power consumption and re-weighted it so the top processor would score 100 points. The i7-3770K came up on top, beating out the i7-2600K by a resounding eight points.
Note: Motherboard pricing data was collected from Newegg’s catalog with the following criteria: retail versions, Asus/Intel/Gigabyte/MSI branded, microATX/ATX form factor, SATA 6 Gbps and USB 3.0 controllers. A price limit of $250 was set on AM3+/LGA1155 boards, while no limit was applied to the LGA2011 models as they are typically premium products.
Another way to access value is to put it purely in monetary terms. As the i7-3770K runs on the same motherboards as Sandy Bridge and it is priced similarly to the i7-2600K/2700K, the cost difference is negligible.
Dividing the overall performance by the platform cost gives us the performance
per dollar (again the results were re-weighted so the best value is awarded
100 points). The popular Core i5-2500K delivers the best bang for your buck
of the bunch, while the Phenom II X4 975 offers fairly good value due to its
low price. The i7-3770K is right up there, 4~5 points ahead of the i7-2600K.
It should be noted that the i7-2700K is similarly priced, but even if its 3%
increase in clock speed generated a linear performance improvement, it would
only bump the score up by about two points. A minimal i7-3960X configuration
will cost you over US$1,200, three times the cost of a i7-3770K setup,
making it an obvious dud unless you’ve got money to burn.
In terms of performance, the new HD 4000 graphics chip is a more substantial
update than the CPU portion of Ivy Bridge, delivering 50~70% higher frame rates
in our gaming benchmarks. Unfortunately needs to be an order of magnitude better
to really make a difference — rendering smooth gameplay is only possible
at lower resolutions. It also can’t quite beat the Radeon
HD 6550D inside AMD’s high-end APU. One nice addition is the ability
to use three displays simultaneously, a milestone for integrated graphics. When
playing H.264 video we encountered higher than expected (but not alarming) power
consumption and CPU usage depending on the playback software used; we chalk
this up to early driver issues.
The i7-3770K and other high power models of Ivy Bridge have a TDP of just 77W
but our sample didn’t really exceed the power consumption of the 95W i7-2600K.
We did find it to be less energy efficient when using integrated graphics. The
GPU used an additional 15W when placed on load, a 50% increase over HD 3000,
but the idle difference was actually bigger. The i7-3770K / DZ77GA-70K
combination had an idle power consumption 7W higher than the i5-2500K
paired with most H67 motherboards. When we switched to our CPU test configuration
using a discrete graphics card with the IGP disabled, the idle power draw was
within 2W of the i5-2500K and i7-2600K running on a P67 board.
As for the CPU, it’s a small step up in performance from Sandy Bridge, breathing
some fresh air into the LGA1155 platform. The Core i7-3770K was faster overall
than the i7-2600K by about 9% (our guess is 7% vs. the i7-2700K) without increasing
the overall power draw. Its overall performance can’t be beat by anything except
an uber expensive Sandy Bridge Extreme. That being said, the increase in CPU
performance won’t make current Sandy Bridge owners envious unless they’re in
a possession of a lowly dual core model.
The i7-3770K’s US$313 price tag allows it to supplant the i7-2600K/2700K,
but it is still a poorer value than slower Sandy/Ivy Bridge offerings. The i5-3570K
is likely to give you better bang for your buck as it should hit the market
at a similar price to the current king of values, the Core i5-2500K.
Our thanks to Intel
for the Core i7-3770K and sample used in this review.
* * *
Articles of Related Interest
Intel DZ77GA-70K Z77 Motherboard: Waiting for Ivy Bridge
Intel Sandy Bridge Extreme: Core i7-3960X LGA2011 Processor
AMD FX-8150 8-Core Bulldozer Processor
AMD A8-3850 Quad Core Desktop APU (updated July 10)
Intel Core i3-2100T & Core i5-2400S Low Power CPUs
Core i5-2400, i5-2500K and i7-2600K CPUs
* * *