Trending

Bell Labs’ CMOS chip changed microprocessor design

In the late seventies of the twentieth century, which is 8 bit Processor It was on the latest cases and Cmos It was the weakest of semiconductor technology and engineers in AT & T‘s Bell laboratory Take a bold leap in the future. They made a high -risk bet on space IBMand IntelAndOther competitors in the performance of chips by combining an advanced 3.5 micron Cmos Manufacturing with a new 32 -bit processor structure.

Despite their creation – BelMac-32 The precise therapist – has never achieved commercial fame from the previous one like Intel 4004 (Released in 1971), its effect was proven more permanent. Almost each slice is in Smartphonesand LED computersToday, the panels devices are based on the principles of semiconductor supplementary conductors that were the Bellmac-32 pioneer.

With the approaching eighties, AT & T He was struggling with the transformation. For decades, the giant telecommunications company took control of the “MA Bell” telecommunications – with US audio contacts with it Western electrical Sub -manufacturing almost every phone in American homes and offices. The US federal government was pressing Get rid of anti -monopolyBut AT & T has a chance to expand computing.

With the already established computing companies on the market, AT & T could not catch up with a knee; Its strategy was to jump forward, and bellmac-32 was the starting point.

The Bellmac-32 chain was honored with IEEE teacher. The dedication celebrations this year are to be held in Nokia Bill Labz campus In Murray Hill, New Jersey, and in Computer History Museum In Mountain View, California.

An unparalleled slice

Instead of simulating the level of industry for the 8 -bit chips, the executives challenge AT & T Bell laboratory Engineers to provide a revolutionary thing: The first is commercially viable Delicate processor It is able to move 32 bits in one hour cycle. It will not take just a new chip but also a completely new structure – one can handle the communications switching and serve as the backbone of future computing systems.

He says, “We were not just building a faster chip,” he says. Michael CondryWho led the architecture team at the Bells’ Holmdel facility in New Jersey. “We were trying to design something that could carry both sound and account in the future.”

This configuration of the Bellmac-32 delicate processor had an improved integrated memory management unit for Unix-Like Operating systems.AT & T Center and History Center

At that time, CMOS technology was considered promising – but it is risky – NMOS and PMOS Designs and then use. Nmos Chips, which only relied on N Type TransistorsIt was fast but thirsty for power. The PMOS chips, which depend on the movement of the charged holes positively, were very slow. CMOS, with its mixed design, provided the ability to provide speed and energy. The benefits were so convincing that the industry soon saw that the need to twice the number of transistors (NMOS and PMOS per gate) deserve to barter.

With the contraction of the transistor sizes with the rapid progress of the semiconductor technology described by More LawSoon, the cost of double the transistor density has become manageable and eventually became minimal. But when Bell Labs took a high -risk gambling, CMOS manufacturing was still not installed and seemed relatively costly.

This did not deter the bell laboratories. By taking advantage of the experience of its deprivation in Holmell and Mauri Hill, as well as in Nabarefil, Illinois, the company collected a dream team of semiconductor engineers. The team included Condry. Song Mo “Steve” KangA rising star in Frame design; Victor HuangAnother processor designer, and dozens of AT & T Bell Labs employees. They started in 1978 to master a new CMOS process and create a 32 -bit small processor of scratch.

Architecture design

The Condry Architecture Group, a Life IEE colleague who will later become CTO in Intel, focused on building a system that would support at all Unix OS and C programming language. Both of them were in their cradle, but it is destined to dominate. To deal with the restrictions of the memory of the times – it was precious kilos – a complex educational group required less steps to implement and can be implemented in one hour cycle.

Engineers also built the slide to support Versamodole Eurocard (VME) parallel busEmpowerment Distributed computing Therefore, many contracts can deal with data processing in parallel. This allowed to make the chip that is also able to use it to control in the actual time.

The group wrote its own copy of UnixWith the actual time possibilities to ensure the compatibility of the new chips design with Industrial automation And similar applications. Bell Labs also invented Domino logicWhich increased the speed of treatment by reducing the delay in the complex Logical gates.

Additional testing and verification technologies have been developed and entered through the Bellmac-32 unit, which is the project to verify and test multiple groups and test it is led by Huang, which allowed the complex Chip To get zero or close to zero. This was the first of its kind in Vlsi Test. The methodological plan for the Bel Labs engineers to form the work of their colleagues eventually made the total design of the multi -slide group family working together smoothly as a complete system of micro -basins.

Then the most difficult part came: in fact building the slide.

Floor maps and colored pencils

Kang, a fellow IEEE who later became president of The Advanced Korea Institute for Science and Technology ((Kaist) In Daijon, South Korea. With no CAD Kang says that the tools available for full verification Calcomp Plots. The plans showed how transistors and circles lines Interconnection It should be arranged inside the slide to provide the required outputs. The team collected it on the floor with an adhesive tape to create a huge square map over 6 meters on the side. Kang and his colleagues follow every hand circle with pens in color, and searching for rest periods, overlap or connected links.

Get it

Once the material design was closed, the team faced another obstacle: manufacturing. The chips were manufactured in a Western electrical A facility in Alnettown, Pennsylvania, but Kang remembers that the rates of return (the percentage of chips on a Silicone chip That meets the standards of performance and quality) was bleak.

To address this, Kang and his colleagues from New Jersey led to the factory every day, refuted their sleeves, and did everything required, including overwhelming floors and calibration of the standards, to build intimate friendship and instill confidence in that the most complex product that plant workers tried to actually produce there.

“We weren’t just building a faster chip. We were trying to design something that could carry both sound and account in the future.” Michael Condry, Bilmak -32 team for architecture

“Building the team was good,” says Kang. “After several months, the Western Electric managed to produce more than the required number of good chips.”

The first version of Bellmac-32, which was ready by 1980, did not make expectations. Instead of reaching the 4MHz performance target, just running 2MHz. Engineers discovered that the modern situation Takka Ricken The test equipment they were using was defective, with the effects of the probe and the test of the test, which leads to inaccurate measurements, so they worked with Takeda Rocke A team to develop correction schedules that correct measuring errors.

It was the second generation of Bellmac chips Clock speeds This exceeded 6.2 MHz, and sometimes up to 9. This was quickly overseeing his time. 16 bits Intel 8008 processor inside the original IBM computer that was released in 1981 Run at 4.77 MHz.

Why did Bellmac-32 go to the prevailing

Despite her artistic promise, Bellmac-32 did not find a wide commercial use. According to Condry, the AT&T axis towards obtaining the equipment factory NCRAnd that began to look forward in the late eighties, means that the company has chosen to support a different set of chips. But by that time, the Beellmac-32 legacy was already growing.

“Before Bellmac-32, NMOS was dominant,” Condry says. “But CMOS changed the market because it turned out to be more effective in FAB.”

At the right time, this perception has reshaped the semi -conductor scene. CMOS will become the basis for talking Microoring treatmentsRun the digital revolution in desktop computers, smartphones and more.

The boldness of Bell Labs – to take an unspected manufacturing process and jump a full generation of chip structure – flows as a historical note in technological history.

Kang says: “We were on the limits of what was possible. We just followed the path – we made a new one,” Kang says. Huang, an IEEA life colleague who later became deputy director Institute of Define Electronics, Singapore“This not only included the structure of chips and design, but also checking the chip on a large scale-with CAD but without digital simulation tools today or even the plate [which is the standard method for checking whether a circuit design for an electronic system that uses chips works before making permanent connections by soldering the circuit elements together]”

Condry, Kang and Huang are looked at this period and expressed their admiration for many AT & T employees whose skill and dedication make a series of Bellmac-32 chips.

Manage it IEEE History Center With the support of the donors, the Milestone program recognizes the distinguished technical developments worldwide. the IEEE North Jersey section Under the nomination sponsorship.

From your site articles

Related articles about the web

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button