by Ilya Gavrichenkov
10/23/2006 | 10:53 AM
The launch of Intel processors based on Core microarchitecture turned into a truly revolutionary event. These processors appeared a very successful solution for the desktop market, especially against the background of their predecessors: Pentium 4 and Pentium D CPUs based on NetBurst microarchitecture. Core 2 Duo processors for desktop PCs turned out not only faster but also much more economical than Pentium 4 and Pentium D solutions.
I believe it won’t be an exaggeration if I say that Core microarchitecture proved a powerful breakthrough that turned NetBurst based CPUs into the least competitive offer almost instantly, at least in the upper price range. Moreover, Intel Core 2 Duo processors became an excellent argument in the eternal rivalry with AMD, which Athlon 64 and Athlon 64 X2 processor families have been offering the best consumer features for a few years running. In other words, Intel Core microarchitecture became a significant milestone for the desktop segment, and the first CPUs based on it have changed the situation in the market radically. According to the result of numerous independent test sessions performed on Intel new processors, Core 2 Duo is undoubtedly the today’s fastest desktop CPU, which is also offering the best performance-to-power consumption ratio.
Our site has already paid a lot of attention to CPUs on Core microarchitecture designed for desktop platforms. However, it is important to mention once again that one of the greatest advantages of this microarchitecture specifically stressed at launch is its universality. The developers’ concept implies that the CPUs based on practically the same core with new microarchitecture inside should undergo very minor modifications in order to be suitable not only for desktop systems, but also for servers and mobile PCs.
Core architecture owes its superb flexibility to the opportunity to vary the maximum frequency and power consumption in relatively broad ranges. In other words, if the working frequency of CPUs with Core microarchitecture is lower than that of desktop CPUs, these processors can suit perfectly for economical mobile systems. So, our today’s article will be devoted to this particular application field.
So, today we are going to talk about mobile processors on Core microarchitecture also known as Merom and about mobile platforms built around them. I would like to point out that looking at Core microarchitecture from a new angle is more likely to give fresh food for thought than another desktop CPU review. The thing is that Merom processors came to replace Core Duo family (also known as Yonah), which were not based on NetBurst microarchitecture. Therefore, it would be incorrect to claim that we already know the outcome of the Core Duo against Core 2 Duo duel. Yonah CPUs feature their own mobile microarchitecture, with a few allowances they can be characterized as dual-core CPUs based on Pentium M that borrowed their microarchitecture from Pentium III. As for Core processors (Conroe, Merom and Woodcrest), they can be regarded as a further development of Yonah. In other words, mobile Core Duo and Core 2 Duo are close relatives, and it is really interesting to compare them against one another from both: theoretical and practical prospective. So, let’s get started!
Although Intel tries to convince unsophisticated users that processors with Core microarchitecture are further evolution of the mobile microarchitecture as well as of NetBurst microarchitecture, this statement arouses certain doubts. We believe that Conroe, Merom and Woodcrest processors hardly inherit anything from NetBurst, and Core microarchitecture should be regarded as the next step in the Pentium III – Pentium M – Core Duo evolution chain. This also results from the fact that the new CPUs feature short 14-stage pipeline and hence are not destined to hit any sky-high frequencies. Even the name of the processors based on new microarchitecture suggests the same: Core 2 Duo (the same name for desktop and mobile solutions).
We have already devoted the entire article to the details of the Core microarchitecture, so please feel free to check out our review called Getting Ready to Meet Intel Core 2 Duo: Core Microarchitecture Unleashed. However, when we discussed the peculiarities of Intel Conroe processors that are architecturally the same as our today’s heroes, the mobile Merom CPUs, we didn’t intend to compare them against Core Duo (Yonah). But today we are about to do it.
Let’s take a look at the improvements in the new Merom processor that are absent by Yonah. Although before we start let’s pay special attention to the similarities that make these two solutions somewhat related. The first thing that catches your eye right away is the fact that both – Yonah and Merom – feature dual-core structure with shared L2 cache memory. Both processors use Intel Smart Cache technology that allows both cores to use the same areas of the cache-memory and engage the amount of memory according to the current needs. Note however, that the size of available L2 cache memory can be different in Yonah and Merom, though it doesn’t affect the concept.
Moreover, both processors feature identical L2 cache memory that implies not only the same capacity, but also the same structure and organization. There are 32KB allocated for the instructions and for data.
The instructions execution scheme is also similar by Yonah and Merom. Both processors feature the execution pipeline of almost the same length, although the pipeline of the newer processor is 2 stages longer. This comes from the significant differences between these seemingly very similar processors. The engineering team that designed Merom introduced a lot of improvements for this new processor. The main improvement is certainly the support of 64-bit extensions of x86 Intel 64 architecture and the so called Intel Wide Dynamic Execution technology that implies that there are much more decoders and execution units in the new processor core.
I don’t want to overload my story with technical parameters, so here is a comparative table that will guide you through the major micro-architectural specs of the Yonah and Merom processors:
Intel Core Duo (Yonah)
Intel Core 2 Duo
L1 data cache
L1 instruction cache
L1 cache latency
L1 cache associativity
L2 cache size
L2 cache latency
L2 cache associativity
L2 cache bus width
1 complex and 2 simple
1 complex and 3 simple
Micro-ops fusion technology
Integer execution units
2 ALU + 2 AGU
3 ALU + 2 AGU
2 (1 Load + 1 Store)
2 (1 Load + 1 Store)
FP execution units
FADD + FMUL + FLOAD + FSTORE
SSE execution units
SSE instructions support
Intel 64 support
Note that besides the larger number of decoders and execution units, newer Merom processors also boast Macrofusion technology support. This technology allows increasing the instructions decoding speed by 1 instruction per cycle if the code has branch prediction algorithms. So, processors with Core microarchitecture can definitely process more instructions per clock cycle than the previous generation Yonah CPUs, at all stages.
As you can see from the table above, another advantage Merom has over Yonah is faster operation with SSE and FP instructions. It is achieved thanks to additional corresponding functional units as well as to expanded length of SSE operands.
Among other advantages of Merom processor that aren’t listed in the table above are greatly improved prefetch algorithms as well as memory disambiguation technology that makes out-of-order commands execution even more efficient.
In other words, despite significant similarities between Merom and Yonah processors, the former represents a tremendous step forward from the micro-architectural standpoint. Therefore, theory suggests that Merom should be a significantly faster mobile solution that Yonah. However, the performance is not the only thing that matters for notebook CPUs. The second most important aspect is power consumption that directly affects the battery life of the platform. That is why before we draw final conclusions regarding the prospects of the new Merom processor in the mobile segment, let’s address some of its other features as well.
Although mobile Merom processors are hardly any different from their desktop analogues aka Conroe, there are still some noticeable differences between these two processor families. In fact, this is not surprising at all, as pure system performance has never been a major parameter for notebook platforms. Mobile users care more about the performance-to-consumed power ratio. That is exactly why Merom processor family differs from Conroe processor family dramatically, even though they are both being sold under the same marketing name of Core 2 Duo.
In fact, the Core 2 Duo processors for desktop and mobile systems differ only by the clock speed and thermal and electrical specifications. In other words, Intel engineers reduced the Vcore and maximum clock speed barrier and this way turned Conroe into Merom that could be used in notebooks. Thus, the maximum Vcore of the mobile Core 2 Duo equals 1.3V and the maximum clock frequency is limited by 2.33GHz. It means that mobile processors on Core microarchitecture are 25% slower in terms of clock frequency than their desktop brothers. However, the typical heat dissipation of the mobile CPU rests within 34W thermal envelope, while the CPUs for desktop platforms can dissipate up to 65W (or even 75W for Extreme models).
However, although Core 2 Duo processors look very economical against the background of the desktop CPUs, they still lose to their predecessors in this respect. Dual-core Core Duo (Yonah) CPUs boasted lower heat dissipation of 31W maximum. Although they were working in about the same frequency range.
In order to better illustrate what has been just said, I would like to present a complete list of dual-core processors for notebooks that Intel is currently offering:
Core 2 Duo T7600
Core 2 Duo T7400
Core 2 Duo T7200
Core 2 Duo T5600
Core 2 Duo T5500
Core Duo T2700
Core Duo T2600
Core Duo T2500
Core Duo T2400
Core Duo T2300
As we see, the major formal specifications of the Yonah and Merom processors are not that very different. The same is true for the prices. Notebooks with similar features set based on Core Duo and Core 2 Duo belong to the same price category. In other words, Intel doesn’t make any principal difference between these two processors.
The close relationship between Yonah and Merom processors shows in the fact that both of them are part of the Centrino Duo mobile platform with the Napa codename. This platform was announced together with the Core Duo processors and hence is pretty mature already. We have reviewed this platform in great detail in our article called Centrino Duo Mobile Platform Review. Here I would only like to remind you that besides the dual-core mobile processors this platform also includes Intel 945PM/GM chipsets and Intel PRO/Wireless 3945ABG network adapter.
Note that it is a temporary thing that they use the same Centrino Duo platform for both, Yonah and Merom processors. Next April Intel is going to refresh its mobile platform with a new solution called Santa Rosa.
Although the CPU in this platform will remain the same, it will be designed for the new socket type. The chipset and the communications unit will also be different. Santa Rosa will include Core 2 Duo processors with 800MHz bus for Socket P, Crestline chipset (mobile analogue of the desktop Intel 965 solution) and Kedron communications unit. Among the advantages of the new promising platform will be much better integrated graphics optimized for Microsoft Windows Vista operating system, support of 802.11n WiFi with much higher data throughput rate, and Intel NAND Technology (Robson) that provides a built-in flash-memory cache speeding up OS and applications loading.
However, this is going to be tomorrow. Today, Core Duo and Core 2 Duo can be used in the same platforms, they are pin-to-pin compatible. In other words, all contemporary notebooks based on Centrino Duo platform can be equipped with either Yonah or Merom processors.
This particular fact helped us a lot when we picked out the hardware for our today’s test session. We tested two absolutely identical mobile platforms that differed from one another only by the CPU. These platforms were two notebooks from ASUS F3Ja family that can be equipped with different dual-core processors.
ASUS F3Ja notebook is a multimedia two-spindle solution with a 15.4-inch widescreen display with a standard resolution of 1280x800. The distinguishing feature of this notebook family is that they use external PCI Express graphics cards alongside with the mobile Napa platform.
The notebooks we got for our tests were built using Core Duo T2400 and Intel Core 2 Duo T5600 processors, Yonah and Merom respectively, working at the same clock speed of 1.83GHz and equipped with 2MB of cache memory.
Intel Core Duo T2400
Intel Core 2 Duo T5600
Both platforms were built around Intel 945PM Express (Calistoga) chipset with ICH7-M South Bridge. This chipset provides mobile platforms with high-performance dual-channel DDR2-667 SDRAM support, which is exactly what both our notebooks had. They were equipped with 1GB of this memory. Unfortunately, the Nanya memory used in ASUS platforms at this time works at the above mentioned frequency only with 5-5-5-15 timings.
ASUS F3J notebook family can come equipped with different discrete mobile graphics cards on ATI as well as Nvidia chips. The F3Ja notebook modification that we had featured ATI Mobility Radeon X1600 with 256MB of graphics memory that could be dynamically expanded to 512MB thanks to HyperMemory technology.
You can check out ASUS official website for the detailed notebook specifications. I would only like to specifically mention a few things that have direct influence on the performance. Since ASUS F3Ja notebooks are based on Centrino Duo platform, these mobile computers are equipped with the Intel PRO/Wireless 3945ABG network adapter with the PCI Express bus. They also featured Fujitsu MHV2120BH Serial ATA hard disk drives with 120GB storage capacity and 5,400rpm spindle rotation speed, as well as 8x DVD RW drives. Both notebooks were equipped with identical 4800 mAh batteries.
As usual, we checked the performance of our systems in general-purpose applications with the help of SYSMark2004 SE test. This benchmark emulates the user’s work in popular applications involving a lot of multi-tasking. Before we pass over to the results, we would like to stress that SYSMark2004 SE is primarily positioned as a desktop testing suite. Nevertheless, it includes a lot of applications that represent typical workload for a mobile system, especially for a system like high-performance Napa platform. That is why we believe it makes more sense to provide the detailed results report focusing specifically on each type of the workload.
In this case we have an image rendered by 3ds max 5.1 into a bmp-file, while the user is preparing web-pages in Dreamweaver MX. Then the user renders some 3D animation into vector graphics format.
Although this work model uses a lot of heavy applications hungry for processor resources, the performance difference between the systems on Yonah and Merom processors is not that significant. The use of Core 2 Duo T5600 processor instead of Core Duo T2400 generates a 5% performance improvement. Frankly speaking, I would expect much more from a 25% theoretical increase of the number of instructions processed per clock cycle and enhanced prefetch mechanism. However, this performance improvement can be observed not only in specifically selected tasks.
Now the test is emulating the user’s work in Premiere 6.5, when he is creating a video movie in raw-format from a few other movies and separate sound tracks. While waiting for the operation to be completed, the user is also modifying and saving to the hard drive a picture in Photoshop 7.01. When the video is finished, the user does the necessary editing and adds special effects to it in After Effects 5.5.
In this case the newer CPU with enhanced microarchitecture ensures greater performance growth of 8%. However, this result is again far below the theoretical expectations. Unfortunately, it makes us think that the increase in the number of Merom’s decoders and execution units wasn’t efficient enough.
This benchmark emulates the work of a professional webmaster. Here, our hypothetical user extracts from the zip-archive the web-site content and at the same time opens an exported 3D vector video in Flash MX. Then the user modifies it by adding some new pictures and optimizes it for faster animation. The final video with applied special effects is then compressed with Windows Media Encoder 9 so that it could later be broadcast via internet. The created web-site is then composed in Dreamweaver MX, while the system is scanned for viruses with VirusScan 7.0.
Core 2 Duo processor demonstrates 13% performance advantage over the predecessor working at the same clock speed in this benchmark. Of course, it owes this result to faster FP and SSE units. However, this advantage of the new microarchitecture is not too impressive, still.
In this case the scenario is quite typical of the average notebook user. Here the test is emulating the user’s work when he is receiving an e-mail in Outlook 2002 with a number of documents in a zip-file attached to it. While the files are being scanned for viruses with the VirusScan 7.0, the user is looking through the e-mails and makes notes in the Outlook calendar. After that the user checks a corporate web-site and some documents through Internet Explorer 6.0.
The new Core 2 Duo processor appears 5% faster here. It is definitely far from that outstanding performance boost that the new Core 2 Duo processors demonstrate in the desktop segment. It could be a clear indication of the great features of the Core Duo processors that have never been used in desktops.
In this test the hypothetical user is editing some test in Word 2002 and uses Dragon NaturallySpeaking 6 to convert an audio file into a text document. The document is then converted into pdf-format in Acrobat 5.0.5. After that the prepared document is used to create a PowerPoint 2002 presentation.
Core 2 Duo is 7% faster in this case, which once again demonstrates the great efficiency of the Pentium III microarchitecture announced in 1999, which then became the basis for the today’s CPUs with Core microarchitecture.
In the next test we see the following situation: the user opens a database in Access 2002 and creates a number of requests. The documents are archived with WinZip 8.1. The request results are exported into Excel 2002 and a diagram is created.
Enhanced data prefetch algorithms together with the effective use of the memory bus in Core 2 Duo processors have finally got a chance to show their real best. The mobile platform built with the new CPU proves 14% faster than the previous generation solution. However, if we once again recall the theoretical advantages of the new Merom platform over Yonah, the result will no longer seem that impressive anymore.
All in all, the average performance improvement provided by the new mobile Core 2 Duo processors in office and digital content creation applications equals about 9%.
PCMark05 is another popular testing suite that explores the general system performance as well as the performance of individual subsystems.
The diagram hardly reveals anything principally new. The notebook built with Core 2 Duo T5600 processor is about 7% faster than the notebook with Core Duo T2400 processor.
When we compare the performance during the execution of algorithms loading the computational processor resources most of all, the new microarchitecture manages to get a little farther ahead: the performance improvement makes 10%.
However, the most interesting results were revealed in the memory test. Two absolutely identical mobile systems equipped with the same dual-channel DDR2-667 SDRAM with 5-5-5-15 timings performed dramatically different. Keeping in mind that both processors, Core 2 Duo T5600 and Core Duo T2400 feature the same L2 cache memory and work at the same clock speed, the 25% performance advantage of the CPU with Core microarchitecture can only be assigned to the new memory technologies aka Intel Smart Memory Access. In reality, it is all about extremely aggressive data prefetch.
As for the performance of the tested notebooks in the gaming graphics applications, we resorted to 3DMark06 here. Note that thanks to ATI Mobility Radeon X1600 graphics adapter used in ASUS F3Ja notebooks, they performed quite well in contemporary 3D graphics applications. In other words, you can play modern games on these notebooks easily, and the results in the diagrams below prove it:
The tested notebooks provide quite acceptable level of graphics performance in 3D mode. The system equipped with Core 2 Duo processor runs about 3% faster than the competitor. This relatively small difference can be explained by the fact that this benchmark is primarily intended to estimate the graphics cards performance.
However, 3DMark06 also includes a test that allows estimating the processor performance during typical calculations in contemporary 3D games, i.e. when calculating the environmental physical and opponents’ AI.
Although the graphics performance is not of primary importance in this case, the difference in results is not dramatic again. Core 2 Duo T5600 is only 3.5% faster than Core Duo T2400.
So, the notebooks equipped with the ATI Mobility Radeon X1600 graphics card, like ASUS F3Ja solutions we are testing today, can be quite suitable for gaming. Therefore, we couldn’t omit any real gaming tests today.
It didn’t matter what CPU the notebook had: both platforms demonstrated pretty acceptable FPS rate in relatively contemporary 3D games. Although I have to admit that the newer Core 2 Duo processors ensure slightly better results. The platforms built with Core 2 Duo T5600 CPU outperformed its counterpart with Core Duo T2400 processor by 12% in Quake 4 and by 4% in Half Life 2. The results are certainly somewhat diverse, so the actual performance of the platforms with different processors in various games will depend on the parameters of the gaming engine. However, you shouldn’t forget that CPUs with Core microarchitecture have a few great advantages very useful in contemporary games: they work more efficiently with the memory bus and perform FP and SSE operations faster.
As we have already seen, contemporary notebooks are as fast and functional as mainstream desktop solutions. Therefore, audio and video encoding may become a typical task for them, too.
First of all we measured how fast they could convert audio files into popular MP3 format.
The advantage of the new mobile processor in Apple iTunes 7 is quite typical and makes around 7%.
Video encoding with the popular Xvid codec reveals much more differences between Yonah and Merom. Since Merom works much faster with SSE instructions, it proves almost 20% faster in this application than the predecessor.
We also see almost the same situation in Windows Media Encoder 9, which we have also used to test the video encoding speed. Core 2 Duo works almost 15% faster than Core Duo running at the same clock speed.
For this part of our test session we picked a few applications that may become quite common for contemporary notebooks these days.
Although the performance during data compression depends a lot on the memory subsystem, WinRAR archiving utility doesn’t reveal any significant difference between the two CPUs. The advanced data prefetch algorithms of the Merom processor may have turned out not effective enough for this particular type of work.
Adobe Photoshop, on the contrary, works faster with Core 2 Duo processor ensuring about 12% performance improvement.
The advantages of the new CPU with Core microarchitecture are even more impressive in Adobe Premiere Pro application for non-linear video editing. The notebook with the newer processor proves 15% faster than the rival. Of course, it is the fast SSE unit that ensures the victory of Core 2 Duo in the last two applications.
Core 2 Duo is faster than the predecessor in 3ds max, too. In viewports the performance difference is not as dramatic, however, during final rendering Merom gets far ahead.
Battery life is as important for mobile computer systems as performance. That is why we decided to pay special attention to this parameter under most typical types of workload. We tested the battery life with the help of Mobilemark2005 testing suite. Note that all the battery life tests were performed with maximum screen brightness and with the disabled automatic shift to StandBy mode.
The first scenario emulated the regular user work in typical office applications. Just as during our performance tests, the hypothetical user ran the following applications on the notebook: Microsoft Word 2002, Microsoft Excel 2002, Microsoft PowerPoint 2002, Microsoft Outlook 2002, Netscape Communicator 6.01, WinZip Computing WinZip 8.0, McAfee VirusScan 5.13, Adobe Photoshop 6.0.1 and Macromedia Flash 5. The script used in this case emulated the routine work of a car dealership employee.
If we compare the results of our performance tests with the results of our battery life tests, we will get quite a bit of new food for thought. Looks like higher performance of the new Core 2 Duo processor has another side to it. The notebook equipped with this processor has slightly shorter battery life that its counterpart equipped with a slower Core Duo CPU. So, the 10% performance increase we have just revealed during our test session costs Merom about 8% of its battery life. Therefore, we shouldn’t retire Core Duo CPUs just yet: they will be a great choice when longer battery life matters more than higher performance.
The second scenario that we used during our battery life test session emulated the usage model when the notebook is used for video playback. In particular, this test shows how long the notebook battery will last during DVD-movie playback via the InterVideo WinDVD 6.0 player.
The qualitative results are the same as in business applications. The system with Core Duo processor will ensure slightly longer DVD playback time when running on battery.
The third experiment we conducted measured the battery life during text reading. The text was displayed in NetScape Navigator 6.01.
Although when you read texts from the notebook screen, you do not spent so much battery resources, the picture hardly gets any different. The mobile systems with Core Duo processor works 6 minutes longer in this mode.
The last fourth scenario was aimed at measuring the battery life when the user is browsing internet. The usage model in this case is very simple: the internet surfing is performed in Microsoft Internet Explorer. The notebooks are connected to the internet via the built-in wireless network controllers that were absolutely identical in our case.
This time the notebooks equipped with Core Duo and Core 2 Duo processors works almost equally long. Although I still have to point out that the system with Core 2 Duo inside shut down 4 minutes earlier.
Summing up all the obtained results, I would like to say that it is not for nothing that the claimed typical heat dissipation of the Core 2 Duo processors is higher than that of the Core Duo. In reality it means that by hitting higher performance they spend more resources when running on battery. However, the maximum difference in battery life we managed to detect equaled only 8%, which can hardly be regarded as a dramatic difference.
The today’s results can hardly help us make any final conclusions. The thing is that we are still under great impression from the success of new CPUs on Core microarchitecture in the desktop market. Unfortunately, the situation with the mobile Core 2 Duo processors is not as rosy. While the desktop Core 2 Duo processors became significantly faster and set new performance records, the mobile Core 2 Duo couldn’t provide the same significant performance improvement compared with their predecessors – Core Duo CPUs.
According to our tests, Core 2 Duo are faster than Core Duo in all applications, however the average performance improvement (provided the clock speed is equal) is less than 10%. And this is a purely evolutional change. In other words, despite all significant micro-architectural improvements that were introduced when shifting from Core Duo to Core 2 Duo, the practical performance increase in the mobile platforms didn’t prove up to our expectations. Of course, video application are working much faster now, sometimes even 20% faster, but video editing is not a really typical task for mobile platforms. So, the significance of the mobile Core 2 Duo launch is not in the higher performance, but in the Intel 64-bit modes support that has been introduced in the notebook platforms with the arrival of the new generation of mobile processors.
In addition, I would also like to say that higher performance of the Core 2 Duo processors is not free and results in higher power consumption. As a result, mobile platforms equipped with Core 2 Duo processors run slightly less on battery than the systems featuring Core Duo inside. However, I have to be fair here: the performance improvement is still higher than the power consumption loss.
In conclusion, I would like to point out again that the arrival of mobile Core 2 Duo processors will not cause any revolutionary changes in the notebook market. Especially, since Intel is not updating the entire platform just yet, but offers to use new CPUs in the old Napa platform. So, if you already have a Napa based notebook with the dual-core Core Duo CPU, there is hardly any real need to upgrade.
A for the real revolution in the mobile market, we should wait for the next spring, when the expanded Core 2 Duo processor family with higher frequency models in it will acquire a new chipset with better integrated graphics and Robson technology, as well as a new wireless network component with higher data transfer rate.