Analysis of "Kourou" AMD k8l platform with quad-core smashing

Source: Internet
Author: User
Tags range

K8 's success gave Intel blow, and the chip giant saw the grim picture. After the core processor was released, Intel recovered part of the "lost ground" and regained its crown of performance, and faced Intel's aggressive offensive, how could amd cope with no "big bang" in the past two years? Please look at the "hammer" that hit "Kourou".

A whole year late.

July 27, 2006, Intel officially released the powerful Core 2 Duo processor, based on the two Conroe core and cores architecture of the new Core 2 Duo, its performance is far more than the current mainstream "pentium-d" performance of more than twice times, But also in the next generation of CPU competition will rival AMD far away in the rear. AMD seems to be too immersed in recent years K8 architecture in the great success of the market, in the core architecture for nearly half a year after the introduction of the new products have not been able to release the confrontation, it is clear that the K8 architecture has long been unable to compete with the core architecture processor for the task of the market. However, AMD finally revealed that the 2007 summer will be the full sale of the new K8L architecture processor, time is just the core architecture released a year later.

Leading K8 an era of the original design of the k8l is four core, its four independent core will be on the same wafer with the latest 65nm manufacturing process production. In addition to adopting the State-of-the-art 65nm manufacturing process, k8l will also introduce the two newest cutting-edge circuit technologies, the IBM patented embedded Silcon Germanium technology and stress memorization technology, This will be very effective to reduce the K8L architecture processor's electronic migration phenomenon and leakage rate and other negative effects.

Double the memory performance

The L1 caching scheme in the K8 architecture is a combination of 1 64KB indicating cache + 1 64KB data caches, each with a concurrent 2-channel/cycle transmission capability. In k8l, AMD promotes its L1 cache transfer capability to concurrent 4-way/cycle, L1 cache and the kernel will have 2 bidirectional 128-bit data channel. The improved cache module has a very low latency and a hit rate of up to 95%. The high speed transfer capability of L1 cache beyond the K8 architecture allows the L1 cache capacity of the k8l architecture to be reduced to 1 32KB indication cache + 1 32KB data cache, thereby increasing its hit/turnover rate. The K8L architecture processor has a growth of more than 80% L1 cache system performance over the K8 architecture processor, while AMD is also considering keeping the L1 cache at 1 64KB indication cache + 1 64KB data cache.

Since the K8 architecture, AMD has integrated memory controllers inside the CPU. This allows the CPU to bypass the L2 cache directly with the L1 cache and memory, and the L2 cache stores other scarce data that is not in the L1 cache. This greatly accelerates the memory performance and core data processing speed of the K8 architecture. The k8l architecture still uses a separate caching scheme for the K8 architecture and a built-in processor memory controller scheme: The L1 cache, L2 cache, and L3 cache eliminate the prefetch and buffering relationships between the data and the memory and the core directly.

The most prominent feature of the k8l architecture, compared to all previous CPU architectures, is that its 4 core processors will have a shared L3 cache. The L3 cache is also born into a wafer with 4 cores, and its capacity is the smallest 2MB takeoff. As with the L2 cache, the L3 cache is independent, and L1 cached data and L3 cached data are not duplicated.

k8l Frame composition

Just as the core architecture shared L2 Cache Advanced Smart Cachecore technology makes its L2 cache performance double the pentium-d cache performance of dual-core L2 processors, AMD's adoption of the more advanced four core shared L3 caching technology will enable the 4 cores of the K8L architecture processor to obtain very fast data exchange capabilities, while more intelligent and dynamic division L3 caching. In addition to the single core load and the other core is more idle, the L3 cache will be able to more quickly than the core architecture of the intelligent sharing L2 cache for the load of a larger kernel to provide faster cache space. The L3 cache's performance boost for K8L architecture processors is enormous, and a new L3 cache design will make k8l architecture processors more than twice times faster than the K8 architecture processor, and will be more than 40% higher than the core architecture, making a significant contribution to the overall performance of the CPU.

On the other hand, the memory system of the K8L architecture will support DDR3 memory and FB-DIMM standards, and will also be upgraded to Hyper-transport 3, but either DDR3, Fb-dimm, or Hyper-transport 3, This is a huge increase in k8l performance. Unlike the graphics card's strong GDDR3 video memory, the performance improvement of DDR3 with K8L architecture processors is even more pronounced.

Battle 2007

AMD has further announced the launch time plan for the K8L architecture processor, where the four core k8l architecture processor code-named Barcelona (Barcelona), which was first sold in summer 2007, supports Hyper-transport 2.0, stepping into Rev H Core, Power consumption will range from 68W to 120W, and the introduction of the quad core and L3 cache does not make the K8L architecture processor a significant 150W-by-side power dissipation, which is in stark contrast to the NetBurst architecture that has been frustrated by Intel's power-control problems. Code Budapest (Budapest), the perfect version of the k8l core will be listed in late 2007, support the updated Hyper-transport 3.0, the larger L3 cache capacity, its model is Opteron 12XX series and Athlon 64 full range In the spring of 2008, the strongest server version of Shanghai (Shanghai) Opteron 22X and 82XX series will become AMD's strongest processor, AMD will introduce more new circuit technology to control its power consumption and so on.

The progress of the k8l architecture relative to the K8 architecture is enormous, and the introduction of a variety of the latest technologies in the K8L architecture has enabled DIY users to be confident that they can replace the current four-core processor that is currently unreachable. At the same time AMD and Intel's contention will be more intense, while the vast number of DIY users are enjoying the fierce fighting and easy access to cheap and powerful processors.

Contact Us

The content source of this page is from Internet, which doesn't represent Alibaba Cloud's opinion; products and services mentioned on that page don't have any relationship with Alibaba Cloud. If the content of the page makes you feel confusing, please write us an email, we will handle the problem within 5 days after receiving your email.

If you find any instances of plagiarism from the community, please send an email to: info-contact@alibabacloud.com and provide relevant evidence. A staff member will contact you within 5 working days.

A Free Trial That Lets You Build Big!

Start building with 50+ products and up to 12 months usage for Elastic Compute Service

  • Sales Support

    1 on 1 presale consultation

  • After-Sales Support

    24/7 Technical Support 6 Free Tickets per Quarter Faster Response

  • Alibaba Cloud offers highly flexible support services tailored to meet your exact needs.