|Introduction and Related Information|
It is fairly safe to say that Intel has far and away been the dominant player in the mobile processor market for what seems like forever. Though AMD's Turion 64 is a solid alternative to Intel solutions, clearly AMD's market share of 13% or so is an obvious sign that the company just hasn't been able to penetrate the Notebook space with the same level of success they have in the Desktop and Server markets. And while the prospects of a combined AMD/ATI total solution approach for notebooks is encouraging, Intel continues its onslaught of Core 2 releases which today also officially takes flight in the Mobile CPU arena.
With the performance profile we saw in Conroe testing on the desktop, it certainly looks like Intel is poised to maintain or widen that market share with Merom in notebooks. We'll step you through key specifics of Merom and how it performs in the pages ahead, so you can make your own judgments.
The long and short of the specification list we've noted above is that a Core 2 Duo processor for Mobile/Notebook applications is essentially built on the identical CPU architecture that Core 2 Duo processors for the Desktop are (and even Core 2's Server versions for that matter).
Clearly, a common architecture approach offers huge economies of scale efficiencies from a manufacturing standpoint for Intel. Beyond its Socket 479 PGA packaging, the only difference between the Desktop and Mobile versions of Core 2 Duo is that Merom is built with more aggressive power savings in mind. Additionally, Conroe chips have a measured operating voltage of around 1.3V while, from our initial observations, Merom runs in the 1.1V range.
A closer look at Merom is next...
|A Closer Look at Merom|
Intel's Mobile Core 2 Duo, otherwise known as Merom, is a completely revamped dual core processor based on Intel's "Core" micro-architecture. This architecture differs greatly from the Core Duo "Centrino" processors that are built with a pair of Intel Dothan (Pentium M) core CPUs under the hood.
Merom, like the Core 2 Duo for Desktops, has a unified 4MB L2 cache that contributes to a significantly larger die size versus Core Duo Dothan-based dual core processors that have only 2MB of cache. The additional cache offers substantial improvements in latency and bandwidth since more operations can be processed completely on-chip rather than having to go out to external system memory.
Finally, the good news for current Core Duo "Centrino" customers is that Merom/Mobile Core 2 Duo is also backwards compatible with most existing Core Duo-based notebooks that are built on Intel's Mobile 945 Express chipset. Some systems may require a simple BIOS update, while others will be a completely plug-and-play affair.
Sony Viao SZ150 Notebook
For reference on how to actually perform a Core 2 Duo transplant, we invite you to check out our recently published Do-It-Yourself Whitebook article here.
|Notebook Test System, Vital Signs and Power Consumption|
We tested Intel's Merom Mobile Core 2 Duo in our Asus Z96JS whitebook that we showed you here recently. It provided a solid test platform to easily compare and contrast performance characteristics of Intel's legacy Core Duo (Dothan) architecture versus Merom (Core 2 Duo).
Taking a look at readings from CPU-Z, we see similarities between the older Core Duo architecture and Core 2 Duo. Below are screen-shots from CPU-Z with a Core Duo T2600 (Yonah) chip, a Core 2 Duo T7600 (Merom) processor and finally a Core 2 Duo E6700 (Conroe) chip.
First, the similarities. For starters, both Yonah and Merom run on a 667MHz FSB. In the case of our T2600 Yonah chip, a 166MHz clock is driven into a X13 multiplier for the core CPU speed and a X4 multiplier for the FSB. Our T7600 Merom is clocked likewise with the exception of a X14 multiplier for its core speed of 2.33GHz versus the X13 multiplier and core speed of the 2.16GHz T2600. And that's where the similarities end for the mobile processors. The rest of the architecture for Merom is much more akin (or should we say "clone" even) to Conroe.
First and foremost, Mobile Core 2 Duo and Desktop chips support Intel's updated SSE instruction set that CPU-Z likes to call SSE4. Also, both Merom and Conroe can process EM64T instructions for Windows XP 64-bit support. And finally both Merom and Conroe have the same 4MB L2 cache and 32K Instruction and 32K Data cache architectures, although there are 2MB variants coming as well. The primary differences between the Mobile Core 2 Duo and Desktop chips are their operating voltages -- 1.163V observed here for Merom and 1.32V observed for Conroe -- along with a much faster FSB for Conroe at 1066Mhz.
But does this upgrade in Intel's Mobile CPU architecture come with a power consumption cost?
Clock-tick for clock-tick the answer is a resounding "No". In the graph here, we've listed total system power consumption for our Asus Z96JS whitebook at idle speeds for the processor and then again under 100% CPU load. We took readings on power draw from the system's AC adapter connected to our lab's wall outlet. Some quick math reveals that Merom's 2.33GHz clock speed advantage over the T2600 (Yonah) 2.16GHz chip almost directly correlates to its additional power consumption. To be exact, our Core 2 Duo T7600 is clocked around 8% faster than the Core Duo T2600 and likewise only draws about 9-10% more juice.
When you consider the fact that both Yonah and Merom are both built on Intel's 65nm process and that Merom has an additional 2MB of L2 cache along with EM64T support, Intel's design engineering triumph with Core 2 Duo becomes crystal clear. Intel's Mobile Core 2 Duo offers more processing power, capacity, features, and does so at power parity versus their legacy mobile Core Duo architecture. Against the backdrop of Intel's Pentium 4 Netburst architecture, Conroe's advancements resulted in impressive and massive gains from a performance-per-watt perspective. In kind, Merom seems to offer similar gains versus the already power-efficient Yonah architecture. Let's take a look at the hard data next and confirm these statements.
|PCMark05 - General Usage Performance|
Futuremark's PCMark05 offers a quick-take benchmark analysis with various standard day-to-day computing usage models. The following is a quote directly from Futuremark that explains exactly what these tests do and how they work.
"The CPU test suite is a collection of tests that are run to isolate the performance of the CPU. The CPU Test Suite also includes multithreading: two of the test scenarios are run multithreaded; the other including two simultaneous tests and the other running four tests simultaneously. The remaining six tests are run single threaded. Operations include, File Compression/Decompression, Encryption/Decryption, Image Decompression, and Audio Compression" - Courtesy FutureMark Corp.
As we noted, our Mobile Core 2 Duo T7600 Merom chip has roughly an 8% clock speed advantage over our T2600 Core Duo Yonah chip. However, in terms of raw CPU throughput, PCMark05's CPU test module shows a 17%+ advantage for Core 2 Duo T7600 at 2.33GHz.
"The Memory test suite is a collection of tests that isolate the performance of the memory subsystem. The memory subsystem consists of various devices on the PC. This includes the main memory, the CPU internal cache (known as the L1 cache) and the external cache (known as the L2 cache). As it is difficult to find applications that only stress the memory, we explicitly developed a set of tests geared for this purpose. The tests are written in C++ and assembly. They include: Reading data blocks from memory, Writing data blocks to memory performing copy operations on data blocks, random access to data items and latency testing." - Courtesy FutureMark Corp.
PCMark05's Memory subsystem test shows roughly a 35% advantage for the Core 2 Duo T7600 Merom processor versus the fastest Yonah Core Duo chip on the market. This is undoubtedly attributed to the chip's additional 2MB of on board cache, more advanced pre-fetch algorithms that Intel's dubs "Smart Memory Access", and the shared-memory cache architecture dubbed "Smart Cache" that is employed in Merom as well as its brethren Conroe desktop chip.
|Office XP and Photoshop 7|
With WorldBench 5.0, we run a series of pre-configured scripts that we believe give a well rounded view of system performance. The tests we focus on are Office XP SP2, Photoshop 7, Nero Burning ROM and Windows Media Encoder while multitasking with a Mozilla-based web browser.
Below we have the results from our WorldBench 5 Photoshop 7 and Office XP tests. These results are recorded in seconds to complete all workloads. Lower times indicate better performance here, so the shorter the bar the better.
Though the Core 2 Duo T7600 (Merom) does show a distinct performance advantage here, the results are more in line with the CPU's clock speed advantage over the fastest Core Duo T2600 (Yonah) chip in our test system. The basic Office applications don't exploit the additional resources available in our mobile Core 2 Duo Merom processor.
Our Photoshop 7 test taxes the memory subsystem a bit more and as a result, Merom at 2.33GHz offers roughly a 14% performance advantage over the fastest Yonah chip in our test and it's over 25% faster than the 1.83GHz clocked Core Duo T2400. Mobile Core 2 Duo's performance advantage is impressive here but maybe we could let Merom stretch its legs a bit more in a few encoding tests.
|Windows Media Encoder/Mozilla MT and Nero|
Media Encoding has always been a strength with the Intel platform and below we have tests from WB5's Windows Media Encoder Multitasking test that also runs an instance of a Mozilla web browser in the background.
Merom once again shows a more significant performance advantage beyond just its clock speed. Here we see the Core 2 Duo T7600 sneaking in over 12% faster than the fastest Core Duo Yonah chip.
Things got a little squirrely with our Nero Burning ROM tests and due to the time crunch we were under with this launch, we had to forego testing the Core Duo T2400 chip. We are confident in the numbers shown for both the Core 2 Duo T7600 and Core Duo T2600 here though, as we re-ran the test several times with repeatable results. This version of Nero doesn't really exploit the new Mobile Core 2 Duo architecture all that much and Merom only shows a smallish 4% edge or so versus Yonah.
|3DMark06 DX Gaming Performance|
As you'll recall from our system specs, the Asus Z96JS whitebook we tested was setup for moderate gaming capabilities with its integrated mid-range Mobility Radeon X1600 graphics subsystem. This gave us the ability to test Merom in a gaming environment and as such we have a few standard game-test metrics in the pages ahead.
3DMark06's built-in CPU test is a multi-threaded "gaming related" DirectX metric that's useful for comparing relative performance between similarly equipped systems. This test consists of two different 3D scenes that are generated with a software renderer, which is dependant on the host CPU's performance. This means that the calculations normally reserved for your 3D accelerator are instead sent to the central processor. The frame rates recorded in each test are used to determine a final weighted score.
The 3DMark06 CPU test definitely enjoys Merom's roomier 4MB cache in addition to its overall processing throughput advantage. From an available CPU resources perspective, our 2.33GHz T7600 is over 18% faster than the T2600 and again, that's with only an 8% clock speed advantage. Further, the T7600 driven system is 34% faster than the T2400 system.
The "default" 3DMark06 test also heavily taxes the graphics subsystem in our Asus Notebook test machine. As a result the scores are significantly less dispersed between all of the processors we tested. Here our 2.33GHz T7600 enabled system is showing only about a 2% advantage over the 2.16GHz T2600. This is a direct result of the test being completely GPU limited by the Asus machine's Radeon X1600 graphics processor. Had we tested with a higher-end Mobility Radeon or NVIDIA GeForce Go 7 series graphics under the hood, scores would have been significantly different.
|Quake 4 Gaming Performance|
Some may consider 3DMark06 scores to be a bit nebulous relative to any sort of real-world gaming performance assessment, so we then fired up the Id's venerable Quake 4 engine for some straight forward frame rate recording with our various Intel mobile processors in the socket.
First we tested various Core Duo and Core 2 Duo mobile processors in what by all rights should be considered a CPU-limited test methodology. Low quality special effects and image quality were selected and a low resolution of 640x480 was set. The results were surprising to say the least...
The moral of the story here? You simply don't need a ton of CPU power to run Quake 4 at reasonable frame-rates; you need GPU power. For example, with a GeForce 7900 GTX powering the graphics pipeline, a Core 2 Duo E6700 chalks up around 190 fps in this testing scenario.
At higher resolutions and image quality, the results scale accordingly and with almost identical variance between each mobile processor type that was utilized in the CPU socket.
|Cinebench 3D Rendering|
To further illustrate the point we've made relative to the test system being GPU-limited, we also ran Cinebench which is a 3D Rendering benchmark based on Maxon's professional-grade Cinema 4D rendering software.
The Cinebench 2003 benchmark is an OpenGL 3D rendering performance test, based on the commercially available Cinema 4D software suite. This is a multi-threaded, multi-processor aware benchmark that renders a single 3D image and tracks the length of the entire process. The time it took each test system to render the entire scene is represented in the graph below (listed in seconds).
Our Cinebench results show us what the Core 2 Duo T7600 is capable of, if it were not held back by the Mobility Radeon X1600 is our test system. In both multithreaded and single-threaded test runs the Core 2 Duo T7600 Merom processors is 25%+ faster than the fastest Core Duo T2600 chip and 48-50% faster than the 1.83GHz Core Duo T2400. This is easily the largest performance advantage we've seen in our testing here with Merom and perhaps a hint of the potential for additional headroom in the future for Mobile Core 2 Duo as third-party software is continually optimized for the architecture.
On a side note, just for reference, a similarly equipped Athlon 64 FX-62 dual core processor based system scores about the same as our Core 2 Duo T7600 posted here. That's a full-power 2.8GHz dual-core desktop chip versus a 2.33GHz notebook CPU -- impressive to say the least.
For our battery life tests, we used the standard benchmark settings from Bapco's MobileMark, along with a few other minor system tweaks. The screensaver was disabled and the volume was set to off.
MobileMark runs an instance of the Netscape web browser reading a large file of the epic novel "War and Peace". Battery time is calculated in number of minutes to exhaustion with 1% battery life left.
In the realm of semiconductors, die size and clock speed is directly proportional to power consumption. The Core 2 Duo Merom processor, with is 2.33GHz clock speed an additional 2MB of L2 cache, as we've shown you, offers significant performance gains over Core Duo at 2.16GHz. However, these gains do not come completely for free. Battery life with a 2.16GHz Yonah-based Core Duo T2600 under the hood is extended by an additional 17 minutes of operation. A Core 2 Duo T7400 at 2.16GHz would definitely offer a more comparable platform test but unfortunately our whitebook test system did not allow us to adjust multipliers or clock speeds.
A rough clock speed/power consumption estimate leads us to believe that an identically clocked T7400 Merom chip versus a T2600 Yonah would still not quite be on par but it would be very close, well under a 10% differential most likely. Our performance testing with the likes of Cinebench, in conjunction with this battery life test, speaks volumes for the performance-per-watt value that Merom brings to the table.
|Performance Analysis and Conclusion|
It's hard not to be thoroughly impressed with the type of performance gains that Merom brings to Intel's notebook platform. Rarely are we able to observe large performance increases like this with the launch of a new processor core, especially at relatively comparable power consumption characteristics versus previous generation micro-architectures. If Intel was noted as the dominant player in notebooks before, it can most likely be said that they're going to "own" that market now. AMD obviously has some serious catch-up work ahead of themselves. Perhaps with ATI's competency in mobile chipsets, AMD will gain some much needed traction, but that is a rather large and complex equation to calculate at this juncture.
Unfortunately, we were not able to get a comparable Turion 64 X2-based notebook in time for testing, but when last we checked, the Turion single core was roughly on par or slightly below a single core Pentium-M performance-wise. Core Duo offered significant performance gains over the Pentium-M, and with Mobile Core 2 Duo's (Merom) debut here today, the performance and features gap widens in Intel's favor. Though AMD has been able to inflict a serious chink in Intel's armor in the Desktop and Server markets, in Mobile computing, "Intel Inside" is an understatement. As an aside, we intend to follow-up on this article with some actual Turion 64 X2 performance testing, in the near future hopefully.
Pricing for Intel's new line-up of Mobile Core 2 Duo processors for notebooks will roll out as follows:
With near immediate availability in the channel and with complete backwards compatibility for socket 479 - 945 Express based machines, we're certain Merom will sell like hot-cakes, just like Conroe has. Not to mention, Intel's Easy-Bake oven is going to punch out these chips like Martha Stewart jacked-up on Red Bull smoothies.