cpu dual core2006.pdf

Upload: melody-cotton

Post on 04-Jun-2018

218 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/14/2019 CPU dual core2006.pdf

    1/38

  • 8/14/2019 CPU dual core2006.pdf

    2/38

    of these new solutions will be not their features but mostly their appealing price.

    Once these three CPUs are launched, the prices on the already existing Core 2 Duo processors will evidentlydrop. As a result, Intels Core 2 Duo price list will then look as follows:

    CPU Clock frequency Bus frequency L2 cache Price

    Core 2 Duo E6700 2.66 GHz 1066 MHz 4 MB $316

    Core 2 Duo E6600 2.4 GHz 1066 MHz 4 MB $224

    Core 2 Duo E6420 2.13 GHz 1066 MHz 4 MB $183

    Core 2 Duo E6400 2.13 GHz 1066 MHz 2 MB $183

    Core 2 Duo E6320 1.86 GHz 1066 MHz 4 MB $163

    Core 2 Duo E6300 1.86 GHz 1066 MHz 2 MB $163

    Core 2 Duo E4400 2.0 GHz 800 MHz 2 MB $133

    Core 2 Duo E4300 1.8 GHz 800 MHz 2 MB $113

    Pentium D 935 3.2 GHz 800 MHz 2 x 2 MB $84

    Pentium D 925 3.0 GHz 800 MHz 2 x 2 MB $74

    As we see, all three upcoming CPUs are quite interesting. Core 2 Duo E6x20 processors offer larger cachememory at the same price point as the well-known Core 2 Duo E6400 and E6300. Core 2 Duo E4400, in its turn,can boast higher clock speed than Core 2 Duo E63x0 but at a lower price.

    Now lets take a closer look at the new processors.

    Core 2 Duo E6420 and Core 2 Duo E6320

    From the technological standpoint Core 2 Duo E6420 and E6320 do not have any new features at all. They arethe same CPUs on Conroe core with 4MB L2 cache memory. Their clock speeds are 2.13GHz and 1.86GHzrespectively. They support 1066MHz FSB. In other words, their only difference from the Core 2 Duo E6400 andE6300 is the large L2 cache. And since they cost the same as their predecessors, Core 2 Duo E6420 and E6320will eventually oust them from the market.

    The detailed specifications of the Core 2 Duo E6420 and E6320 prove that these processors are analogous to thetop CPUs on Conroe core. Only the clock speeds are different:

    Core 2 Duo E6420 Core 2 Duo E6320

    Processor core Conroe

    Nominal frequency 2.13 GHz 1.86 GHz

    Bus frequency 1066 MHz

    Multiplier 8x 7x

    L2 cache 4 MB

    Packaging LGA775

    Manufacturing process 65 nm

    Core stepping B2

    TDP 65 W

    Vcore 1.187-1.325 V

    Enhanced Halt State (C1E) Technology Yes

    Enhanced Intel Speedstep Yes

    Execute Disable Bit Yes

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    8 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    3/38

  • 8/14/2019 CPU dual core2006.pdf

    4/38

    However, if you take a closer look at the screenshots you will notice that the popular CPU-Z utility is not yetquite familiar with the Core 2 Duo E6x20 processors mistakenly recognizing them as Core 2 Duo E6x00.

    Core 2 Duo Cache Memory and Performance

    Since new Core 2 Duo E6x20 processors with larger L2 cache memory will be priced just as the Allendale basedCPUs working at the same frequencies (this core is the same Conroe core with reduced L2 cache), it will beextremely interesting to find out how the performance of CPUs on Core micro-architecture depends on the sizeof the L2 cache. Therefore, we decided to compare the performance of the Core 2 Duo E6420 and Core 2 DuoE6400 side by side in the most popular applications. The obtained results are given in the table below:

    Core 2 Duo E6420

    4MB L2

    Core 2 Duo E6400

    2MB L2Performance Hit

    3DMark06 5948 5928 0.3%

    3DMark06, CPU 1797 1774 1.3%

    PCMark05 6673 6582 1.4%

    PCMark05, CPU 5196 5174 0.4%

    Word 2007 (Document Compare), sec 51 54 5.9%

    Excel 2007, sec 14.4 18.4 27.8%

    7-Zip 4.44, Compressing, KB/s 3023 2789 8.4%

    7-Zip 4.44, Decompressing, KB/s 21076 20983 0.4%

    MPEG-4 Encoding, AutoGK 2.4/Xvid 1.2, fps 32.16 30.91 4.0%

    H.264 Encoding, Apple Quicktime Pro 7, sec 493 498 1.0%

    mp3 Encoding, Apple iTunes 7, sec 156 157 0.6%

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    8 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    5/38

    Adobe Photoshop CS3 (ps7bench 2.0), sec 105 107 1.9%

    Windows Photo Gallery (Print), sec 31 32 3.2%

    Adobe Premiere Pro 2.0, sec 245 246 0.4%

    Windows Movie Maker, sec 102 103 1.0%

    3ds Max 8 (Space Flyby), sec 263 265 0.8%

    CINEBENCH 9.5, Rendering 694 693 0.1%

    POV-Ray 3.7 829.1 820.1 1.1%

    Quake 4, 1024x768 High Quality 90.01 83.24 8.1%

    F.E.A.R., Medium Quality 104 98 6.1%

    Unreal Tournament 2004, 1024x768 79.9 76.8 4.0%

    Company of Heroes, 1024x768 115.4 112.6 2.5%

    Valve VRAD map build benchmark, sec 335 336 0.3%

    Valve Source Engine particle benchmark 34 31 9.7%

    Fritz 9 Chess Benchmark 3106 3078 0.9%

    The results we obtained in 22 popular applications prove that cache size does have significant influence on the

    system performance. Although the average advantage of the Core 2 Duo E6420 over Core 2 Duo E6400 equaledonly 3%-4%, the actual performance gain depends a lot on the type of workload.

    Office applications and games stand out among all other benchmarks. In this type of applications cache poses thebiggest influence on the system performance. In both: contemporary games as well as typical office tasks, Core 2Duo E6420 can be as much as 6-8% faster than Core 2 Duo E6400. There are even greater examples of the

    practical value of the large L2 cache. For instance, Valve Source Engine particle benchmark that shows theprocessor performance when calculating the physical model of the gaming world demonstrates a 10%performance advantage of the newcomers over the older processor models. But the most impressive result of allcan definitely be observed in Excel 2007. The CPU with 4MB L2 cache hit the impressive performance gain of27.8% during typical statistical operations.

    However, there are also some applications where L2 cache memory has very little influence on the results. Herewe could list final rendering tasks as well as some sound, video and image editing procedures.

    Nevertheless, we should give due credit to Intel for the introduction of fully-fledged Conroe core in the youngermodels from the Core 2 Duo E6000 series without increasing their price. 4MB L2 cache memory makes Core 2Duo E6420 and E6320 processors even more attractive for the end-users than Core 2 Duo E6400 and E6300.

    Core 2 Duo E4400

    Since the youngest Core 2 Duo processors from the E6000 series will acquire twice as large cache-memory andfully-fledged Conroe core, Core 2 Duo E4000 family will remain the only CPUs with smaller 2MB L2 cache.

    Some time ago we discussed Core 2 Duo E4300processor. It was the first CPU on Allendale core, which hasonly 2MB L2 cache on the physical semiconductor die. This was what allowed Intel to reduce their productioncost and set very attractive prices for these processors.

    Core 2 Duo E4400 processor that will come out together with the Core 2 Duo E6420 and E6320 will be thesecond CPU in the todays youngest processor line-up on Core micro-architecture. Just like Core 2 Duo E4300,the new Core 2 Duo E4400 will feature not only smaller 2MB L2 cache, but also 800MHz system bus. As for theclock frequency, this newcomer will run 200MHz faster than E4300: at 2.0GHz.

    The detailed processor specs are given in the table below:

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    8 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    6/38

    Core 2 Duo E4400

    Processor core Allendale

    Nominal frequency 2.0 GHz

    Bus frequency 800 MHz

    Multiplier 10x

    L2 cache 2 MB

    Packaging LGA775

    Manufacturing process 65 nm

    Core stepping L2

    TDP 65 W

    Vcore 1.225-1.325 V

    Enhanced Halt State (C1E) Technology Yes

    Enhanced Intel Speedstep Yes

    Execute Disable Bit Yes

    Intel EM64T Yes

    Intel Thermal Monitor 2 Yes

    Intel Virtualization Technology None

    As we see, unlike the CPUs from the E6000 family, Core 2 Duo E4400 is based on a different core stepping: L2.As far as we know, the overclocking potential of this core stepping is slightly worse than that of the fully-fledged B2 core stepping. Among other distinguishing features we should definitely mention the absence ofIntel Virtualization Technology support. However, this technology is hardly ever used in desktop systems ofregular computer users, so its absence is not that much of a loss in the end.

    CPU-Z utility recognizes the core stepping as L2:

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    8 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    7/38

    By the way, the electronic components located at the bottom of the CPU also indicate that it belongs to L2 corestepping. Take a look:

    Conroe B2 stepping (left), Allendale L2 stepping (right)

    Clock Frequency vs. FSB Speed and L2 Cache Size

    One of the main attractions of the Core 2 Duo E4400 processor is its price. The thing is that this processorworking at 2.0GHz clock speed is at the very bottom of Intels price-list, below all Core 2 Duo E6000

    processors. In other words, it costs even less than Core 2 Duo E6320 with lower clock frequency of 1.86GHz.Looks like Intel believed that 800MHz FSB and 2MB L2 cache would make Core 2 Duo E4400 slower thanCore 2 Duo E6320.

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    8 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    8/38

    However, we have our concerns regarding this. To prove one way or another we decided to compare theperformance of the Core 2 Duo E4400 and Core 2 Duo E6320 in the popular applications:

    Core 2 Duo E44002.0GHz, 2MB L2, 800MHz

    FSB

    Core 2 Duo E63201.86GHz, 4MB L2, 1066MHz

    FSB

    3DMark06 5830 5771

    3DMark06, CPU 1663 1594

    PCMark05 6079 6084

    PCMark05, CPU 5039 4726

    Word 2007 (Document Compare), sec 57 57

    Excel 2007, sec 20.5 16.3

    7-Zip 4.44, Compressing, KB/s 2605 2841

    7-Zip 4.44, Decompressing, KB/s 19719 18455

    MPEG-4 Encoding, AutoGK 2.4/Xvid 1.2,fps

    28.68 28.36

    H.264 Encoding, Apple Quicktime Pro 7,

    sec

    531 564

    mp3 Encoding, Apple iTunes 7, sec 168 178

    Adobe Photoshop CS3 (ps7bench 2.0), sec 112 117

    Windows Photo Gallery (Print), sec 34 35

    Adobe Premiere Pro 2.0, sec 259 269

    Windows Movie Maker, sec 113 113

    3ds Max 8 (Space Flyby), sec 283 297

    CINEBENCH 9.5, Rendering 643 609

    POV-Ray 3.7 775.2 722.8

    Quake 4, 1024x768 High Quality 77.2 85.51

    F.E.A.R., Medium Quality 89 98

    Unreal Tournament 2004, 1024x768 71.5 71.6

    Company of Heroes, 1024x768 106.1 108.2

    Valve VRAD map build benchmark, sec 358 381

    Valve Source Engine particle benchmark 28 31

    Fritz 9 Chess Benchmark 2856 2734

    The obtained results turned out pretty ambiguous. Core 2 Duo E6320 cannot boast any indisputable advantageover the cheaper Core 2 Duo E4400. There still are a lot of tasks where pure computational power of the CPU

    plays depending on its clock frequency the most important role. Among these tasks we can list image, sound andvideo editing applications as well as final rendering. In other words, Core 2 Duo E6320 is only faster than Core 2Duo E4400 in office tasks and 3D games.

    So, if you are looking for a low-cost dual-core processor from the Core 2 Duo family, you may face a relativelydifficult task: you need to determine what type of applications youll be running most often on this system andwhat type of performance will be most critical for your needs. However, since Core 2 Duo E4400 is $30 cheaperthan Core 2 Duo E6320, your choice may stay with the former even though it is not that fast in gamingapplications.

    Testbed and Methods

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    8 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    9/38

    Massive price reduction and launch of the new low-cost Core 2 Duo processors are two remarkable events thatcan change the situation in the corresponding market segment dramatically. Therefore, we decided to perform afull round of benchmarks in the new Microsoft Windows Vista OS when Core 2 Duo E6420, E6320 and E4400hit our lab. We will look at all dual-core processor models that will be priced below $200 this coming April. As aresult, the complete list of our todays testing participants included 18 processors from Athlon 64 X2, Core 2Duo and Pentium D families, including Core 2 Duo E6600 and Athlon 64 X2 5000+, although their price is alittle higher than the 200-dollar maximum we set.

    Here is the list of all hardware including the full list of participating processors that was used in our test session:

    CPUs:AMD Athlon 64 X2 5000+ (Socket AM2, 2.6GHz, 2x512KB L2, Windsor );AMD Athlon 64 X2 4800+ (Socket AM2, 2.5GHz, 2x512KB L2, Brisbane);AMD Athlon 64 X2 4600+ (Socket AM2, 2.4GHz, 2x512KB L2, Windsor );AMD Athlon 64 X2 4400+ (Socket AM2, 2.3GHz, 2x512KB L2, Brisbane);AMD Athlon 64 X2 4200+ (Socket AM2, 2.2GHz, 2x512KB L2, Windsor );AMD Athlon 64 X2 4000+ (Socket AM2, 2.1GHz, 2x512KB L2, Brisbane);AMD Athlon 64 X2 3800+ (Socket AM2, 2.0GHz, 2x512KB L2, Windsor );AMD Athlon 64 X2 3600+ (Socket AM2, 1.9GHz, 2x512KB L2, Brisbane);Intel Core 2 Duo E6600 (LGA775, 2.4GHz, 1067MHz FSB, 4MB L2, Conroe );Intel Core 2 Duo E6420 (LGA775, 2.13GHz, 1067MHz FSB, 4MB L2, Conroe );Intel Core 2 Duo E6400 (LGA775, 2.13GHz, 1067MHz FSB, 2MB L2, Allendale);Intel Core 2 Duo E6320 (LGA775, 1.86GHz, 1067MHz FSB, 4MB L2, Conroe );Intel Core 2 Duo E6300 (LGA775, 1.86GHz, 1067MHz FSB, 2MB L2, Allendale);Intel Core 2 Duo E4400 (LGA775, 2.0GHz, 800MHz FSB, 2MB L2, Allendale);Intel Core 2 Duo E4400 (LGA775, 1.8GHz, 800MHz FSB, 2MB L2, Allendale);Pentium D 945 (LGA775, 3.4GHz, 800MHz FSB, 2x2MB L2, Presler);Pentium D 935 (LGA775, 3.2GHz, 800MHz FSB, 2x2MB L2, Presler);Pentium D 925 (LGA775, 3.0GHz, 800MHz FSB, 2x2MB L2, Presler).

    Mainboards:ASUS M2N32-SLI Deluxe (Socket AM2, NVIDIA nForce 590 SLI);

    ASUS P5N-E SLI (LGA775, NVIDIA nForce 650i SLI).Memory: 2048MB DDR2-800 SDRAM (Mushkin XP2-6400PRO , 2 x 1024 MB , DDR2-800, 4-4-4-12);Graphics card: PowerColor X1900 XTX 512MB;HDD: Western Digital WD1500AHFD.OS: Microsoft Windows Vista Ultimate (64-bit).

    We ran the tests with the BIOS Setup of the mainboards adjusted for maximum performance.

    Performance

    Futuremark: Synthetic Benchmarks

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    8 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    10/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    11/38

    In the popular 3DMark06 test AMD Athlon 64 X2 and Intel Core 2 Duo show some parity. The CPUs from thesesimilarly priced processor families perform equally fast with a slight advantage in favor of products on Coremicro-architecture. As for the Pentium D performance, it proved surprisingly decent for solutions based onobsolete NetBurst micro-architecture. Although these CPUs are slower than Core 2 Duo, they offer better price-to-performance ratio than the Athlon 64 X2 processors.

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    12/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    13/38

    Another test, Futuremark PCMark05, checks the general system performance, not just the gaming speed. Thistest demonstrated slightly different results. Core 2 Duo family takes the lead here leaving competitors fromAMD far behind. Pentium D family also performs quite well here. These processors can even catch up with afew more up-to-date Core 2 Duo representatives in the CPU subtest.

    Gaming Applications

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    14/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    15/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    16/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    17/38

    The situation in contemporary games is quite logical. Core 2 Duo processors offer the best performance here.AMD CPUs from the Athlon 64 X2 family show much lower fps rate. As for the old Pentium D processor family,these guys can no longer compete neither with Core 2 Duo, nor with AMD rivals.

    Besides a few real games, we have also included the results of two benchmarks based on Valve Source code thatwill be used for future games development. The first benchmark estimates how fast the testing participants copewith building the lighting maps, and the second evaluates the performance during environmental physics

    processing.

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    18/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    19/38

    Typical computational task, building lighting maps is performed faster on Core 2 Duo processors. As for physicalcalculations, the results are not so straightforward here. Athlon 64 X2 processors on the older 90nm core work asfast as the Core based CPUs. However, newer 65nm Brisbane core that finds its way into dual-core AMD

    processors cannot yet boast good results. Athlon 64 X2 based on Brisbane core is not fast enough because of itsslower cache. By the way, we observe similar situation in regular gaming benchmarks, too, although thistendency stands out most vividly in Valve Source particle benchmark.

    We have also included a chess benchmark on Fritz 9 engine into our gaming test session.

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    20/38

    All in all, the situation here is hardly any different from what we have seen in 3D games. Core 2 Duo processorfamily ensures higher performance.

    Office Applications

    Unfortunately, complex benchmarks modeling the user behavior, such as SYSMark 2004 SE, do not work inWindows Vista. Therefore, we decided to estimate the typical office performance using the time it takes our testsystems to resolve certain specific tasks. These tasks were the comparison of two versions of a large document in

    Microsoft Office Word 2007 and the calculation of a table with typical statistical operations in Microsoft OfficeExcel 2007.

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    21/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    22/38

    Here Core 2 Duo processors perform phenomenally fast. Athlon 64 X2 can only compete against Intel PentiumD here, which by the way cost much less.

    We have also tested the systems performance with the benchmark built into the 64-bit version of the 7-ziparchiving tool. We looked at the archiving speed and data extraction from archives.

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    23/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    24/38

    Two charts and two totally different pictures. Archiving is done best on Intel processors, while extraction of filesfrom the archive is performed faster on AMD CPUs.

    Audio and Video Encoding

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    25/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    26/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    27/38

    During audio and video encoding with major codecs CPUs on Core micro-architecture prove fast enough tooutpace K8 based CPUs from AMD price similarly. Note that even Pentium D processors on NetBurst micro-architecture can boast pretty decent performance level (in relation to their price point). As we remember, thismicro-architecture was optimized specifically for work with streaming data.

    Image Editing

    Besides our standard Adobe Photoshop CS3 test session we have also checked the performance of our testing

    participants in Microsoft Windows Photo Gallery that was used to prepare a digital photo gallery for printing.

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    28/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    29/38

    The performance results during image editing are not surprising. Core micro-architecture allows Core 2 DuoCPUs to easily outperform Athlon 64 X2, which can only compete with Intel Pentium D in this test.

    Nonlinear Video Editing

    We tested the performance during video movie rendering in Adobe Premiere Pro 2.0 and during the preparationof the video movie for posting on YouTube using a standard Windows Vista application aka Windows MovieMaker.

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    30/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    31/38

    I have to say that the performance of inexpensive Core 2 Duo and Athlon 64 X2 processors doesnt differ asmuch in video editing applications as we have just seen in image editing tasks. However, Intel processors again

    prove faster than the AMD solutions priced identically.

    Final Rendering

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    32/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    33/38

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    34/38

    Final rendering is one of the few tasks where the performance of Athlon 64 X2 processors corresponds to theirprice. In other words, this is the type of tasks where Athlon 64 X2 will be able to compete successfully againstCore 2 Duo and Pentium D even after the upcoming April price drop.

    Core 2 Duo E6420, E6320 and E4400 Overclocking

    Younger processor models are especially attractive for overclockers who strive to put together ahigh-performance system at minimal possible cost. Therefore, we couldnt leave out the overclocking tests of the

    new processors.

    For our experiment we put together a special testbed with a different mainboard. ASUS P5N-E SLIthat provednot very efficient during our previous overclocking attempts was replaced with a more predictable ASUS P5BDeluxe, which is free from FSB Hole defect and is guaranteed to allow raising the FSB speed above 500MHz,which is very important for successful overclocking of younger Core 2 Duo processors. Our overclockingexperiments were performed using Zalman CNPS9700 LED air-cooler. Also we clocked Corsair DominatorTWIN2X2048-8888C4DF memory synchronously with the FSB to avoid additional problems. The stability ofthe overclocked processor was tested with the ORTHOS utility that proved very reliable during our previoustests.

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    35/38

    The first one to get into our testbed was Core 2 Duo E6420 CPU. This processor is remarkable for its 8xmultiplier that makes it almost ideal for overclocking. With this multiplier you can overclock it practically to itsmaximum in most mainboards, because you will hardly need to push FSB beyond 500MHz in this case.

    First of all we decided to find out what frequency our test Core 2 Duo E6420 will support without any Vcoreadjustment. This processor runs at the nominal Vcore of 1.325V, which is the maximum possible setting byspecifications.

    In this case the maximum FSB frequency when the CPU remained stable equaled 370MHz. In other words, thisCPU can run stably at 2.96GHz without raising its Vcore, which is a 40% increase above the nominal speed.

    The second experiment was performed with the same CPU working at 1.6V Vcore. Core 2 Duo processors areknown to be very tolerant to processor voltage increase, which leads to significant improvement of theiroverclocking potential. And this time, it was no exception: we could push the FSB frequency to 465MHz.

    As a result, the overclocked CPU was working stably at 3.72GHz, which is 75% higher than the nominal. This isvery good and impressive result, which makes Core 2 Duo E6420 a very attractive choice for overclockers.

    We can conclude that Core 2 Duo E6420 processors can overclock as good as their elder brothers. And this isactually not surprising at all, because with the Core 2 Duo E6x20 the youngest CPUs on Core micro-architectureacquired fully-fledged Conroe core.

    The second testing participant to undergo some overclocking was Core 2 Duo E6320 with the nominal multiplierof 7x. Note that this processor is harder to overclock because of the low clock frequency multiplier, which wehave already mentioned before during our Core 2 Duo E6300 overclocking experiments. In order to achievemaximum frequencies for the Conroe core, the mainboard may need to allow raising FSB frequency beyond500MHz, which far not every mainboard can do, even from a leading mainboard maker. Moreover, you are morelikely to hit against the FSB Wall the maximum FSB frequency for the given processor that is not connected

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    36/38

    with its clock speed.

    As for the practical results, our Core 2 Duo E6320 with the nominal Vcore of 1.3125V could speed up to3.29GHz without any core voltage increase. It is equal to 470MHz FSB.

    Strange as it might seem, but raising the Vcore didnt have any effect. Even when we hit 1.6V on the CPU (andthe corresponding higher voltage on the chipset North Bridge, which is necessary for higher FSB speeds), ourCore 2 Duo E6320 didnt get not even 1MHz faster.

    It looked like we hit against the so-called FSB Wall effect, which wouldnt let us overclock the CPU beyond3.29GHz. So, the overclocking potential of this model turned out not so exciting as those of the previous CPU.However, it would be unfair to draw any conclusions basing only on the results of one single processoroverclocking. Moreover, FSB Wall is a very individual parameter, which may differ dramatically by differentCPU samples. So, what we can claim is that Core 2 Duo E6420 has higher chances to overclock to themaximum, although Core 2 Duo E6320 can also be pretty efficient and has every chance to hit the maximum ifthe conditions allow.

    As for the overclocking results for our third CPU, Core 2 Duo E4400, we didnt pin a lot of expectations on itfrom the very beginning. This CPU is based on the cut-down Allendale core, which is known to be not asimpressive as Conroe when it comes to overclocking. Even the relatively high clock frequency multiplier of 10xdoesnt help here (it is so high because the CPU supports 800MHz bus). If you remember, we managed tooverclock a similar Core 2 Duo E4300 processor only to 3.42GHz. So, we expected the newcomer to perform atabout the same level.

    However, the actual results turned out even worse than we had expected. At the nominal Vcore of 1.2875V, thisCPU could only run at 2.93GHz, i.e. at 293MHz FSB taking into account the 10x multiplier. By raising the Vcoreto 1.6V, we could only push FSB frequency to 325MHz, not any higher than that.

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    37/38

    So, the CPU could run stably at 3.25GHz maximum, which doesnt look very impressive against the backgroundof the previous two models. However, Core 2 Duo E4400 will hardly disappoint you with its performance evenat this speed. Therefore, it is definitely far from failure.

    However, as you can see from the results of all three younger Core 2 Duo models, the overclocking success ismore of a lottery. There are too many factors that can negatively affect the final result. Here I have to mentionFSB Wall effect, limited mainboard abilities when it comes to FSB frequency increase, and even insufficientfrequency potential of the younger processor models, because they are often built on the cores that didnt getselected for more expensive CPUs. In other words, the overclocking result on Core 2 Duo E6420, E6320 orE4400 will have a lot to do with simple luck.

    Conclusion

    Unfortunately, we have no information about the upcoming AMDs pricing policy; therefore we cannot makeany definite conclusions about the inexpensive dual-core processors from the price-to-performance stand point.However, there are a few things we are pretty certain of.

    After the upcoming price drop and the launch of the new processor models, Core 2 Duo processors will finallybe able to fit into relatively inexpensive computer systems. Since Core based processors used to sell for at least$200, they couldnt enjoy vast popularity among budget and low-end mainstream users. However, this situationwill very soon change. Especially, since their performance level will allow them to compete successfully againstother inexpensive dual-core processors, such as AMD Athlon 64 X2 and Intel Pentium D.

    There is the whole bunch of popular applications, such as office tasks, games, image editing and media contentprocessing, where Core 2 Duo processors perform faster than Athlon 64 X2 from the same price range.

    In this case, AMD will either have to put up with the loss of the mainstream price segment, or correct the pricelist accordingly. However, relatively high production cost of the 90nm dual-core processors on K8 micro-

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e

    38 8/22/2012

  • 8/14/2019 CPU dual core2006.pdf

    38/38

    architecture as well as too slow adoption of the 65nm manufacturing process may hinder this correction.

    The obtained results suggest that AMD Athlon 64 X2 5000+ should cost no more than $180, and Athlon 64 X24800+ and 4600+ - no more than $140, if they want to successfully compete with the new Intel CPUs. However,we doubt that a price cut like that will ever be possible.

    However, inexpensive Core 2 Duo processors may face another problem on the way to the market: the absenceof adequately priced feature-rich mainboards for them. Hopefully, the mainboard makers resolve this issue

    ASAP, or the effect from this great Intel offer will be completely ruined.

    In conclusion I would like to say that Core 2 Duo processors have another very important advantage, which willbe valued by hardware enthusiasts. All these CPUs feature pretty good overclocking potential. Even the slowestmodels can run at 3GHz speed with very little effort on the users part, which pushes their performance to thelevel of Core 2 Extreme and hence cannot be outpaced even by overclocked Athlon 64 X2 based platforms. Thisshould definitely become another reason for the growing popularity of the Core 2 Duo family after the upcoming

    price reduction and the launch of the new CPUs.

    labs - Print version http://www.xbitlabs.com/articles/cpu/print/core2duo-e