x86-64 architecture

After AMD’s last revolutionary processor, the x86 architecture have not encountered any major changes at all – only minor ones, such as when Intel introduced the Pentium MMX that uses the x86 architecture. After such a long wait however, AMD might bring again the hype and a new achitectural revolution. [no_toc]

AMD has now decided to improve the x86 architecture to increase its flat memory in 64 bit mode. By increasing the flat memory for 64 bit, they are also efficiently increasing its performance because of the added address space and increased number of general-purpose registers. However, many average computer users may still opt to stay with 32 bit mode because it has been the staple for many years. Still, 64 bit mode has many performance benefits over 32 bit mode. Let’s take a look through some of its benefits below.

opterons2
opterons

In 32 bit mode, the RAM is limited to 4GB. 4 GB of RAM is not enough for users of CAD, designers, and even simulators will experience performance decrease with the limited RAM. Intel introduced its Xeon processors that could support 36 bit addressing mode albeit with a performance decrease by about 30 %. With a performance decrease like that, the 64 bit mode is more convenient due to the fact that it will not reduce performance as well as the added address space.


lme-modes
Huge data bases users. Especially companies that have a huge data base, a 64bit address space is much more advantageous because it efficiently increases the speed and ease of programming compared to the 32bit address space.


hammer-core
Users with huge data bases such as large companies require a much more efficient tool for its data processing. A 64 bit is recommended in this situation because it efficiently increases speed and ease of programming compared to the 32 bit address space.

In calculating long scientific equations, speed and memory is a must for accurate calculations. 64 bit has all that without any performance decrease.


execution-unit
Because of the increased speed and performance, many complex combinations for security ensuring applications are now possible even in cryptography.

multiplication delay

Many users will opt to stay using a 32 bit address space because of the fear that the 64 bit mode will be expensive and many current 32 bit applications will still run in 32 bit mode even in a 64 bit address space because they are coded as such. According to AMD however the processor needed for the 64 bit will only have a 2-3 % increase in its transistors, effectively having a good price to performance ratio.We all know though, that increasing the transistor count will also lead to the die’s area to be increased thus, affecting cost and compatibility for a bit.
additional-regidters
With plenty of benefits, the new IA64(x86-64) architecture will surely replace the old x86 architecture. Still, the users and installed data base of the x86 architecture will still opt to stay with the x86 because many developers will probably have doubts into spending money for the development of applications and programs for the IA64.

core frequency

Chipset

A chipset is probably the most awaited part of a processor because with this, we will know the advantages and disadvantages of the processor, as well as its compatibility. Knowing the chipset also allows us to have a sneak peek in the processor’s functions as well as the price of it.


memory-speed
AMD has taken a new approach for their chipset. We all know that a chipset consists of a memory controller, AGP and bridges called the Northbridge and Southbridge. The Northbridge connects the processor with the RAM slots, PCI-E slot, and the Southbridge. Meanwhile, the Southbridge (also known as I/O controller HUB) connects all secondary peripherals to the Northbridge. These two chips are both connected by a bus. Well, what type of bus will AMD have to install? It is just logical that they will choose the newly announced Hyper Transport Bus. It has the greatest speed and a less sophisticated layout compared to other buses.

core frequency

So what do you think AMD has probably chosen as their bus? Of course the Hyper Transport Bus!


dram-latency-1
They are left with no other better choice than this. However, AMD is still not finished in implementing new technologies in their chip.

chipset delays

Then what about the chips functions?

processor improvement

There are five, the memory controller, AGP, I/O controllers, and a PCI controller.


io-latency
The processor already has the memory controller as well as the AGP, but what about other parts? An infamous NVIDIA approach is just integrating all those parts into one single chip but AMD decided otherwise. AMD opted to create three different chips under the AMD8000 series.


scalability
These three chips will have different parts installed into them, thus, users will not have to pay for the functions they don’t need, they will just buy a specific chip according to their needed function.


hyper-transport
Days are gone when you purchase an expensive chip with a PCI-X bus controller that is rarely used for home.


tpi
AMD’s technical support team will also have an easier time solving technical problems because each chip have an individual function and parts that can be classified easier compared to different chipsets with many so called functions and parts that finding a technical solution for them will be hard because there are many set of parts to set their eyes upon.

Because of this new architecture, chipsets from other manufacturers will be almost rendered inferior compared to AMD’s. However, the CrushK9 from NVIDIA has made a good solution for a single chip. It will have integrated video cards that users can rely instead of buying dedicated video cards. With this, cheaper pc build for offices will be made without sacrificing the possibility of gaming. With NVIDIA’s reputation of having good gaming performance, users can rely on this for a more cheaper gaming experience.