Sign in with
Sign up | Sign in
Intel Core i5 And Core i7: Intel’s Mainstream Magnum Opus
By ,
1. Introduction

Intel’s emphasis right now is on Clarkdale, the Nehalem-based mainstream lineup centering on a 32nm process shrink. Clarkdale will be the foundation on which upcoming Core i5 and Core i3 CPUs are based. It’s a big deal for Intel. So big, in fact, that I was told jokingly two weeks before the Lynnfield launch that the whole company had been focusing on Clarkdale, not the Core i5 and Core i7 we’re seeing today.

Of course, that’s only really funny for the folks who’ve already seen how the Lynnfield-based processors actually perform and know they’re not as anemic as an enthusiast might expect, given the fact that Intel is aggressively pursuing integration, aiming for a SoC-type design in the not-so-distant future.

But Clarkdale is six months away, at least. Today is all about Lynnfield—the Core i5 and Core i7 CPUs for Intel’s LGA 1156 interface.

The Venerable Core 2 Rides Off…Sort Of

With the divulging of its Core i7, Core i5, and Core i3 branding, Intel quietly rang the death knell of its Core 2 family, which has been with us for more than three years now, gently massaging away memories of a day when the company ravenously chased after faster clocks.

That transition won’t happen immediately, though—or even quickly for that matter. Well into the fourth quarter of next year, Intel’s Core 2 architecture will remain a value play. Even today it’s going to persist as a viable option for entry-level buyers.

Core 2 Quads span from $163 to $316 in the company’s August 9th price list. Core 2 Duos range from $113 to $266. Does the trio of CPUs being launched today wreck a number of those price points? Absolutely. Do the three Lynnfield processors we’re seeing now, from $199 to $555 smother Core 2 Quad and Core 2 Duo to the point that everyone will spend at least $200 on their next CPU? Obviously not.

Wait, Define Mainstream

To make a long story a little shorter, Bloomfield sits at the top of Intel’s stack as Core i7 for LGA 1366. Lynnfield now occupies a space between the high-end and the mid-range segments. Yorkfield (Core 2 Quad) becomes this transitional family that tides Intel over until Clarkdale launches in Q1’ 2010. And Wolfdale continues on in the dual-core Pentium family through the course of 2010.

If you would have considered a Core 2 Quad or Phenom II X4 previously, the lone Core i5 will be of interest to you. If you were previously pondering a Core i7 for LGA 1366, the Core i7-860 and -870 are now vying for your attention with price points disturbingly similar to the i7-920 and -950, respectively. How’re you supposed to choose between CPUs when architecture, functionality, and pricing are all so similar?

Intro to Core i5/i7


Good question—this is one of the areas where we’re going to be particularly critical of Intel today. The naming is a mess if you’re not already familiar with the technology. Fortunately, Intel puts function ahead of marketing, so there’s a lot more exciting innovation to cover than confusing branding. Let’s just get that out of the way, first.

2. What’s In A Name?

With Core 2 Quad and Core 2 Duo, you had a general idea of what you were dealing with upfront. Adding a Q8000- or E7000-series designator wasn’t particularly descriptive, but at least there was one tangible identifier in there for the layman to digest.

When Intel introduced Core i7 last year, the fact that there was only one Nehalem-based desktop family made it easy to say “Core i7—yeah, high-end stuff,” regardless of whether you were actually talking about the $300 i7-920 or the $1,000 i7-965.

Core i7, Core i5, and Core 2 QuadCore i7, Core i5, and Core 2 Quad

Now you have Core i5 and another strain of Core i7; at this point the nomenclature is just jargon for everyone except for the power users who live and breathe this stuff. Let’s break it down in a neat little table, though.


Core i7 (LGA 1366)
Core i7 (LGA 1156)
Core i5
Core 2 Quad
Processor Interface
LGA 1366
LGA 1156
LGA 1156
LGA 775
Number of Cores
4
4
4
4
Turbo Boost
Yes
Yes
Yes
No
Hyper-Threading
Yes
Yes
No
No
L1 Cache
32KB/32KB per core
32KB/32KB per core
32KB/32KB per core
32KB/32KB per core
L2 Cache
256KB per core
256KB per core
256KB per core
Up to 12MB shared
L3 Cache
8MB shared
8MB shared
8MB shared
No
Memory Channels
3
2
2
2
Max. Memory Rate
DDR3-1066
DDR3-1333
DDR3-1333
DDR3-1600
Chipset
X58
P55
P55
X48
Price
$284-$999
$285-$555
$199
$163-$316


As you can see, Core i7 for LGA 1366 remains the enthusiast-class offering, sporting the most PCI Express 2.0 connectivity via Intel’s X58 Express chipset, up to three available channels of DDR3 memory support, Intel’s first generation of Turbo Boost, and Hyper-Threading.

Core i7-870Core i7-870Core i5-750Core i5-750

Core i7 for LGA 1156 integrates the PCI Express connectivity (albeit 16 lanes instead of 36), sheds one memory channel, incorporates an updated implementation of Turbo Boost, and maintains Hyper-Threading support.

Core i5 includes the same on-board PCI Express subsystem and dual-channel integrated memory controller. It employs Intel’s improved Turbo Boost (though it’s slightly less aggressive than i7’s). What it lacks, however, is Hyper-Threading—apparently a noteworthy-enough capability to turn an i7 into an i5. Of course, descending the stack also results in lower base clock rates.

The Making Of A Core i5/i7

Architecturally, most of what you get in a Core i5 or Core i7 processor is borrowed from technology already found in Intel’s LGA 1366-based Core i7 and Xeon 3500-series processors. The Lynnfield die is different from Bloomfield though, estimated at 774 million transistors packed into 296 square millimeters (versus 731 million in 263 millimeters for the first Core i7s). More than 400 million of those transistors make up the CPU's cache.

As with Bloomfield, Lynnfield is a monolithic design divided into four cores (execution pipelines, L1 data/instruction cache, TLBs) and the uncore (L3 cache, integrated PCI Express, the memory controller, QPI, and the PLLs). The power for these two “halves” remains separate, independently adjustable in your motherboard's BIOS.

Bloomfield: core vs. uncoreBloomfield: core vs. uncore

Each of the four cores retains its 32KB L1 data cache (still 8-way set-associative), 32KB L1 instruction cache (still 4-way set-associative), and 256KB L2 cache (you guessed it—still 8-way set associative). An inclusive 8MB L3 cache 16-way set-associative) should look familiar as well. 

With nothing really new to report in the cores themselves, we move to the uncore, where Intel has added PCI Express 2.0 connectivity, axed a single memory channel, dropping the total to two, and altered QPI.

3. QPI, Integrated Memory, PCI Express, And LGA 1156

The company won’t get specific about the changes made to its QuickPath Interconnect, the point-to-point interface first used to attach Core i7 to X58. But there is a QPI link inside of Lynnfield, serving as the glue between on-die PCI Express and the uncore. Intel has a lot more freedom with this technology now, since it doesn’t have to leave the die. Thus, the performance and timings are both reportedly better here than they were on Bloomfield.

This interconnect is important today, naturally, but will become a key performance enabler when Clarkdale launches next year. With graphics and memory control on the same piece of silicon, memory bandwidth—one of the biggest Achilles’ heels of integrated graphics designs—is delivered much more effectively. As a result, expect to see Intel adjusting the speed of its internal QPI link up or down to differentiate its Clarkdale SKUs.

Of course, none of that is really a concern with Lynnfield, which employs 16 lanes of integrated PCI Express 2.0 to interface with discrete GPUs. We’ve already spent considerable time comparing the performance of single- and dual- card configurations on P55 against X58, P45, and 790GX using 2.8 GHz processors. We've determined that there’s very little gain or loss resulting from Intel’s Lynnfield implementation. In other words, the integration of PCI Express is just about as close to transparent to end-users as possible. This is the way integration should work (though most of us have been conditioned to think it automatically leads to performance sacrifices).

Much more impactful is the licensing of SLI and CrossFire, which allows most P55-based motherboards to support either technology, just like X58. Of course, the difference is that P55 platforms are going to be significantly cheaper. More on the chipset shortly.

Finally, you have the processor’s integrated DDR3 memory controller, which is cut from three 64-bit channels in Bloomfield to two 64-bit channels with Lynnfield. Officially, Bloomfield is rated for up to DDR3-1066, yielding a theoretical maximum of 25.6 GB/s of bandwidth. We’ve all seen the LGA 1366-based Core i7s scale as high as DDR3-2133 though, so the official spec means very little to enthusiasts. In contrast, Lynnfield is rated for DDR3-1333 memory modules, offering up to 21.3 GB/s. For the most part, the loss of that third channel is made up for by an increase in attainable data signaling rate. We’ll put this theory to test in a couple of pages.

LGA 1156: More Socket Segmentation

The table below should look somewhat familiar; I used it back in February to chart the last nine years of desktop socket launches from Intel and AMD. With the debut of LGA 1156, the Intel column is looking pretty darned crowded.

Disruptive Socket Launches
Year
AMD
Intel
2001

Socket 478
2002


2003
Socket 754

2004
Socket 939
LGA 775
2005


2006
Socket AM2

(Intel launches Core 2 Duo, most motherboards need to be replaced)

2007


2008

LGA 1366
2009
(Socket AM3new processors work in old motherboards, but not the other way around)LGA 1156


I’m sure there will be a number of enthusiasts who’ve spent a lot of money on LGA 1366, X58, and Core i7. You'll see the performance of Lynnfield, compare the cost, look at the upgrade path to Clarkdale, and ask me why I couldn’t have given a more vocal heads-up about what was to come. To those folks, let me give you one hopefully-welcome piece of information: the upcoming Gulftown design (32nm, hexa-core, Hyper-Threading-enabled) will be a drop-in upgrade to your existing platform. Meanwhile, those waiting on Clarkdale have a dual-core design to look forward to—certainly not as sexy, unless you’re really itching for integrated graphics. 

What, then, was the impetus behind LGA 1156 less than a year after LGA 1366? And why couldn’t Intel integrate PCI Express on the LGA 1366 platform, too?

Right from the get-go, 1366 was designed as a DP interface for Intel’s Tylersburg platform. The fact that it surfaced first on the desktop is similar to how LGA 771 made a brief appearance in Intel’s Skull Trail gaming configuration. The difference is that motherboards based on X58 made their way under $200 and the cheapest processors dropping into the interface are less than $300. That’s not little money, to be sure. But it’s affordable enough to tempt the folks who know how fast an overclocked i7 machine can be.

LGA 1366 didn’t get PCI Express because it would have made the pin-out far too complex. Plus, the servers and workstations that now use the 5520 and 5500 chipsets need access to more than just 16 lanes of connectivity for storage controllers, professional graphics cards, and high-speed networking.

LGA 1156: the real mainstream interfaceLGA 1156: the real mainstream interface

Thus, LGA 1156 becomes the true successor to LGA 775, despite now-convoluted segmentation. It’s simpler than LGA 1366, its locking mechanism applies pressure more evenly on the CPU’s heat spreader, and it helps protect against EMI—particularly important when Clarkdale exposes on-chip graphics.

You can expect this processor interface to remain pervasive until the end of 2011, when LGA 1155 (planned launch in the first half of 2011) starts grabbing market share big time.

4. Intel’s Turbo Boost: Lynnfield Gets Afterburners

The integrated PCI Express, two-channel memory controller, and tweaked QPI link are notable changes from Bloomfield to Lynnfield, but none are as story-altering as Intel’s latest version of Turbo Boost technology.

You’ll remember from our Bloomfield analysis that Core i7-900-series CPUs feature a PCU (power control unit), an on-die controller with approximately the same number of transistors as a complete 486 processor. At regular intervals, the PCU samples temperature, current, power consumption, and operating system states.

What does it do with this information? In the case of Bloomfield, which has a 130W TDP, the processor can almost completely shut off cores that are not in use by dropping them into C6, cutting consumption. Obviously, idle cores (those in C3 or C6) result in a larger gap between actual power use and the imposed maximum. So, in threaded workloads, where three or four of Bloomfield’s cores are working (but still under the PCU’s programmed limits), this feature called Turbo Boost increments the CPU’s clock ratio by one. Multiplied by a 133 MHz base clock, that’s an additional 133 MHz of free clock rate. With only one core active (in C0 or C1), Turbo Boost could take things up a second performance bin, adding 266 MHz to the chip’s standard clock.

Now, back when the Core i7-975 Extreme launched, I observed that it was actually very rare to see two bins of Turbo Boost in practice, since Vista’s scheduler has a bad habit of bouncing threads from one core to another, keeping multiple cores in action. I was able to screenshot the 975 running at 3.6 GHz, but it only lasted a fraction of a second. In that case, two bins wouldn't be what I’d consider a tangible benefit.

Fast forward to today. All three Core i5 and Core i7 CPUs now sport a 95W TDP (and an 89A ceiling), making the PCU’s power-policing duties even more critical. Adding further to the controller’s role is a more aggressive implementation of Turbo Boost. With three or four cores active, the Core i5-750 and Core i7-860 get a one-bin improvement each (the Core i7-870 gets two). But with only two cores active, all three models enjoy a four-bin (533 MHz) speed-up. And with one core active, the two Core i7s get five bins (667 MHz) so long as you’re still under 95W.

Turbo Boost: Available Bins (Under TDP/A/Temp)
Processor Number
Frequency
4 Cores Active
3 Cores Active
2 Cores Active
1 Core Active
Core i7-870
2.93 GHz
2
2
4
5
Core i7-860
2.8 GHz
1
1
4
5
Core i5-750
2.66 GHz
1
1
4
4
Core i7-975
3.33 GHz
1
1
1
2
Core i7-950
3.06 GHz
1
1
1
2
Core i7-920
2.66 GHz
1
1
1
2


At the end of the day, this is really the legacy of Lynnfield. In threaded environments, you see the benefit of a quad-core processor. In titles like WinZip, Lame, or iTunes—benchmarks we’ve seen time and time again favor higher-clocked dual-core chips—Lynnfield kicks into gear to serve up better single-threaded speed.

In order to keep an even closer handle on consumption, the sampling rate of Lynnfield’s PCU is increased versus Bloomfield. Consequently, the new Core i5 and Core i7 processors can ramp voltage up or bring it back down more aggressively than Bloomfield-based CPUs, and thus react faster to a given single- or multi-threaded workload. It’s all fine-tuning really, but when you’re talking about switching between idle and active states, every little bit counts.

Turbo enabledTurbo enabledTurbo disabledTurbo disabled

Testing Turbo

It’s actually really easy to see Turbo Boost in action, thanks to the TMonitor beta at cpuid.com. By simply assigning an application to a single processor core through Windows’ Task Manager, it’ll be constrained to that one CPU, realizing as much Turbo Boost acceleration as is available. Incidentally, this isn’t the best way to actually use a Core i5- or Core i7-based machine, as we realized better performance by default (even though iTunes looks to be jumping around between available cores).

Setting affinity...Setting affinity...

iTunes by default: 1:22iTunes by default: 1:22Manual affinity: 1:25Manual affinity: 1:25

As you might expect from a technology that can “overclock” a Core i5-750 from 2.66 GHz to 3.2 GHz in single-threaded workloads, the performance potential is significant. Our upcoming benchmarks compare the new i5 and i7 processors against Bloomfield, Yorkfield, and AMD’s Deneb, but here you’ll see i5-750 with Turbo on and off under iTunes.

The result is compelling, especially when you look at how the i5 does against the rest of the pack.

Put the i5-750 ($199) up against the i7-920 ($279-ish) in single-threaded apps and you get even more interesting results. Turbo Boost is the “killer app” that’s going to allow these Lynnfield processors to beat out more expensive Bloomfield configurations in the right environments.

Core i7 Coming Up Short With One Core Active?

In trying to get the very most out of Turbo Boost, we did observe some interesting behavior with single-core scenarios. The Core i5-750 had no trouble realizing its four-bin boost, jumping from 2.66 GHz to 3.2 GHz in a number of workloads.

Manually setting affinity yields 27x...Manually setting affinity yields 27x...Running at default yields 26x...Running at default yields 26x...

However, as we observed in our Core i7-975 Extreme review, the Core i7-870 and Core i7-920 failed to reach their 3.6 GHz and 2.93 GHz peaks, respectively, unless you force the running program into a single core manually.

Now, according to Intel, most applications that report clock rates rely on ACPI-based P-states, and are thus unable to correctly detect frequencies modified by Turbo Boost. But we’ve been assured that TMonitor does properly reflect the actual frequency on a per-core basis. Consequently, it’s more accurate to think of the Core i7-800s as receiving four bins with one core active and the Core i7-900s as getting one extra bin, rather than counting on five and two bins, respectively.

5. Hyper-Threading: Differentiating Core i7

So now we have Core i7-800s on LGA 1156 and Core i7-900s on LGA 1366. Confusing, yes. But I’ve tried to make it clear here, at least, that what makes an i7 an i7 is its inclusion of Hyper-Threading technology. Indeed, Intel also uses clock rate as a differentiator, and at least in theory you should get more Turbo Boost from an i7-800-series CPU than an i5 as well.

As you probably already know, Hyper-Threading is Intel’s simultaneous multi-threading technology that presents each physical core as two logical cores to an operating system. Thus, when you open the Task Manager on a Hyper-Threading-enabled Nehalem-based CPU, you see eight threads. This doesn’t mean you suddenly have the equivalent of an eight-core processor. Instead, Intel duplicates certain resources in each physical core to make the technology available; it’s better to think of Hyper-Threading as allowing the execution resources on a quad-core i7 to be better-utilized in threaded workloads.

We know that MainConcept Reference, for example, is well-threaded. By simply turning Hyper-Threading on, our transcoding workload falls from 1:48 to 1:26.

Similarly, the latest version of AVG’s anti-virus software realizes a massive gain thanks to Hyper-Threading. Less-optimized or simply multi-tasked environments will yield less-pronounced results, but there certainly seems to be a reason to at least consider a Core i7 if you’re able to benefit from what Hyper-Threading offers.

6. Memory Architecture: Does Losing One Channel Hurt?

You “lose” two things in stepping down from an LGA 1366-based interface to the LGA 1156 Core i7 (three things if you go for a Core i5). There’s the triple-channel memory architecture, enough PCI Express 2.0 via X58 to give each graphics card in a CrossFire or SLI config its own x16 link, and, in the case of i5, you also lose Hyper-Threading.

We already know that Hyper-Threading can be a big boon if you’re running the right apps. We know that the PCI Express situation really isn’t that big of a deal. But what about the memory subsystem? Technically, Lynnfield’s two channels of DDR3-1333 come within 4 GB/s of Bloomfield’s three DDR3-1066-capable channels. But those specs mean very little to the power users willing to shoot for 1,600, 1,866, or 2,000 MT/s.

Because we’ve found very little reason to recommend anything faster than DDR3-1333 (at least as far as performance goes), we’re arming our X58 platform with two and three channels of DDR3-1333 memory and our P55 test bench with two channels of the same stuff running 7-7-7-20-1T timings.

There’s clearly a massive throughput advantage with three channels of DDR3 memory. But as we’ve seen over and over, it doesn’t necessarily translate over into the real world. If you were worried about a negative impact on performance due to Lynnfield’s memory controller, don’t.

7. P55: The Chipset’s Responsibilities Dwindle

Say farewell to Intel’s conventional three-component platform design. P55 (and very likely every desktop chipset moving forward) centers on a two-chip implementation consisting of the CPU and one piece of motherboard core logic. Surely, there’s a team of Nvidia engineers feeling pretty gosh-darned vindicated right now.  

With the memory and PCI Express controllers now part of Lynnfield (and graphics migrating that direction with Q1’s Clarkdale launch), there’s little else for a chipset to do except the functionality formerly handled by Intel’s ICH southbridge lineup. Thus, P55 gives you six 3 Gb/s SATA ports, a Gigabit Ethernet MAC, 14 USB 2.0 ports, HD Audio, and eight lanes of PCI Express 2.0 for peripheral connectivity. As an indication of how far southbridge technology has come in the last two years or so, P55 is wholly uninspiring.

There is some notable power savings here compared to X58, though. To begin, Lynnfield sports a 95W TDP. Bloomfield is 130W. The X58 Express IOH is a 22W part. That vanishes completely. P55 uses up to 4.7W. And ICH10R consumes up to 4.5W. Add it all up and you’re down more than 56W right off the bat.

Making The Connection

Intel’s most recent three-chip desktop platform, X58, employed a 25.6 GB/s QPI link between the Core i7 CPU and X58 IOH. It then used a 2 GB/s DMI connection between X58 and the ICH10 chipset component.

As we shift to P55 and its two-chip design, the northbridge gets absorbed into Core i5/Core i7, and we’re left with what amounts to a southbridge attached to the processor, even if Intel refers to this as a platform controller hub. As with the ICHes before it, P55 connects to its host (Lynnfield) through a DMI connection.

According to Intel, the DMI link between Lynnfield and P55 runs at 2 GB/s, similar to past-generation ICHes. Previously, that connection handled six lanes of PCI Express 1.1, SATA, USB 2.0, Gigabit Ethernet, and HD Audio. With the move to P55, most of those subsystems remain unchanged. However, the chipset now supports eight lanes of PCI Express 2.0.

Intel is nevertheless confident that its DMI link won't be saturated. The math doesn't lie, though. With the right combination of add-in storage and SSDs, it wouldn't be difficult to jam things up there.

8. Windows 7: Microsoft Listens To Intel, Finally

Over the course of its life, Windows Vista has taken a load of abuse—much of which is deserved.

One area we’ve seen both Intel and AMD affected is power management. In AMD’s case, enabling Cool’n’Quiet technology on its original Phenom processors caused a sizable performance hit as Vista’s scheduler moved threads from active to idle cores running at half-speed in a process called migration.

Cores parkedCores parked

Why did it do this? In order to maintain the symmetry of a system under full load, you don’t want I/O to become dependent on just one core. If you keep threads rotating between cores running at their maximum performance (this whole concept goes out the window when you start talking about spinning cores down), you get better responsiveness.

This was an implementation decision made during Microsoft’s Windows NT kernel design, and based on our experiences with both processor vendors' hardware, it wasn't considered a "feature" to either company. Of course, it affected Intel in a much different way than AMD. The problem Intel had in Vista was one of power consumption. For every migration, you have to write-combine the Nehalem architecture’s L3 cache, which costs power.

And loadedAnd loaded

This changes with Windows 7 and a feature called ideal core. If a task’s needs are being addressed by one core, the operating system will let you stay there. This means two things to Intel: first, you don’t use power on the migration, and second, idle cores are able to remain in a C6 state. Purportedly, this migration fix alone will yield an extra 10 to 15 minutes of battery life on Nehalem-based notebooks, though this won’t become a major issue until the mobile dual-core Arrandale launches later this year. Perhaps more interesting, though, is that processors without C6 will not realize this gain (including AMD’s CPUs).

Core parking is a second optimization, based on the observation in previous operating systems that you might have four cores running background processes at 10% utilization each. The idea is to load all of those tasks onto one core and let the others idle if operating load levels allow for it. Now, you can see how these two features working together might have a significant impact on power, as ideal core prevents rabid thread migration, while core parking optimizes loading. Taken together, the pair intelligently maximizes the number of idle cores, and then keeps them from being spun up unnecessarily, yielding the theoretical power gains.

If you want to know more about the changes incorporated into Windows 7, check out this interview with Mark Russinovich, a Technical fellow for Microsoft.

We’re Making The Switch

Based on reader feedback to Windows Vista, access to the final Windows 7 code, and great app compatibility with our current benchmark suite, we updated as many software versions as possible and made the leap to Windows 7 for our review here (along with the gaming analysis of CrossFire and SLI graphics configurations, published separately).

But before we did, we wanted to quantify these power-saving claims from Intel for ourselves. So, we logged runs of PCMark Vantage on clean drives with an install of Windows Vista updated to Service Pack 2 and the Windows 7 RTM code, both x64 builds.

The results were actually counter to what we expected. The Windows 7-based build averaged six watts higher over the course of its run, but finished the test three minutes faster than the Vista machine. Also noteworthy, though, is that when the Windows 7 machine has a chance to idle (which is where we'd expect to see ideal core and core parking actually having an effect), it does dip down lower than the Windows Vista box.

We checked these results over with Intel, and came away with the following interpretation: the Windows 7 P-state promotion policies are more aggressive than Vista's, meaning a Windows 7 system ramps to Turbo Boost faster, resulting in the better performance and higher power consumption. At idle, the previously-discussed features enable the Windows 7 config to dip below the idle power draw of the Vista machine.

Overall, though, Windows 7 actually averages higher power consumption in this experiment, even if it simply idles for the three minutes while the Vista box finishes its run. We are fairly certain of why, exactly, this is, but will hold off on comment until we're able to present power data substantiating the claim. However, that doesn't change the fact that, in this case, Windows 7 won't be cutting your power bill. In order to show Windows 7 cutting consumption, we'd have to spend a lot more time at idle (admittedly more representative of how most PCs are typically used), replace a certain component, or disable certain settings in the OS.

9. Test Setup And Benchmarks
Test Hardware
Processors
Intel Core i7-920 (Bloomfield) 2.66 GHz, LGA 1366, 4.8 GT/s QPI, 8 MB L3, Power-savings enabled

Intel Core i7-870 (Lynnfield) 2.93 GHz, LGA 1156, 8 MB L3, Power-savings enabled

Intel Core i5-750 (Lynnfield) 2.66 GHz, LGA 1156, 8 MB L3, Power-savings enabled

Intel Core 2 Extreme QX9770 (Yorkfield) 3.2 GHz, LGA 775, 1,600 MHz FSB, 12 MB L2, Power-savings enabled

Intel Core 2 Quad Q9550S (Yorkfield) 2.83 GHz, LGA 775, 1,333 MHz FSB, 12 MB L2, Power-savings enabled

AMD Phenom II X4 965 BE (Deneb) 3.4 GHz, Socket AM3, 4 GT/s HyperTransport, 6 MB L3, Power-savings enabled
Motherboards
Asus P6T (LGA 1366) X58/ICH10R, BIOS 0707

Gigabyte P55-UD6 (LGA 1156) P55, BIOS F3

Intel DX48BT2 (LGA 775) X48/ICH10R, BIOS 1902

Asus M4A79T Deluxe (AM3) 790FX/SB750, BIOS 1103
Memory
Corsair 4 GB (2 x 2 GB) DDR3-1600 7-7-7-20 @ DDR3-1333

Corsair 6 GB (3 x 2 GB) DDR3-1600 7-7-7-20 @ DDR3-1333
Hard Drive
Intel SSDSA2M160G2GC 160 GB SATA 3 Gb/s

Intel SSDSA2MH080G1GN 80 GB SATA 3 Gb/s
Graphics
Sapphire Radeon HD 4870 X2 2GB
Power Supply
Cooler Master UCP 1100W
Heatsink
Vigor Gaming Monsoon III LT
System Software And Drivers
Operating System
Windows 7 x64 RTM
DirectX
DirectX 11
Platform Driver
Intel INF Chipset Update Utility 9.1.1.1015
Graphics DriverCatalyst 9.8


Although we tested the Core i7 and Core i5 configurations on Gigabyte's P55-UD6, our Turbo Boost, Hyper-Threading, memory performance, and Windows 7 versus Vista power measurements were all taken on Intel's DP55KG motherboard.

Remember that we ran all of our gaming analysis separately in order to fit in single- and dual-card CrossFire and SLI configurations across two resolutions and seven different titles. That story can be found right here.

Benchmarks and Settings

Audio Encoding

iTunes

Version: 8.2.1.6 (64-bit), Audio CD ("Terminator II" SE), 53 min., Default format AAC

Lame MP3

Version: 3.98.2 (32-bit), Audio CD ""Terminator II" SE, 53 min, wave to MP3, 160 Kb/s

Video Encoding

TMPEG 4.7

Version: 4.7, Import File: "Terminator II" SE DVD (5 Minutes), Resolution: 720x576 (PAL) 16:9

DivX 6.8.5

Encoding mode: Insane Quality, Enhanced Multi-Threading, Enabled using SSE4, Quarter-pixel search

XviD 1.2.2

Display encoding status=off

Mainconcept Reference 1.6.1

MPEG2 to MPEG2 (H.264), MainConcept H.264/AVC Codec, 28 sec HDTV 1920x1080 (MPEG2), Audio: MPEG2 (44.1 KHz, 2 Channel, 16-Bit, 224 Kb/s), Mode: PAL (25 FPS), Profile: Tom’s Hardware Settings for Qct-Core

Applications

Autodesk 3ds Max 2009 (64-bit)

Version: 2009 Service Pack 1, Rendering Dragon Image at 1920x1080 (HDTV)

WinRAR 3.90

Version 3.90 (64-bit), Benchmark: THG-Workload (334 MB)

WinZip 12.1

Version 12.1, Compression=Best, Benchmark: THG-Workload (334 MB)

Adobe Photoshop CS4
Radial Blur, Shape Blur, Median, Polar Coordinates filters
AVG Anti-Virus 8.5
Virus scan of 334MB of compressed files

Synthetic Benchmarks and Settings

3DMark Vantage

Version: 1.02, GPU and CPU scores

PCMark Vantage

Version: 1.00, System, Memories, TV and Movies, and Productivity benchmarks, Windows Media Player 10.00.00.3646

SiSoftware Sandra 2009 SP4

CPU Test=CPU Arithmetic/MultiMedia, Memory Test=Bandwidth Benchmark

10. Benchmark Results: Synthetics

Beginning with PCMark Vantage, we can already see the strength of Intel’s Core i7-870 processor. Remember, though, that just because it’s a Lynnfield part and purportedly mainstream doesn’t make it an affordable option. At $555, we actually think the Core i7-870 is overpriced, just like the Core i7-950. More notable is the triumph of the Core i5-750 against Intel’s Core 2 Extreme QX9770, a $1,000 CPU, and even more so versus the Core 2 Quad Q9550. As of the launch, the Core 2 Quad is priced $10 higher than the i5.

Synthetic benchmarks are especially prone to favoring Intel’s Hyper-Threading technology given an emphasis on parallelism in most future-looking tests. This would seem to be the case here judging from the CPU Suite and Overall benchmark scores that show Intel’s Core i7-870 and -920 as victorious.

The GPU Score reflects a more even playing field, though Bloomfield is still at a slight advantage. In the overall suite score, i5-750 again bests Core 2 Quad Q9550, pulling even to the 3.2 GHz Core 2 Extreme.

While the X48-based memory controller feeding Intel’s Core 2 Quad family pales in comparison to AMD’s integrated DDR3 memory controller, Lynnfield’s dual-channel arrangement armed with the very same DDR3-1333 modules running the very same timings realizes significantly more throughput. To that end, the Core i7-920 is even faster still.

11. Benchmark Results: Media Apps

Well-threaded, MainConcept gets a clear benefit from Hyper-Threading. Even without HT though, Intel’s Core i5-750 passes up the Core 2 Extreme. It’s not able to maneuver past AMD’s 3.4 GHz Phenom II X4 965 BE, though, which bests the $199 chip by two seconds.

Here’s where we’d expect to see Turbo Boost shine, and indeed the Core i7-870 and Core i5-750 both fly past the Core i7-920. The next-closest contender is the once-$1,000 Core 2 Extreme QX9770, though the fact that it’s being outclassed by a $199 Core i5 speaks volumes.

Our DivX transcode readily runs across eight threads, but seems to get a substantial boost from Turbo as the i7-870 and i5-750 both scoot past the Core i7-920. It’s great for AMD that its flagship is now able to topple the fastest Core 2-based chip here. But there’s a new mainstream sheriff in town and its name is Core i5.

Xvid is lightly threaded, so the fact that Intel’s Core i7-870 and Core i5-750 take first and second place is attributable to Turbo Boost yet again. Core i7-920 takes third, followed closely by Core 2 Extreme.

We weren’t seeing any benefit from the 64-bit build of Lame used in previous processor reviews, so we’ve switched to a newer 32-bit build here. Still single-threaded, Turbo Boost seriously bolsters the performance of the two Lynnfield chips, leaving Core i7-920 and Core 2 Quad Extreme behind.

12. Benchmark Results: Productivity

We would have thought Turbo Boost would give the 2.66 GHz Core i5 a larger lead over the i7-920. More surprising is the Core 2 Extreme’s first-place finish ahead of even the Core i7-870—capable of hitting 3.6 GHz with one active thread.

The latest version of WinRAR, 3.90, is optimized for 64-bit environments and threading—evident in the massive advantage held by both Core i7 chips. Core i5 takes a third-place finish, followed by Core 2 Extreme and Phenom II X4.

Also able to take advantage of Hyper-Threading, 3ds Max 2009 best handles its rendering workload using either of our two Core i7s. The Core i5 follows pretty close behind, though, and the rest of the pack is right on its tail.

Again we see Hyper-Threading take a massive chunk out of one of our benchmarks—in this case, it’s AVG’s latest anti-virus app. The $199 Core i5 and $1,000 Core 2 Extreme place similarly, but both are edged out by AMD’s Phenom II X4 965 BE.

Our Photoshop CS4 test runs through a series of thread-aware filters, though the Core i7-870’s clock rate advantage seems to propel it through the metric faster than any of the other CPUs. The Core i5-750 and Core i7-920 turn back comparable scores, though both are narrowly edged out by Intel’s Core 2 Extreme.

13. Power Consumption

The reduced platform power consumption of the Core i5 and Core i7 CPUs is immediately apparent, even at idle. Both Lynnfield-based designs dip in around 20W underneath the Core 2 Quad Q9550S, which we’ve been using up until this point to replicate the performance of a standard Q9550. The ‘S’ model has a 65W TDP though, so the fact that Core i7-870 and Core i5-750 suck up less juice at idle is impressive. So too is the idle consumption of AMD’s Phenom II X4 965 BE, which also ducks in under the Core 2 Quad.

Fire up the Small FFT test in Prime95, add a FurMark Burn-In test, and the power usage jumps through the roof. Here’s where Intel’s low-power Q9550S shines, turning in the best results. But the two Lynnfields continue to impress with the second and third lowest power consumption figures. AMD’s Phenom II X4 965 comes in fourth, followed by the Core 2 Extreme, and trailed by the 130W Bloomfield-based Core i7-920.

It can be difficult to take thermal design power specs and give them real-world meaning. However, when you do the math, these load numbers make good sense. The TDP of Intel’s low-power Core 2 Quad is 30W below Lynnfield’s spec. Subtract out a power-hungry northbridge and you’re looking at the gap we see here in practice. Add 22W to the X58’s power budget and then take Bloomfield’s 130W ceiling into account; it’s no wonder Core i7-920 sits at the other end of the spectrum.

14. Conclusion

I’ll be honest—when I first got my hands on a pre-production Core i5 three months ago, the processor took me by surprise, even with an artificial cap of 2.8 GHz on its Turbo Boost functionality. That was before final specs or pricing was available. Now that we’ve had a couple of weeks with final hardware the Core i5 and Core i7 processor families are even more fascinating.  

To begin, they make it much harder to recommend LGA 1366-based Core i7s. We know the i7-900-series is supposed to be higher-end, and it’s hard to ignore the fact that next year we’ll see hexa-core Gulftowns that drop right into our X58 motherboards. But seriously. Motherboards priced under $100? Core i5s under $200? We’re talking a possible contender next time we tackle an Intel-based $650 System Builder Marathon story (AMD fans rejoice—this month we’ll be doing an all-AMD series for you guys). That’s $10 less expensive than a Core 2 Quad Q9550 and $45 less than a Phenom II X4 965.

Alright, so the Core i5-750, specifically, is priced well. What is there to like about it? Reasonable power consumption, a base clock rate comparable to Intel’s Core i7-920, a more-aggressive Turbo Boost able to take the chip to 3.2 GHz in single-threaded workloads, CrossFire and SLI compatibility—it’s a pretty compelling list, actually.

What about the two LGA 1156-based Core i7s? We tested the Core i7-870 and are fairly convinced that, like the Core i7-950, it sits in a no-man’s land. Nearly two times the price of Core i7-860 and only marginally better-looking on a spec sheet, the Core i7-870 becomes Lynnfield’s version of an Extreme Edition processor—without the unlocked multiplier. More attractive for the folks who stand to benefit from Hyper-Threading is Core i7-860. Its price tag puts it in the realm of Core i7-920, its Turbo Boost helps make it faster, and a complementary motherboard is going to cost you between $75 and $50 less.

But based on our benchmarks here and our game testing with single and dual Radeon HD 4870 X2s and GeForce GTX 285s, we’re most excited about the value of Core i5. The fact that it’s regularly able to smack around the current Core 2 flagship (QX9770) is just crazy.

Of course, this launch isn’t all bad news for the AMD enthusiasts out there. When the Phenom II X4 965 BE debuted in August, I hinted that you should wait until today before taking a leap. Now you see why. With i5-750 selling at $199, AMD has no choice but to compress its price list. At the very least, it’ll likely slash the prices on its high-end Phenom IIs. If you held off, great deals are quite likely in your future.

But any price action in the Phenom II or Core 2 lineups is going to be a result of a solid showing today by Core i5, which is why it earns the first Recommended Buy award I’ve given to a processor in almost a year and a half managing Tom’s Hardware.

Stay tuned. Patrick Schmid is working on comprehensive overclocking coverage using these two Lynnfield processors. I've had one sample up to 4.1 GHz in the lab on air, and am excited to see what his story reveals.