Sign in with
Sign up | Sign in
Core i7-3720QM: Ivy Bridge Makes Its Mark On Mobility
By ,
1. Understanding Ivy Bridge's Real Target

More than one year has passed since Intel hit a home run with its Sandy Bridge-based processors, which we first reviewed in Intel’s Second-Gen Core CPUs: The Sandy Bridge Review. The architecture delivered impressive performance at the company's designated price points. And, thanks to impressive efficiency, our follow-up Core i7-2820QM: Sandy Bridge Shines In Notebooks showed the mobile incarnation to be a real winner.



Intel's third-generation Core CPUs, based on the Ivy Bridge design that we first reviewed in Intel Core i7-3770K Review: A Small Step Up For Ivy Bridge, are characterized by a shift from 32 to 22 nm manufacturing and a significant boost to 3D graphics alacrity. In the mobile space, the transition promises to be more pronounced. Augmenting notebook performance, efficiency, and battery life in equal measure, Intel believes its Ivy Bridge architecture is ideal for catapulting the Ultrabook product segment into the mainstream—and perhaps blunting the accelerating momentum of tablets.

Mobile Ivy Bridge Today: Core i7 At 55, 45, And 35 W

Intel continues using its Core i3, i5, and i7 brands to create a good/better/best hierarchy. Unfortunately, for the time being, only the company's Core i7 models are being made available (worse, only three of six planned models are currently listed on Intel's price sheet). Ivy Bridge-based Core i5 and i3 chips will emerge later this year.

Mobile Third-Gen Core i7 Family
CPU SKU
Cores / Threads
Base Freq.
Max. Turbo
L3 Cache
HD Graphics
Graphics Base Freq.
Graphics Max. Freq.
TDP (W)
Price
Mobile Third-Gen Core i7 Family
-3920XM

4/8

2.9 GHz

3.8 GHz

8 MB

4000

650 MHz

1.3 GHz

55

$1096

-3820QM

4/8

2.7 GHz

3.7 GHz

8 MB

4000

650 MHz

1.25 GHz

45

$568

-3720QM

4/8

2.6 GHz

3.6 GHz

6 MB

4000

650 MHz

1.25 GHz

45

$378

-3615QM

4/8

2.3 GHz

3.3 GHz

6 MB

4000

650 MHz

1.2 GHz

45

?

-3612QM

4/8

2.1 GHz

3.3 GHz

6 MB

4000

650 MHz

1.1 GHz

45

?

-3610QM

4/8

2.3 GHz

3.1 GHz

6 MB

4000

650 MHz

1.1 GHz

35

?


The functionality of Intel's Core i7-3920XM, -3820QM, and -3720QM is similar to the already-reviewed desktop Core i7-3770K. The only differences are in CPU clock rate, maximum graphics frequency (which tops out at 1.15 GHz on the -3770K), and rated TDP.

Hyper-Threading is enabled on all of these mobile Core i7s, giving them four-core/eight-thread configurations. The most egregious deviation from Intel's familiar naming convention is that four mobile Core i7s offer 6 MB of shared L3 cache, rather than the 8 MB we'd expect on a Core i7-class processor.

In fact, it's not easy to decipher the meaning of Intel's nomenclature. On the server side, the company makes a deliberate effort to explain the significant of each character. We covered that in-depth on page two of Intel Xeon E5-2600: Doing Damage With Two Eight-Core CPUs. This time around, though, we're really only certain that the i7 translates to a Hyper-Threading-enabled quad-core chip, and that the first alphanumeric character that follows, the -3, indicates the Ivy Bridge architecture. The XM suffix is indicative of the highest-end Extreme model, while QM is a dead giveaway of quad-core performance. The irony there is thick. Intel's branding has become so meaningless that it takes an additional letter at the end of the model name to clarify.

We still expect new Low Voltage (LV) and Ultra Low Voltage (ULV) parts, but the model numbers for those haven't been announced yet, presumably in a move to allow partners like HP the opportunity to sell off some of their extra Sandy Bridge-based inventory.

We covered the architectural details of Ivy Bridge in our aforementioned desktop Core i7-3770K review, and all of that information applies here as well. However, there are three enhancements that distinguish Intel's 7-series platforms from their predecessors, including native USB 3.0, provisions for up to three display outputs, and the ability for board designers to attach a Thunderbolt controller via four processor-based PCI Express 3.0 lanes.

A trio of display outputs is perhaps the most exciting addition in our opinions, overcoming a long-time limitation of integrated graphics that allowed a notebook to drive its own screen and one attached monitor. Now, you're able to use a mobile machine's panel and up to two external displays, improving productivity.

Test Setup

In the desktop world, it's easy for us to use one motherboard and swap multiple processors in and out. That's far less common in the mobile space, where form factors are always designed to support certain thermal profiles, making them far less flexible. Consequently, comparing mobile processor architectures requires notebook PCs with more significantly different configurations.

We attempt to eliminate potential variables that affect power consumption, though. For example, we test using an external display output instead of the notebook's own panel. We standardize on Crucial's 256 GB m4 SSD as the main system drive, and connectivity-related benchmarks are performed using a LAN to eliminate power differences related to the wireless networking subsystem. This doesn't magically isolate the CPUs we're looking at, but it's a step in the right direction.

Test Hardware: Mobile Systems
Processors
AMD A8-3520M (Quad-Core, 1.6 GHz)Intel Core i5-460M (Dual-Core, 2.53 GHz)Intel Core i7-2820QM (Quad-Core, 2.3 GHz)Intel Core i7-3720QM (Quad-Core, 2.6 GHz)
Memory
8 GB DDR3-13338 GB DDR3-10668 GB DDR3-13338 GB DDR3-1600
Graphics
AMD Radeon HD 6620GIntel HD Graphics
AMD Mobility Radeon HD 5730
Intel HD Graphics 3000Intel HD Graphics 4000
Nvidia GeForce GT 630M
Notebook
HP Pavilion dv6-6c35dxLenovo IdeaPad Y560Unknown Clevo modelAsus N56Vm
Hard Drive
Crucial m4 256 GB SATA 6Gb/s
DirectX
DirectX 11
Operating System
Windows 7 Ultimate 64-bit
Graphics Driver
Catalyst 12.4Intel 8.15.10.2696
Catalyst 12.4
Intel 8.15.10.2696Intel 8.15.10.2696
Nvidia 301.24
2. Benchmark Results: PCMark 7

In PCMark 7, the Ivy Bridge-based Core i7-3720QM at 2.6 GHz (with a 3.6 GHz maximum Turbo Boost frequency) outpaces the Sandy Bridge-based -2820QM by ~40%, and it surpasses the Arrandale-based Core i5 by ~275%. The -3720QM's score is even more impressive compared to AMD's A8-3520M, as the Llano-based APU barely manages to outperform Intel's Core i5-460M.

Drilling down, the testing reveals some interesting finds (which actually start to call PCMark's weighting into question). Even though the Core i7-3720QM clearly leads the pack, its performance advantage is evident in only a few scenarios.

PCMark 7 Results
i7-3720QM
i7-2820QM
i5-460M
A8-3520M
Video Playback
23.12 FPS
23.16 FPS
23.14 FPS
22.99 FPS
Video Transcoding
169 531.73 KB/s
17 416.99 KB/s
1828.96 KB/s
2460.25 KB/s
Storage: Gaming
15.22 MB/s
14.87 MB/s
12.13 MB/s
14.11 MB/s
Graphics DX9
22.99 FPS
17.47 FPS
13.29 FPS
4.19 FPS
Image Manipulation
11.55 Mpix/s
9.3 Mpix/s
4.00 Mpix/s
7.41 Mpix/s
Storage: Importing Pictures
26.37 MB/s
26.58 MB/s
25.42 MB/s
24.12 MB/s
Web Browsing
15.17 pages/s
14.24 pages/s
6.7 pages/s
9.34 pages/s
Data Decrypting
145.54 MB/s
101.31 MB/s
39.10 MB/s
49.95 MB/s
Storage: Windows Defender
5.37 MB/s
5.25 MB/s
4.22 MB/s
4.97 MB/s


The Core i7-3720QM particularly shines in tests involving:

  • Video Transcoding
  • DX9 Graphics
  • Web Browsing

Because transcoding is so demanding, Intel built fixed-function hardware into its Sandy and Ivy Bridge-based architectures to address it, and the effect of that hardware is made apparent in PCMark 7, grossly inflating the synthetic's score compared to what you'd get from, say, an SSD or discrete graphics card.

With regard to 3D performance, the HD Graphics 4000 engine delivers ~30% higher results than HD Graphics 3000 in the DX 9 benchmark. Futuremark says that the Web browsing test has a 50/10/40% workload distribution between CPU, RAM, and GPU, making it an interesting measure of productivity.

Even though the AMD A8-3520M has four physical cores (compared to the Core i5-460M’s two cores), PCMark 7 works the Llano-based APU harder than the dual-core chip (indicated by overall CPU usage). And yet, our power consumption graph makes it clear that the APU still uses less power.

The Core i7-3720QM and Core i7-2820QM don't exhibit the same spikes in utilization, likely as a result of QuickSync offloading the demanding transcode workload from the general-purpose compute cores. Nevertheless, their higher performance incurs a greater power consumption penalty, and they both use more than AMD's A8-3520M.

3. Benchmark Results: Adobe Photoshop CS 5

Our Photoshop benchmark consists of four threaded filters applied to a large (~15 MB) TIFF image in a script, which effectively demonstrates the performance of each processor in a real-world application. Naturally, the CPUs best optimized for threaded tend to rise to the top.

Unfortunately, that's not the case absolutely. Both the Core i7-3720QM and Core i7-2820QM do really well, the Ivy Bridge-based design trumping Sandy Bridge. As expected, our dual-core Arrandale-based sample performs a lot worse, eating up almost four minutes to finish the same task. But then, surprisingly, AMD's quad-core A8-3520M eats up an additional two minutes.

Seemingly, sharing a 35 W TDP between graphics and processing cores is a real handicap for the APU. So, let's take a look at how utilization and power consumption come into play. In a mobile environment, those two variables are important as well, after all.

As mentioned, our filters are threaded, which explains why even the fastest quad-core chip is pegged at 100% throughout the test. And the performance picture is once again painted by the lines where each CPU drops back to near-zero, indicating the task is complete.

But it takes a look at the power consumption chart to reveal how much battery life each of these platforms is going to gobble up as it gets its job done. The Core i7-3720QM is a 45 W chip, and its system power use hovers in the 40 W range. The older Sandy Bridge-based part, Core i7-2820QM, is also a 45 W processor. Its system power use spikes much higher, though. The Core i5-460M is a 35 W CPU that nearly matches Ivy Bridge's power use, despite its far inferior performance. AMD's A8, also rated for 35 W, clearly underperforms the other contenders. However, it hovers between 20 and 30 W total power use.

Now, that'd be great if the A8 got its job done faster. Using less power over a long period of time doesn't turn out to be a benefit. In terms of energy efficiency, even the Core i7-2820QM outperforms AMD's A8-3520M because it consumes a little more than two times as much power, but finishes the job in one-third of the time. Ivy Bridge improves on that to an even greater degree.

4. Benchmark Results: WinRAR 4.11

WinRAR is one of three archival apps normally found in our processor reviews. Historically, it's far better-threaded than WinZip (though a recent update to WinZip 16.5 might change this), but it doesn't always scale as well as 7-Zip. It's a good middle-ground, and it remains a very popular title amongst our readers.

We rarely see WinRAR achieve high processor utilization numbers, likely as a result of a storage bottleneck. But the outcome is similar to Photoshop anyway. The Ivy Bridge-based platform soars into a lead, followed by the Core i7-2820QM, which itself is trailed by the Core i5-460M. Again, AMD's A8 simply cannot keep up.

Both Bridges enjoy relatively low processor utilization numbers. The Core i7-3720QM gets its job done faster with a spike in power consumption, though, while Core i7-2820QM takes a lot longer, but maintains remarkably low power use. The result is roughly a wash in terms of energy efficiency.

Core i5-460M takes longer and uses more power (not a good combination). Although it's the slowest solution of all, at least AMD's A8 maintains steady consumption under 20 W.

5. Benchmark Results: iTunes 10.6.1

iTunes is single-threaded, so it reflects the performance of each CPU running at its highest Turbo Boost/Turbo Core clock rate. The Ivy Bridge-based Core i7-3720QM enjoys a raw frequency advantage and better instruction-per-cycle throughput, putting it in the lead. The other two Intel chips follow closely behind. Despite a peak 2.6 GHz Turbo Core clock rate, AMD's A8 simply cannot keep up.

With up to one core utilized at any given time, a workload like this is generally going to look pretty light (though it's interesting that AMD's chip gets as high as 50%).

Although the Ivy Bridge-based Core i7-3720QM only spins up a single core, it pushes that piece of logic to 3.6 GHz, exploiting as much available TDP as possible. Core i7-2820QM, manufactured at 32 nm, only hits 3.4 GHz, but it's not as efficient, so power consumption spikes slightly higher.

Pushing one-fourth of a quad-core chip with a 45 W TDP translates into modest power use. It's almost counter-intuitive that the 35 W Core i5 would use even more energy. However, remember that our Arrandale sample is a dual-core model, so we're monopolizing half of its on-die resources to do the same task.

As we've come to expect, AMD's A8 maintains very low power use, but over a painfully long period of time, negating any benefit the APU might have seen from its sub-20 W figure.

6. Benchmark Results: WoW, Call Of Duty, And Battlefield 3

If you want the architectural run-down of the Ivy Bridge architecture's HD Graphics 4000 core, head on over to Intel Core i7-3770K Review: A Small Step Up For Ivy Bridge and check out page three. Briefly, the new engine adds four execution units, which now total 16, and a number of optimizations for performance that yield better benchmark results than Intel's HD Graphics 3000 implementation.

The advantage is more significant on the mobile side than it was in our desktop-oriented measurements. The flagship Core i7-3920XM offers a maximum graphics clock rate of 1.3 GHz, where as the Core i7-3770K's HD Graphics 4000 component tops out at 1.15 GHz. The base clocks for both the mobile and desktop processors are 650 MHz.

The Core i7-3720QM in our notebook sample employs a maximum graphics frequency of 1.25 GHz, putting it just behind the Core i7-2820QM's highest bin. However, the increase in resources dedicated to higher frame rates means HD Graphics 4000 still delivers superior speed.

In our evaluation of the Core i7-3770K, we found that AMD's 100 W Llano-based APUs delivered better graphics performance than Intel's Ivy Bridge architecture. The A8's bigger power budget gives AMD more of an opportunity to emphasize its GPU component. But when you scale all the way back to a 35 W TDP, that's no longer true. Intel's manufacturing advantage more palpably kicks into play, and HD Graphics 4000 is able to shine within a 45 W thermal envelope.

Radeon HD 6620G is the highest-rated graphics implementation in AMD's mobile Llano family. But with only 400 shader cores operating at 444 MHz, the HD 6620G poses no threat to HD Graphics 4000 (or even the HD Graphics 3000 that came before).

In Call of Duty: Modern Warfare 3, AMD's Radeon HD 6620G basically matches the performance of HD Graphics 3000 with anti-aliasing disabled. However, HD Graphics 4000 blows both solutions out of the water, delivering more than twice as much performance. Even  with anti-aliasing enabled, AMD's Radeon HD 6620G is ~30% slower than Intel's HD Graphics 4000.

The older Core i5's vanilla HD Graphics engine is too old to support Battlefield 3. Given the numbers we see from the other systems that do manage to run it, performance would fall in the single-digit range anyway, though.

Again, we see the Radeon HD 6620G match the performance of HD Graphics 3000. But HD Graphics 4000 runs away with the victory, averaging higher (but still only marginally-playable) average frame rates. The speed-up ranges from 70 to 80%, depending on the resolution and quality settings you consider.

7. World Of Warcraft: CPU Utilization And Power Consumption

High Quality

Despite its higher clock rate, the Core i7-3720QM reports higher CPU utilization than the Core i7-2820QM in World of Warcraft: Cataclysm. It's power consumption is roughly the same, though. That's a notable advantage when you also consider the HD Graphics 4000 engine's better performance.

Good Quality

Fair Quality

8. 3D Performance And Power Profiles, Demystified

Notebooks generally employ different power policies, depending on whether the power source is a wall wart or a battery. Many of these settings are not user-accessible, so, we run our benchmarks on battery power in order to maintain consistency.

The graphics subsystem's power profile is controlled independent of the Windows-based power controls. There are three options available on an Intel-based system: Maximum Battery Life, Balanced (the default), and Maximum Performance. AMD's platforms default to “Maximum Battery Life,” and they don't include a “Balanced” middle setting.

All of the benchmark results we just presented to you employ the Intel and AMD defaults. However, when you start testing across profiles, the results start changing in a way that bears examination. The Core i7-3720QM and -2820QM Maximum Battery Life profiles cut performance by almost 50%, but also help the processors consume between 40 and 50% less power. By virtue of advanced 22 nm lithography, HD Graphics 4000 is the more efficient design when you divide power use into frame rate, of course.

In comparison, Arrandale's lack of granular graphics control means you don't see any change in performance or power use when you alter the profiles.

Meanwhile, AMD's Radeon HD 6620G enjoys a 35% speed-up when we switch to Maximum Performance, simultaneously incurring a ~40% increase in power consumption.

Core i7-3720QM

Core i7-2820QM

Core i5-460M

A8-3520M

9. Quick Sync: Performance And Power Consumption

Beyond its impact on gaming, the HD Graphics 4000 engine also presents enhanced Quick Sync functionality. Previously transcoding a complete Blu-ray movie took about 27 minutes. Now, you can get the same job done in about 17 minutes. 

When you factor in the advantages of a new lithography node, Quick Sync appears to be much more efficient, too. Although the CPU utilization of our Core i7-3720QM proved to be consistently higher (perhaps a result of getting more work done in less time), power consumption concurrently dropped by 15-30%. 

Performance Preset

Quality Preset

10. Benchmark Results: Blu-ray Playback Efficiency

With Blu-ray media becoming more commonplace these days, we thought it fitting to measure Blu-ray playback power consumption efficiency. Surprisingly, AMD's decode hardware turns out to be the most energy-efficient, even though it requires more CPU resources than Intel's implementation.

In comparison, Ivy Bridge appears to be less efficient than Sandy Bridge, if only by a small margin, even though CPU utilization is largely identical.

The worst performer in this test is Intel's old Arrandale design, which requires almost ~50% more power than Ivy Bridge, Sandy Bridge, and AMD's Llano.

11. Mobile Ivy Bridge: Paving the Way For Ultrabooks

We already know that Ivy Bridge is a tick in the company's cadence. It represents a shift to 22 nm manufacturing, and incorporates much of what Intel introduced alongside Sandy Bridge. Naturally, it's an evolutionary step forward more than anything.

In the desktop space, Ivy Bridge doesn't have much effect at all on the performance of our tests. Improved IPC throughput shaves off a few seconds here and there from our processor-bound benchmarks. And of course, the graphics engine is significantly faster. But an enthusiast puts very little weight on integrated graphics, and a couple of percentage points certainly aren't enough to compel an upgrade from the previous generation.

Asus's N36Vm, Ivy Bridge Based NotebookAsus's N36Vm, Ivy Bridge Based Notebook

Ivy Bridge's impact on the mobile space is certainly more profound, though. That 22 nm adoption helps bring down power consumption, even as Intel's newest processors maintain similar performance. Moreover, a notebook is far more likely to exploit good-enough built-in graphics. By beefing up its GPU, HD Graphics 4000 proves capable enough to satisfy more of Intel's customers than any past effort. It's even capable of slogging through Battlefield 3 (albeit using low resolutions and modest quality settings).

Lower power, better compute performance, and faster graphics. Those are all massive boons to partners designing small, thin, and light mobile platforms based on Ivy Bridge. 

Where does the competition land? Well, none of AMD's Llano-based parts come equipped with fixed-function logic able to match Quick Sync. We know that upcoming Trinity-based APUs will include the company's VCE capability, but because that feature isn't even enabled on the Radeon HD 7000-series add-in cards, we don't know how it'll match up.

AMD does have a great graphics solution. Unfortunately, throttling down from the 100 W TDP of its desktop Llano-based parts down to 35 W affects 3D performance in a big way. The A8-3520M boasts 400 shader cores, but they have to operate at much lower clock rates. The result is generally good enough to match Intel's HD Graphics 3000 implementation, but the new Ivy Bridge design simply pulls away effortlessly.

In the days to come, we'll be seeing AMD's answer to Ivy Bridge in its Trinity design. Armed with Piledriver-based processor cores and a more efficient graphics architecture, the company probably won't try to compete against Intel's highest-end mobile models. We do expect a big value push, though, and an almost-certain counter to Intel's Ultrabook initiative.

Until then, Intel's most recent step forward looks like it'll be a major enabler for a number of the company's partners. The high-performance, lower-power Ivy Bridge design facilitates the newest generation of Ultrabooks, embodied in a fresh form-factor that many folks hope will breathe new life into the diminishing notebook marketplace. With tablets increasingly grabbing mind and market share, Intel is placing its bets on Ivy Bridge-powered Ultrabooks to help turn the tide.