Top Spec

I’ve always been interested in Apples stand in their product architecture, when Apple decided to make the move to the x86 platform openning new leases to Linux based OS not to mention being able to run windows. Any way moving back to the topic!

Apple have bought the Chip manufacturer PA Semi. On behalf of Apple they current develop low-power processors for their PowerPC architecture. Apple have used the firm for previous products such as the PowerBook and Power Mac.

As you are all aware Apple have used Intel Processors for their current x86 architecture systems. Apple have decided to use Intels Atom microprocessors for their future IPhone release, the iPhone currently uses the ARM-base microprocessor. The ARM-base based iPhone owned by StrongARM have been integrated into Intels organisation so the move to the x86 architecture is appropriate.

The migration to the new architecture ‘x86′ is problematic. There are many advantages of moving to the architecture from a developers point of view, create once and run forever, using the Atom based processor! compare this to the ARM processor, it has been created by different manufacturers causing incompatible software between versions, no future/ backward compatibility, in conclusion causing entirely separate versions of applications and Operating Systems. The consumer must also note that the architecture will run on any desktop and laptop system, well the majority, x86 based.

With the Apple buy out of PA, Apple has now said in their future specification of products they will influence PA’s future product base.

Bookmark Article:

The University of Manchester (My Home City) have managed to carve transistors using the material Graphene one atom by 10 atoms in size.

Graphine is a one atom thick sheet which consists of a densely packed honeycomb crystal lattice. In basic terms under a microscope it is said to look like chicken wire! Simple.

By using this material they used a technique by chopping the honeycomb shaped material into quantun dots

“A quantum dot is a semiconductor whose excitons are confined in all three spatial dimensions. As a result, they have properties that are between those of bulk semiconductors and those of discrete molecules”

This gave a switchable conductivity which gave the graphene technique its transiting properties. By using this technique has Moore’s law been shattered? Will this process decrease the amount of transistors needed in microprocessing?

“Moore’s Law describes an important trend in the history of computer hardware: that the number of transistors that can be inexpensively placed on an integrated circuit is increasing exponentially, doubling approximately every two years. The observation was first made by Intel co-founder Gordon E. Moore in a 1965 paper.The trend has continued for more than half a century and is not expected to stop for another decade at least and perhaps much longer.”

There are other techniques like the graphene implementation but the transistor run at such high temperatures a extreme method of using volatile gases for cooling has to be used.

The advantage of Graphene allows the transistors to work within room temperatures. The technique is so new and in prototype form that there is no current industrial production for the technology. Only time can tell the fait of Moore’s Law and the current transistor implementations.

Bookmark Article:

Hacking software is said to be the old technique to gaining access to someone else’s computer system. The University of Illinois say the next level is for microprocessor hacking itself.

The research found that altering the processor could actually leave computer systems helpless to back-door attacks. Apparently hacking the chip was the easy part. The process requires few changes to only few parts of the processors circuitry.

The process of hacking the processor was executed by altering 1,341 logic gates of the chips one million logic gates. From the software aspect the processor would then need to be sent a special network packet which loaded a malicious firmware giving complete access to the attacker.

The only problem with the attack technique the attacker would need to physically get access to that computer. The likely way of using this technique would be to implement the hack during development stages of the processors. Would any employees be bribed to do so?

Bookmark Article:

High-K dielectric refers to a material with a high dielectric constant. The High-k is used in semiconductor manufacturing processes to replace the silicon dioxide within gate dielectrics. By implementation of the material, high-k further lowers the die size of processors. The technology is currently used in the Penryn Intel Processors.

IBM and partners have the lead with the implementation of HKMG high-k metal gate. IBMs semiconductor manufacturing has lead to the 32nm wafer. By implementing the 32nm technology they have found a 35 percent increase in performance compared to the 45nm process. The 32nm also consumed between 30 to 50 percent less power.

IBMs implementation of the high-k/metal gate within transistors that controlled primary on/off switching functionality lead to the development of 32nm chip circuitry. Intel Have already shown the 45nm process, so is there any point in upgrading your processor? Intel will no-doubt be designing their 35nm process.

Bookmark Article:

The Cell processor architecture developed by IBM Toshiba and Sony is said to start sampling under the name of SpursEngine 1000 or SE1000. The processor design was originally used on the Sony PS3 using seven cores synchronously.

Their targeted market is set to be for the consumer electrical products. The powerful design and advanced capabilities of the cell processor is to target video processing capabilities in particular. The cell processor will ideally be used for the growing high definition market especially with high resolution encoding/ decoding.

The SE1000 co-processor will include four slave cores used for the streaming of media. The chip is built on a 65nm process using 10W to 20W of power. The team are considering shrinking the die to reduce the power consumption.

You can see the SpursEngineis implemented into a PCI-E board which runs at a X1 bus speed. Toshiba are also working with third party companies to create applications and the development of HDTV set top boxes.

Bookmark Article:

The new socket is to be named LGA1366 for the Bloomfield and Gainestown Processors. The socket contains 600 extra pins than the socket LGA775 not to mention the three 64bit DDR3 memory channels. The Socket design seems similar to the current Intel Socket but the installation implementation is different.

I can imagine a lot of enthusiast out there are contemplating the new hardware, there no ability of backward compatibility! New heat sinks, new motherboard, new Die size! Well to be honest it is a new socket!

The LGA1366

Bookmark Article:

The Intel Nehalem Processor as we know of uses 2 to 8 cores. The processor also supports simultaneous multi threading which allows 4 to 16 thread computation. You have to remember this is all done on one processor, impressive!. The new Processor has a so called building block design, this allows non-CPU cores to be added to the architecture. Could this be implementation for their long awaited Larrabee integrated graphics?

The processor is currently clocking at 3.20Ghz using 256kib of L2 Cache. The processor will also share 8mib of level 3 Cache. The processor surprisingly has integrated tri-channel DDR3 memory running at 1333 MHz. Not to mention communicating with a whole new I/O hub. I previously mentioned Larrabbee and how I though the integrated memory was ingenious, something that has been long awaited.

I’d also like to point out that the architecture could change the communication between the C.P.U and the motherboard completely. The Processor will always need to communicate with the main-board but the trick is to completely remove bottlenecks between the two. Hopefully this is it. Read my previous post Collision of the top players which just explains the Larrabee compettion.

Nehalem is ready to hit the shelf’s in Q1 of 2009 this will be a long wait but I’m really contemplating upgrading at this stage and waiting near enough another year. Exciting Stuff!

Bookmark Article:

When AMD first released their Phenom quad core processors AMD had a TLB problem. The problem was based with the instruction where cache is asked to overwrite another memory location. Since the processor shared L3 Cache when two cores tried to access the same section of Cache memory the cores would race to access that memory location. Each core would then try and rewrite this new memory location causing the conflict.

Now that AMD have sorted their problem and increased clock speeds how do they compare to Intels processors? As we already know, AMD are well behind from where Intel’s research stands. Intel seem to be increasing the glab between the contenders.
Benchmark results have shown that using the Q9300 processor one of the lower end Intel processors still performs better than AMDs x4 9850 processor. I know what i’d prefer! If you dont then I’d consider getting a low end intel quad core and then when money is not an issue :S get the extreme quad version.

So I guess Intel are still leading and will always lead unless AMD come back with a new graphical architecture such as Larrabee cough.

Bookmark Article: