1.Hypertransport (super transmission):
Hypertransport was originally AMD proposed a bus technology proposed in 1999. With the release and promotion of the AMD64 platform, Hypertransport applications are increasingly widely known.
Hypertransport is an end-to-end bus technology designed for integrated circuits on the motherboard, which can provide higher data transmission bandwidth between memory controllers, disk controllers, and PCI bus controllers. HyperTransport is a high-speed sequence connection function that simulates two independent data links in the same parallel bus. Under 400 MHz, the bidirectional 4BIT mode is 0.8Gb / sec, the bus bandwidth of the two-way 8bit mode is 1.6Gb / sec; 800MHz, the bidirectional 8Bit mode bus bandwidth is 3.2Gb / sec, the bidirectional 16bit mode bus bandwidth is 6.4Gb / sec, the bus bandwidth of the two-way 32Bit mode is 12.8Gb / sec.
Hypertransport has a major feature, which is that when the data bit is not 32 bit, it can transmit data batch to achieve the same effect as 32 bit. For example, 16bit data can be divided into two batches. The 8bit data can be divided into four batches. This data subcontract transmission is given, giving Hypertransport's greater elastic space.
In February 2004, Hypertransport Technology Alliance officially released Hypertransport 2.0 specification, due to the DUAL-DATA technology, the frequency successfully increased to 1.0GHz, 1.2GHz and 1.4GHz, data transmission bandwidth by each channel 1.6GB / sec increased to 2.0 Gb / sec, 2.4Gb / sec and 2.8GB / sec, the maximum bandwidth was raised from the original 12.8Gb / sec to 22.4Gb / sec.
When Hypertransport is applied to the memory controller, it is actually similar to the traditional front-end bus (FSB, Front Side Bus), so the frequency of Hypertransport is equivalent to the front end for the CPU of the Hypertransport technology for the memory controller. The frequency of the bus.
Original: http://www.donews.net/aidem/archive/2005/03/14/301499.aspx
2. Hardware-level virus protection
In the new 939-pin Athlon FX / Athlon 64 processor and the 940-pin Operton processor to be released in the future, AMD adds an important hardware-level virus protection. This feature is Intel, AMD and Microsoft in order to cope. Early virus software is particularly developed. We know that the buffer overflows the primary means of virus attacks, but the hate is that this means is widely known, but it is better to solve the problem. Forcing Microsoft can only get a patch, with this technology. All this will be fully improved. AMD is called "Execution Protection" (Intel's called method), which is a locking mechanism integrated within the CPU. When the operating system activates this function, the data of the CPU buffer will read only, and Can't be executed, which can effectively prevent the virus buffer overflow attack. The current CPU does not do this, can only be ravaged by the virus. After activating this feature, the intentions spending data security for data security will be greatly reduced, and the virus library update will no longer be so frequent, and the updated more frequent Windows patches, large parts are no longer necessary. The habit of operating the operating system rely on "eat patch" is also changed. Windows XP's Service Pack 2 supports and implements viral protection in the AMD 64 processor. How big is the specific role of this hardware anti-virus function, our evaluation room is planning a test. The original text is taken from: http://youth.scnu.edu.cn/union/Article_show.asp? ArticleId = 409
3. Interface type:
We know that the CPU needs to work with the motherboard through an interface. After so many years of development, the CPU has an interface, card, contact, pin, etc. At present, the current CPU interface is a pin interface, and there is a corresponding slot type corresponding to the motherboard. Different CPU interface types, in the number of jacks, volume, and shapes, so they cannot be connected to each other.
Socket 775 Socket 775, also known as socket t, is currently applied to the interface corresponding to the CPU packaged in Intel LGA775, which is currently using the Pentium 4, Pentium 4 EE, Celeron D, etc. of the LGA775 package, and CPU. Unlike the previous Socket 478 interface CPU, there is no traditional pins at the bottom of the Socket 775 interface CPU, which is 775 contacts, ie, non-needles but contact, through 775 in the corresponding socket 775 slot. The root stylus contacts to transmit the signal. The Socket 775 interface not only effectively enhances the signal strength of the processor, improves the processor frequency, but also improves the productivity of processor production, and reduces production costs. As the Socket 478 gradually faded, Socket 775 will become the standard interface of all Intel desktop CPUs in the future.
Socket 754 Socket 754 was the CPU interface initially released in September 2003, which currently uses this interface with low-end Athlon 64 and high-end Sempron with 754 CPU pins. With the popularity of Socket 939, Socket 754 will eventually fade out.
Socket 939 Socket 939 is the 64-bit desktop platform interface standard in June 2004 in June 2004, which is currently using high-end Athlon 64 and Athlon 64 FX, with 939 CPU pins. Socket 939 processor and the past Socket 940 slot cannot be mixed, but Socket 939 still uses the same CPU fan system mode, so the fan used for Socket 940 and Socket 754 can also be used in Socket 939. .
Socket 940 Socket 940 is the earliest AMD64-bit interface standard with 940 CPU pins, which currently adopt the server / workstation used by this interface and the original Athlon 64 FX. With the new Athlon 64 FX, Socket 940 will become an Opteron's dedicated interface with the SOCKET 939 interface. The use of Socket 603 Socket 603 is relatively professional, applied to the high-end server / workstation platform in Intel, using this interface of the CPU is Xeon MP and early Xeon with 603 CPU pin. The CPU of the Socket 603 interface can be compatible with the socket 604 slot.
Socket 604 is similar to socket 603, and Socket 604 is still a servers / workstation platform applied to Intel aspects, using this interface of the CPU of 533MHz and 800 MHz FSB Xeon. The CPU of the Socket 604 interface cannot be compatible with the Socket 603 slot.
The Socket 478 Socket 478 interface is the interface type used in the Pentium 4 Series processor, and the number of pins is 478 pins. The Pentium 4 processor area of Socket 478 is small, and the pin is arranged closely. This interface is adopted by Intel's Pentium 4 Series and P4 Celeron Series.
Socket A Socket A interface, also called Socket 462, is the socket interface of AMD Athlon XP and DURON processors. The Socket A interface has 462 inserts, which can support 133MHz outgoing.
Socket 423 Socket 423 slot is the standard interface of the initial Pentium 4 processor, and the shape of the socket 423 and the slot of the first few Socket classes are similar, the corresponding CPU pin is 423. The Socket 423 slot is mostly based on the Intel 850 chipset motherboard, which supports the Pentium 4 processor of 1.3GHz ~ 1.8GHz. However, with the popularity of DDR memory, Intel has developed an I845 chipset that supports SDRAM and DDR memory, and the CPU slot is also changed to Socket 478. The Socket 423 interface is also hidden.
The Socket 370 Socket 370 architecture is Intel's development instead of the Slot architecture, and the appearance is very similar to the socket 7, and also uses a zero-plug-in slot. The corresponding CPU is 370 pin. Intel's famous "copper ore" and "Tu Latin" series CPU use this interface.
Slot 1 Slot 1 is the CPU interface developed by Intel's replacement Socket 7, and the patents apply. Such other vendors cannot produce the product of the Slot 1 interface. The CPU of the slot1 interface is no longer a familiarity, but it becomes a flat rectangular square, and the interface has also become a gold finger, which is no longer a pin form.
Slot 1 is Intel's slot designed for Pentium II Series CPU, which makes Pentium II CPU and its associated control circuit, and the second-level cache is on a piece of card, most Slot 1 motherboard uses 100MHz Overseas. The technical structure of Slot 1 is more advanced, providing greater internal transmission bandwidth and CPU performance. This interface has been eliminated, and there is no product for such interfaces on the market.
Slot 2 SLOT 2 is used in a professional, all systems in high-end servers and graphics workstations. The CPU used is also a very expensive Xeon series. Slot 2 has many different than Slot 1. First, the slot 2 slot is longer, and the CPU itself has to be large. Second, SLOT 2 is competent for higher demand multi-purpose calculations, which is the key to entering high-end enterprises to calculate the market. In the standard server design, general vendors can only use two Pentium II processors in the system, and after Slot 2 is designed, 8 processors can be used simultaneously in one server. And Pentium II CPUs with Slot 2 interface have adopted the most advanced 0.25 micron manufacturing process at the time. The motherboard chipset supporting the Slot 2 interface has 440Gx and 450nx. The Slot A Slot A interface is similar to Intel's SLOT 1 interface for AMD's K7 Athlon. In terms of technology and performance, the Slot A motherboard can be fully compatible with the original peripheral expansion card devices. It is not an Intel's P6 GTL bus protocol, but the ALPHA bus protocol EV6 of Digital. The EV6 architecture is a more advanced architecture that uses multi-threaded points to point topology, supports 200MHz bus frequencies.
Original: http://publish.it168.com/cword/1179.shtml
4. Dual channel:
Dual Channel Memory Technology is actually a memory control and management technology that relies on the memory controller of the chipset, which can double the bandwidth provided by the two equivalent specifications memory. It is not a new technology, which was used in the server and workstation system, just to solve the increasingly embarrassing memory bandwidth bottleneck problem in the desktop, it walked to the front desk of the desktop motherboard technology. In a few years ago, Intel has introduced an I820 chipset that supports dual-channel memory transmission technology. It has a pair of gold partners with RDRAM memory, which makes it a maximum highlight of the market, but production costs Excessive defects have caused the situation that is good, and finally eliminated by the market. Since Intel has given up support for RDRAM, the dual-channel memory technology at the mainstream chipset is a dual channel DDR memory technology. The mainstream dual-channel memory platform Intel is the Intel 865/875 series, while AMD is NVIDIA NFORCE2. series.
Dual Channel Memory Technology is a low price, high performance solution to solve the contradiction between CPU bus bandwidth and memory bandwidth. Now the CPU's FSB (front-end bus frequency) is getting higher and higher, and Intel Pentium 4 has much higher demand for memory bands than AMD Athlon XP. The Intel Pentium 4 processor and the North Bridge chip data transmission use QDR (Quad Data Rate, four data transmission) technology, and its FSB is 4 times that of the outer frequency. The Intel Pentium 4's FSB is 400/533 / 800MHz, respectively, and the bus bandwidth is 3.2Gb / sec, 2.7GB / sec and 6.4GB / sec, and the memory bandwidth provided by DDR 266 / DDR 333 / DDR 400 is 2.1. GB / Sec, 2.7GB / sec and 3.2Gb / sec. In single-channel memory mode, DDR memory cannot provide the data bandwidth required for the CPU to become a system's performance bottleneck. In dual-channel memory mode, the memory bandwidth of the dual channel DDR 266 / DDR 333 / DDR 400 is 4.2GB / sec, 5.4GB / sec and 6.4GB / sec, here you can see, dual channel DDR 400 memory is just a bandwidth requirement of 800 MHz FSB Pentium 4 processor. For the AMD ATHLON XP platform, the data transmission technology of the processor and the North Bridge chip uses DDR (Double Data Rate, Double Data Transport) technology, and FSB is twice as far as it is, and its demand for memory band is far lower. In Intel Pentium 4 platform, FSB is 266/333 / 400MHz, respectively, and the bus bandwidth is 2.2Gb / sec, 2.7GB / sec and 3.2Gb / sec, which can be satisfied with single-channel DDR 266 / DDR 333 / DDR 400 Its bandwidth demand, so uses dual channel DDR memory technology on the AMD K7 platform, it can be said that it is not much efficient, and performance is not as obvious as the Intel platform, and the performance impact is the integrated motherboard for integrated display chips. NVIDIA's NFORCE chipset is the first chipset that extends the DDR memory interface to 128-bit, and Intel uses this dual-channel DDR memory technology on its E7500 server motherboard chipset, Sis and Via have also In response, actively develop this technology that can make DDR memory bandwidth. However, due to various reasons, it is necessary to achieve this dual-channel DDR (128 bit parallel memory interface) for many chipset manufacturers. DDR SDRAM memory and RDRAM memory are completely different, and the latter has a high-delayed characteristic and is a serial transmission mode. These features determine the difficulty and cost of designing a two-channel RDRAM memory chipset. It is not too high. However, DDR SDRAM memory has its own limitations. It itself is a low delay characteristic. It is parallel transmission mode, and the most important point: When the DDR SDRAM operating frequency is higher than 400MHz, its signal waveform tends to distortion. Problem, these are unmatosed, which supports a chipset that supports dual-channel DDR memory systems, and the manufacturing cost of chipsets will also increase accordingly, and these factors have restrictions on this memory control technology.
The ordinary single-channel memory system has a 64-bit memory controller, and the dual-channel memory system has 2 64-bit memory controllers, which have 128bits in the dual channel mode, which is theoretically in theory. An increased double. Although the bandwidth provided by the double 64-bit memory system is equivalent to a bandwidth provided by a 128-bit memory system, the effect is different. The dual-channel system includes two independent, complementary intelligent memory controllers, in theory, both of the two memory controllers can operate simultaneously in the case of zero. For example, two memory controllers, one is A, another is B. When the controller B is ready for the next access memory, the controller A is read / written by the main memory, and vice versa. This complementary "nature" of the two memory controllers can make the waiting time reduce 50%. Two memory controllers in dual-channel DDR are functionally exactly the same, and the timing parameters of the two controllers can be set separately. Such flexibility allows the user to use two different constructs, capacity, speed DIMM memory strips. At this time, the dual-channel DDR simply adjusts to the lowest memory standard to achieve 128bit bandwidth, allowing DIMM memory strips for different density / waiting time characteristics. Reliable work together. Supported dual-channel DDR memory technology desktop chip group, Intel platform has Intel's 865P / 865G / 865GV / 865PE / 875P and after 915/925 series; VIA's PT880, ATI's Radeon 9100 IGP series, SIS 655 SIS 655FX and SIS 655TX; AMD platforms have a VIA KT880, NVIDIA NFORCE2 Ultra 400, NForce2 IGP, NFORCE2 SPP, and later chips. Original: http://publish.it168.com/cword/989.shtml
