Apple Officially has launched the iPad, essentially a scaled up iPhone, and judging by the model number, it started development several years ago around the time of the first iPod Touch. Many of the components are similar to the iPhone, if not the same. Obviously the biggest processor ews is the Apple A4 ARM processor at its heart. Its developed by Apple, and built by Samsung. a multi-die package, it includes the ARM processor, and PowerVR Graphics on one die, and then two 128MB DRAM dies as well.

iPad Motherboard - A4 CPU by Samsung

What is interesting about the A4? It has a 64bit memory bus, rather then the more standard 32bit but found on most ARM devices. This likely for faster memory access to support faster graphics.

Wifi, Bluetooth, and FM functions are handled by a Broadcom BCM4329 which includes two processors of its own (the documentation of this part does not state what architecture they are, but MIPS or XA-RISC is likely)

The screen controller is the Broadcom BCM5973/5974 which has been used by Apple for several years. The Baseband processor in the 3G version is the Infineon X-Gold 608 which contains a 312MHz ARM926 processor, a 2007 design, so probably saved Apple some money.

So all told the iPad 3G contains at least 4 seperate processors.

Instrinsity: ARM Processor Design House

In related news rumors are flying about Apple acquiring Intrinsity. Instrinsity is a processor design house which specializes in, you guessed it, ARM processors. Its likely they have helped Apple more on the A4 then P.A. Semi (another Apple acquisition) This is further supported by the fact that Intrinsity has worked a lot lately with Samsung in developing a 1GHz ARM CPU (the Hummingbird). Who fab’s the A4? Yah, Samsung.

Sources:
EE Times: Inside the iPad
EE TImes: Inside the iPad: Broadcom/Samsung
iFixit: Teardown
ars technica: Apple buys Instrinsity

AMD today unvelied their new CPUs for gamers, which seem to readily appeal to poker players. The New Lottery Core series contains 12 cores. However, as is normal with production of such a complex device some, or all of these cores may not work.  AMD has designed the cores/CPU such that if a core fails, the others will function fine, this of course saves them testing, as well as packaging costs as each chip can use the same package and markings.

Obviously these are targeted specifically for the hobbiest, and priced accordingly. You may get one with all 12 cores functioning and if so AMD considers you ‘very lucky’ apparently not so much if you get one with no working cores.

Currently these are only available in the UK, but if successful should spread elsewhere.

More Info at PC Pro UK

Routing internet traffic takes a vast amount of processing power. Gone are the days of dialup, where data could be processed with relative ease. Now with every connection pumping out data at rates measured in Mbits, high end devices are needed to route, inspect, and shape that traffic. These devices literally inspect, and operate on EVERY packet that goes through them, that require incredible speed to do so without affecting the speed of the data.

Netronome NFP-32xx block diagram

Enter such companies as Netronome, a maker of network flow processors, compatible with the Intel IXP28xx series. Each one of these processors has a 700MHz ARM CPU, as well as 8 1.4GHz RISC based Microengines are capable of delivering 56700MIPS. Expect to see more such products as companies work to meet President Obama’s National Broadband Plan.

Earlier we talked about why old designs are not going away. Today lets look at a new design. The Intel Corei7-980X Extreme. Currently one of the most powerful processors in existence.  For comparisons sake I am going to loosely compare the 980X to the Motorola 68000 which arguably was the first 32 bit CPU. (yes the 980x is technically 64 bit but for the sake of comparison we’ll use it).

Too put that in perspective you could stick over 29,000 MC68K cores on a 980X die.

Intel Core i7-980X Processor die

So while old designs are still in wide use (and will continue to be) the new processors are nothing short of amazing.

980G Review may be found here at Hot Hardware

This topic comes up a lot. Why are old CPU designs still used? and used a lot. Every few years a company will make a statement about their new 32bit or 16bit CPU/MCU design that makes ‘migration from an 8-bit design easy’ or seeks to replace 8-bit microcontrollers entirely. It does not happen, and will not for decades.

1981 Motorola 68000 a 16/32bit CPU

8-bit processors first debuted in 1972 with the Intel 8008, so they are pushing 40 years. What many don’t realize is that 32bit processors debuted in 1979 (Motorola 68k and National Semiconductor 32k). So 32 bit is nothing new, and the same reason applies today as it did 30 years ago. Why use a Ferrari when a Chevy will do? Most designs simply don’t need that power, or complexity, 8-bits is MORE then enough to run a toaster oven, a bread machine, or your TV remote.

Embedded.com had an article talking about this issue just last week. So why do companies say that their design will replace 8-bit? It’s good PR, it gets people talking about their new processor, and thats not a bad thing.

One thing we talk a lot about is power budgets, getting as much performance out of a chip, with the least power draw possible. There is of course other budgets, such as space, or one we all know, cost.

When you buy a new Intel Core i7 processor you expect to spend $300 or so. Its common the CPU’s in phones etc cost $10-$20 each. However many applications don’t need or cannot afford that much power. In fact, most CPUs are NOT vastly powerful, or expensive.  For example ST just annoucned a ARM Cortex-M3 based 32bit CPU, running at 24MHz, its cost? $0.85. TI has a whole line of its popular MSP430 16bit microcontrollers, running at up to 25MHz, they cost $0.25, yup less then a pack of gum for a fully featured processor.

The Nvidia Tegra series of processors 9really SoCs as they integrate memory controller, I/O bridges, etc all on one chip) just keeps getting bigger and better.

Tegra Processor

The new Tegra 2 announced at CES 2010 now integrates a dual core ARM Cortex-A9 running at up to 1GHz.  That’s all well and dandy but it doesn’t stop there. Nvidia also adds separate processor units for HD encoding, still image processing, audio, 2d/3D, HD playback and mobile phone baseband functions.  Each core can be independently clocked, throttled, or turned off as needed, resulting in a processor that can run at 500mW.

This is the type of processor you will be seeing in devices like the forthcoming Apple tablet, and future smart phones.  All this again begs the question: When will Microsoft port Windows 7 to ARM?

More Info here

EETimes has another article about a person being charged/convicted of selling counterfeit chips to the US navy. This has been a growing problem in the electronics industry for the last decade, but has its roots much earlier then that.

It was common in the 90’s for counterfeiters (aka remarkers) to take a processor, wipe the markings, and mark it with a higher speed. This was rather common with the Pentium era and newer, but occurred with 486’s as well. To a computer user this typically meant that their computer ran much warmer, and often times less stable.

To a collector this means you must be VERY careful when looking at processors in your museum to ensure that rare sample you have, is not in fact a clever forgery, or that Pentium 133 is not in fact a remarked 75.

Having your computer crash or having a few fake CPUs in your collection is a mere annoyance, but what about actual use? For example a part listed as mil-spec, with a wide temperature operating band, that controls a ships defensive systems? If this is in fact a fake (remarked from a commercial spec IC, which has been happening). The system could and likely WILL fail at the worst time. The result? People lose their lives.

Perhaps 2 of the most famous RISC architectures, ARM, and MIPS, have been around for years.  Like many well designed architectures, they continue to be adapted, enhanced, and used in litterally billions of devices.

Recently Altera (maker of FPGAs and CPLD’s) licensed the use of the MIPS32 core.  This is a first for MIPS, as they now join ARM, and PowerPC as major cores available in FPGAs. Source: EETimes

On the topic of ARM, they have been working extensively recently with GlobalFoundries to provide customers with proven IP, specifically the ARM Cortex-A9.  They are developing as a SoC core on a 28nm process.  ARM currently works with TSMC to test, and prove various cores on TSMC’s process, and now will do so with GlobalFoundries. What this does in enable system builders to have IP blocks that are PROVEN on a given manufacturing node. It takes some of the guess work, and certainly risk, out of developing a ASSP, ASIC, or SoC.

When you think of Atmel what do you typically thing? High Speed 8051 microcontrollers and AVR RISC processors.  Maybe the occasional EEPROM. But there is another side of Atmel.

Atmel AT697F Rad-Hard SPARC

Atmel also makes a line of radiation hardened space qualified SPARC CPU’s. These are used extensively by the  European Space Agency and other satellite builders.  Atmel just released anew one too. The AT697F, a revision of the AT697E. What can it do? well oits a full 32bit SPARC V8 core, running at 100MHz (90MIPS). Its made on 0.18u which is very impressive for a space based processors. Most of Atmels other designs are basing on a half micron process.

A larger process like half a micron gives increased radiation resistance, but at the expense of speed. At 0.18u Atmel has got the speed up to 100MHz, AND increased radiation tolerance to 300krads.  To put that in perspective, a dose of only 1 krad (1000 rems) will kill you 99% of the time so these processors can continue to function at over 300 times that.

Source:  EE Product Center