Foreword

Intel’s portfolio knows how to impress for some time. After the introduction of the first 45-nm-high-end processors, led by the flagship Intel Core advance now piecewise 2 Extreme QX9650 last October, the processor giant in the less expensive regions. The beginning of the new core 2 makes this duo E7200, which is to complete the difficult available 8000 series down. With this processor, which will cost about 100 euro, the 45 nm production really comes into the mainstream space. The core 2 Duo of the 8000 series with an average price of 170 euros a piece still was too expensive, many potential customers so that they now perhaps could have found the first real alternative with the E7200. We will go the new CPU in our review on the ground and thereby clarify whether the performance for the price fits, as the power consumption goes down, and what you can do else with the supposedly small processor.

It is logical that Intel had to make cuts in the processor, which is like the great Wolfdale family,. They concern specifically the halving of the L2 cache 6 MB to 3 primarily but also the control of the frontside bus will otherwise. This runs all new Yorkfield Processor 9000 series with four cores and also the Wolfdale processors of the core 2 Duo E8x00 (based on two cores) physically 333,33 MHz (effective after Intel’s quad-pumped bus 1,333 MHz), the first small representative E7200 the well-known 266,66 MHz running. The resulting FSB 1,066 MHz was already in the popular processors core 2 Quad Q6600 and the older E6x00 series used. Achieved by a “crooked” multiplier of 9.5 to the core 2 Duo E 7200 thus 2.53 GHz. With time the official successor of the E6000 have E4000 series – in the E8000 enters its successor already found – from now on the battlefield.

Overview

As usual we take a look at the new processor with CPU-Z. Previously we had to update but the BIOS at our gigabyte motherboard, because we reached the maximum, but normal clock frequency not without a setting of the half multiplier. After we our BIOS of gigabyte X 38 motherboards so were updated with the latest version of “F6″, a new setting was added, which in addition can specify also half to the whole multipliers. From now on, our core 2 ran duo E7200 the 2.53 GHz, which he is to make according to specification. A little mystery, however, was the set voltage. While the BIOS reported us that runs the processor 1,2125 Volt, always only 1.18 volts are displayed under Windows. In the power save mode, this is reduced then to 1.12 volts; like all Intel processors, the multiplier is then typically “6″.

The L2 cache was halved compared to a conventional Wolfdale processor from 6 to 3 MB. The missing 3 MB are not present on the chip, so that a software is excluded. The following image shows a Wolfdale core with 6 MB L2 cache. The orange area which almost the entire left half of the processor, is the L2 cache. Expressed in figures a Wolfdale has a total of 410 million transistors; approximately 272 million of which are required for the cache. It is organized into 24 ways, as you can see on the thick vertical lines.

The Wolfdale Penryn with only 3 MB is a new processor design, which suggests also the used M0 stepping. In the simplest sense, the traditional Penryn processor was cut off at the end. This helps that “Penryn 3 MB” has a total area of only 81 instead of 107 mm² and reduced its number of transistors towards a “Penryn 6 MB” of 410 on 274 million. Instead of the previous 24 routes, the associativity of the L2 cache was reduced to 12 ways.

With the image, also the last doubts should be removed. The “Penryn 3 MB” or his “Wolfdale, 3 MB” desktop version is also in the core 2 Quad Q9300 used, while all other quad-core processors of the 9000 series from two “Penryn 6 MB” together are set. Fairy tales in forums around the reactivating of the L2 cache should these facts definitively the past belong – because the processor in the M0 stepping comes along, then quite simply no cache to reactivate exists.

Test system

• Processor
  • Intel
    • Quad-core
      • Intel Core 2 extreme QX9770 – 3.20 GHz, 12 MB L2 cache, FSB1600 (Yorkfield C0)
      • Intel Core 2 Extreme QX9650 – 3.00 GHz, 12 MB L2 cache, FSB1333 (Yorkfield C0)
      • Intel Core 2 extreme QX6850 – 3.00 GHz, 8 MB L2 cache, FSB1333 (Kentsfield G0)
      • Intel Core 2 Quad Q9450 – 2. 66 GHz, 12 MB L2 cache, FSB1333 (Yorkfield C0)
      • Intel Core 2 Quad Q6700 – 2.66 GHz, 8 MB L2 cache, FSB800 (Kentsfield B3)
      • Intel Core 2 Quad Q6600 – 2.40 GHz, 8 MB L2 cache, FSB800 (Kentsfield G0)
    • Dual-core
      • Intel Core 2 Duo E7200 – 2.53 GHz, 3 MB L2 cache, FSB800 (Wolfdale M0)
      • Intel Core 2 Extreme X 6800-2.93 GHz, 4 MB L2 cache, FSB800 (Conroe B1)
      • Intel Core 2 Duo E6850 – 3.00 GHz, 4 MB L2 cache, FSB1333 (Conroe G0)
      • Intel Core 2 Duo E6750 – 2.66 GHz, 4 MB L2 cache, FSB1333 (Conroe G0)
      • Intel Core 2 Duo E6600 – 2.40 GHz, 4 MB L2 cache, FSB800 (Conroe B1)
      • Intel Core 2 Duo E6420 – 2.13 GHz, 4 MB L2 cache, FSB800 (Conroe B1)
  • AMD
    • Quad-core
      • AMD Phenom 9950 2.6 GHz, 2 MB L2 cache, 2 MB L3 cache, FSB200, HT link 2,000 MHz, DDR2 1.066 (Barcelona simulates B3)
      • AMD Phenom 9850 2.5 GHz, 2 MB L2 cache, 2 MB L3 cache, FSB200, HT link 2,000 MHz, DDR2 1.066 (Barcelona simulates B3)
      • AMD Phenom 9750 2.4 GHz, 2 MB L2 cache, 2 MB L3 cache, FSB200, HT link 1,800 MHz, DDR2 1.066 (Barcelona simulates B3)
      • AMD Phenom 9650 to 2.3 GHz, 2 MB L2 cache, 2 MB L3 cache, FSB200, HT link 1,800 MHz, DDR2 1.066 (Barcelona simulates B3)
      • AMD Phenom 9600 to 2.3 GHz, 2 MB L2 cache, 2 MB L3 cache, FSB200, HT link 1,800 MHz, DDR2 1,066 (Barcelona B2)
      • AMD Phenom 9550 2.2 GHz, 2 MB L2 cache, 2 MB L3 cache, FSB200, HT link 1,800 MHz, DDR2 1.066 (Barcelona simulates B3)
      • AMD Phenom 9500 2.2 GHz, 2 MB L2 cache, 2 MB L3 cache, FSB200, HT link 1,800 MHz, DDR2 1,066 (Barcelona B2)
      • AMD Phenom 9150e – 1.8 GHz, 2 MB L2 cache, 2 MB L3 cache, FSB200, HT link 1,600 MHz, DDR2 1.066 (Barcelona simulates B3)
      • AMD Phenom 9100e – 1.8 GHz, 2 MB L2 cache, 2 MB L3 cache, FSB200, HT link 1,600 MHz, DDR2 1,066 (Barcelona B2)
    • Dual-core
      • AMD Athlon 64 X 2 6000 + – 3.00 GHz, 2 MB L2 cache, FSB200, HT link 1 GHz, DDR2-750 (Windsor F3)
      • AMD Athlon 64 X 2 6000 + – 2.20 GHz, 2 MB L2 cache, FSB200, HT link 1 GHz, DDR2-733 (Windsor F3)
• Motherboard
  • AMD: Gigabyte MA790FX-DQ6 (790FX) – Rev. 1.0 BIOS: F3
  • Intel: Gigabyte X38T-DQ6 (X 38)-Rev. 1.0 BIOS: F3h, for E7200 BIOS F6 (needed for the sake multiplier)
• Memory
  • AMD
    • 2 x 1024 MB DDR2-1066 OCZ (5-5-5-15, dual-channel)
    • 2 x 1024 MB DDR2-1066 Aeneon (5-5-5-15, dual-channel)
    • 2 x 1024 MB DDR2-700, DDR-714, DDR2-733, DDR2-750, DDR2-766 Aeneon (4-4-4-18, dual-channel)
  • Intel
    • 2 x 1024 MB DDR3-1600 OCZ Platinum (CL7-8-8-24, dual-channel)
    • 2 x 1024 MB DDR3-1333 OCZ Platinum (CL7-7-7-20, dual-channel)
    • 2 x 1024 MB DDR3-1066 OCZ Platinum (cl6xxe-6-6-16, dual-channel)
• Graphics card
  • ATi Radeon HD 2900 XT (742 / 828), 512 MB
• Peripherals
  • Samsung HD501LJ (hard disk)
  • MSI DR8-A (burner)
• Power supply
  • Tagan TG480-U22
• Driver versions
  • ATi catalyst 7.10
  • Intel chipset driver 8.3.1.1009 (for X 38 chipset)
• Software
  • Microsoft Windows Vista Ultimate 32-bit, build 6000
  • Microsoft DirectX 9.0 c August 2007
  • Microsoft Direct3D 10

Explanations
To enable a possible fair and realistic comparison between the competitors, all in a closed MIDI tower with factory fan Assembly tests (a fan back sucking, blowing one on the hard drive cage in the front), to be attentive so also on thermal problems in the car. A “cooler master stacker RC-832″, which was made available to us by Caseking is used. The housing allows the use of up to nine 120-mm fans, of which the two used fans are included.

Benchmarks

All the benchmarks were carried out on Windows Vista in a resolution of 1280 x 1024 pixels. Especially in games can occur however slight variations in the resolution and video settings. What exactly this is mentioned on the spot in the text.

Many of the programs of our selected freely available, so that you can understand the tests on the local PC. Attached please find the exact version numbers or program variants that we have chosen for the test.

• Synthetic
  • SiSoft Sandra XIIc (version 2008) & XIIc SP1 for AMD Phenom
  • Super PI xmod 1.5 XS
  • Sciencemark 2.0 32 bit
  • Everest 4.20
  • PCMark05 1.2.0
  • PCMark Vantage 1.00
• System & rendering
  • 7-Zip 4.42
  • WinRAR 3.71
  • Cinebench 2003
  • Cinebench R10
  • Lightwave 8.5
• Multimedia
  • Nero Recode (Nero 8 suite version 8.1.1.0)
  • DivX 6.7
  • Tsunami MPEG video encoder Xpress 4.3.1.222
  • Apple iTunes 7.4.3.1
  • Lame 3.97 b
  • Lame 3 97a
  • Vorbis Oggdrop XP
• Games
  • 3DMark03 3.6.0
  • 3DMark05 1.3.0
  • 3DMark06 1.1.0
  • Fear MP demo
  • Quake 4 1.42
  • Anno 1701 demo
  • Company of Heroes 1.71
  • World in conflict
  • Crysis SP demo
• Other required tools
  • CPU-Z 1.41, 1.42, 1.43, 1.44
  • Prime95
  • Fraps

Synthetic tests

SiSoft Sandra XIIc

No matter whether it concerns mainboard, memory, hard drive, peripherals, cards, processor, network, interface BIOS, Windows or DirectX, SiSoft Sandra has extensive answers at hand. For much of the hardware in the PC, there is also a benchmark tests, the PC test can on its performance compared to some reference computers. All these values are however almost exclusively purely theoretical nature and have little relation to the practice, but processors in their theoretical capabilities compare well.

Download: SiSoft Sandra

Super PI

Super PI is a fairly simple and above all small software, with which the district number pi to more than one million jobs is calculated after the comma. The time required is measured and can be used for benchmarking of processors. We test the standard test “1 m”, which is practiced by many readers in our forum with a modified version 1.5 XS.

Download: Super PI

Sciencemark

In a further test of the bandwidth of the memory, the corresponding motherboard and the processor has to face the tool Sciencemark 2.0. This determines not only the bandwidth, but also the latency of memory. Still, the cache of the processor of a review may be subjected.

Everest

We round off the image of the storage assessment program Everest version 4.20. This tool has some built-in benchmark functions and tools that will completely deplete the computer and check for errors. We have used Everest beside temperature measurement to reveal the memory benchmark and the cache data throughput.

Download: Everest

PCMark05

The PCMark suite provides a detailed overview of the performance of the components such as processor, memory, graphics card and hard drive installed in the PC of the user since June 2005. To be carried out different individual tests, whose Einzelergebnisse are offset concluded as total value. Some tests are still highly topical, especially when it comes to synchronous work. Therefore, the supposedly dusty PCMark05 is reflected also in our new benchmark course.

Download: PCMark05

PCMark Vantage

Something over two years after the launch of PCMark05, a Futuremark benchmark for assessing the performance of a computer in a variety of application scenarios, the Finnish manufacturer presents the PCMark07, called “PCMark Vantage”. Once you want to be more private users and companies on the basis of a compact program able as possible objectively to evaluate the performance of a computer based on a broad basis of tests. Our article on PCMark Vantage provides all details on the new benchmark. The biggest hurdle in the system requirements of PCMark Vantage is the operating system, because the benchmark version of 2007 performed their service – a reason only on Windows Vista more, to take our new test system with Windows Vista of a thorough examination.

Download: PCMark Vantage