Intel 8008: An abandoned chip that pioneered a dynasty
Latest update time：2022-03-09 11:22
Reads：
Source: The content is compiled from eejournal by Semiconductor Industry Observer (ID: icbank) , thank you.
Early microprocessors spanned two major eras of computing. The first era was from the late 1960s to the 1970s, when computer system engineers implemented minicomputer processor architectures and processor boards using TTL parts, bipolar PROMs, stone knives, and bearskins. Each minicomputer manufacturer, including Digital Equipment Corp (DEC), Data General (DG), Prime, Computer Automation, IBM, Burroughs, HP, Four-Phase, NCR, and Univac, had its own proprietary minicomputer architecture, ISA, and specialized peripherals. Those were wild times.
In the second era, which partially overlapped with the first, some computer manufacturers and some semiconductor suppliers began designing LSI chips that integrated larger and larger processor blocks onto a single IC, and
the first commercially successful single-chip microprocessor was also an early product of the second era.
That was the 4-bit Intel 4004, which just celebrated its 50th birthday on November 15, 2021.
The 4004 microprocessor arrived with great fanfare, but Intel was developing a second microprocessor at the same time.
That microprocessor was the 8-bit Intel 8008.
The stories of these two processors are intertwined in many ways, but they are independent in other ways.
As with the Intel 4004 microprocessor, an external customer drove the development of the Intel 8008 microprocessor. In the case of the Intel 4004 microprocessor, the external customer was Busicom, a Japanese calculator company that wanted the Intel 4004 to build a high-end desktop calculator. Intel designed and built the 4004 and three companion chips for Busicom, then negotiated to sell the Intel 4004 microprocessor to other companies in exchange for Busicom's price concessions on the parts. Intel introduced the 4004 microprocessor to the world on November 15, 1971.
The 4-bit Intel 4004 microprocessor was part of a 4-chip set that included a microprocessor, ROM, RAM, and shift registers for I/O expansion. This 4-chip set, known as the Intel MCS-4, represented a walled garden. The microprocessor's unique multiplexed 4-bit bus formed the garden wall. If another chip wanted to communicate with the Intel 4004 microprocessor, it had to implement the control and timing logic required to interface with the multiplexed bus.
At the time, Intel's main business was selling memory, specifically RAM and ROM. These memories all had parallel address and data buses. None were directly compatible with the Intel 4004's unique bus. This was not the vision for the Intel 8008 microprocessor, which was developed with a simpler system bus in mind. The 8008 was designed to use standard RAM and ROM, which Intel also manufactured.
Intel's external customer for the 8008 was the Computer Terminal Corporation (CTC) in San Antonio, Texas. CTC made significant contributions to the architecture and ISA of the Intel 8008 microprocessor. The definition of the microprocessor was based on CTC's existing plans for an 8-bit board-level processor that had about 100 SSI and MSI TTL chips built into it. Using their experience with the development of the 4004, Ted Hoff and Stan Mazor at Intel reviewed and tweaked the architecture of CTC's planned processor to improve it slightly in order to simplify its manufacture as a single chip and allow Intel to cram the microprocessor into a tiny 18-pin DIP. The architectural tweaks involved changes to the ISA.
CTC built dumb terminals, and Intel provided a custom 512-bit circular shift register for the application. CTC wanted to enter the growing minicomputer terminal business and was developing an 8-bit embedded processor board as the basis for the smart terminals it planned to build. As part of this development, Victor Poor, CTC's technical director, looked at Intel's 64-bit (not Kbit or Mbit) bipolar SRAM as a possible way to implement CTC's processor registers. He asked Intel if they could make a custom version of the SRAM by adding counters to the RAM design to enable it to be used as a push-down stack register.
Intel's Stan Mazor, who had been involved in the early definition of the Intel 4004, discussed the requirements with Poor, gained a deeper understanding of the CTC processor architecture, and then wrote three proposals for a custom chip. Mazor's first proposal was for an 8-bit register with a stack counter, per Poor's initial request. The second proposed chip was a register stack with an additional arithmetic unit (similar in concept to the four-phase AL1). The third proposal was for a full 8-bit CTC CPU on a single chip. That one caught Poor's attention.
Mazor did not even describe the CTC processor architecture or its ISA in detail when he created this proposal, but Poor was so interested in the third proposal that he sent Mazor the programming manual for the processor, which described the architecture at the assembly-language level. Mazor and Ted Hoff, fresh off the Intel 4004 project (leaving Federico Faggin to develop and research silicon gate processes and logic implementation details), dug into the programming manual for the CTC processor and created a more detailed proposal for a single-chip version of the CTC eight-bit CPU processor. Intel's CTC sales staff then got involved, and CTC placed a $3 million purchase order for the single-chip processor for 100,000 parts at $30 per piece on March 18, 1970.
Newly hired Hal Feeney became the chip designer for the 8-bit processor project. Mazor and Feeney began their development with a proposal that Intel had provided to CTC. However, the project soon came to a halt as questions arose as to whether CTC was really committed to developing such a custom chip. (A similar pause occurred in the development of Intel's 4004, but the fault lay entirely with Intel.) With the 8-bit microprocessor project stalled, Feeney began helping Federico Faggin complete the development of the 4004 microprocessor and the MCS-4 chipset.
There is something very interesting about the detailed microprocessor proposal that Intel sent to the CTC. The proposal accidentally included a design flow that prevented the microprocessor from properly handling interrupts. As originally defined, the interrupt mechanism would cause the microprocessor to call the interrupt service routine without first placing the return address on the processor's stack, so the interrupt service routine could not return from the interrupt. This flaw rendered the proposed interrupt mechanism useless.
Around this time, Texas Instruments (TI) also began developing a single-chip processor for CTC, based on CTC's specifications and at CTC's request. TI's processor would be called the TMX 1795. Although TI had originally proposed a 3-chip set for CTC's processor, it switched to a single-chip design sometime after Intel made its own proposal to CTC. TI built the TMX 1795 on a very large die, which was not economical in mass production.
TI manufactured the TMX 1795 but failed to sell it to CTC and never successfully marketed the device. Instead, TI successfully sold a large number of TTL, calculator, and other chips. The TMX 1795 microprocessor survives in the story and several artifacts at the Computer History Museum in Mountain View, California, including a 1996 video of the device in action.
Meanwhile, the 6-month project hiatus for the 8008 microprocessor actually helped Intel debug and improve the design. First, it provided time to reflect on and refine the architecture of the 8-bit processor. The original CTC instruction set included a bitwise branch instruction. The Intel design team determined that the instruction was not needed and removed it to simplify the hardware design of the processor. At the same time, the Intel design team determined that the processor would benefit greatly from increment and decrement instructions, so they added these two instructions to the ISA of the 8008. Interrupts also allowed the Intel 8008 design team to catch and fix a flawed interrupt mechanism.
Additionally, the interruption provided time for Intel's groundbreaking 1103 1Kbit DRAM to go into production. This was a significant event for many reasons, but an immediate benefit to the 8008 project was the 18-pin DIP that housed the first large-capacity DRAM. Because this package was now officially approved by Intel's production group, the Intel 8008 design team could use it to add two valuable pins to the Intel 8008. Previously, the 8008 design team had been limited to 16 pins because that was the package Intel's production group already had on hand.
According to Feeney, the two extra pins were a much-needed improvement in 8-bit processors. One benefit: One of the extra pins was used to bring extra status information out of the microprocessor, which helped implement the microprocessor's stack and allowed the interrupt mechanism to work properly.
The Intel 8080 microprocessor came in an 18-pin DIP package, which Intel also used for its 11031Kbit DRAM.
Through careful design and Faggin's improvements to semiconductor processes and design methods (some of which he had developed for Intel's MCS-4 project), Feeney's 8008 microprocessor chip was only slightly larger than the Intel 4004 microprocessor chip (3500 transistors compared to 2200 for the 4004), even though the 8008 required 50% more transistors. Feeney's design was therefore very manufacturable. Unfortunately, the Intel 8008 microprocessor implemented with silicon-gate MOS ran significantly slower than the bit-serial TTL version of the processor designed and implemented by CTC. It also required a large number of support chips to create a complete system, although not as many as the 100 chips on the CTC processor board.
CTC evaluated the Intel 8008 microprocessor in late 1971 and said, "No thanks." Too little, too late. The company had already developed the first Datapoint 2200 terminal with a bit-serial TTL processor and was working on a faster parallel implementation for the next generation of terminals. CTC sold the Datapoint 2200 machines until 1979, upgrading the design of the TTL processor board several times along the way. With each revision, the TTL processor got faster.
Datapoint 2200 intelligent terminal, with a removable 2.5MB hard drive box on the top.
The Datapoint 2200 terminal was more than just a terminal. It was a small computer that you could program in BASIC or PL/B and came with one or two digital cassette tape drives, a companion 2.5Mbyte hard drive, and later an optional floppy drive. Some historians call it the first personal computer, and it was clearly designed to be a computer, but it wasn't microprocessor-based. It just produced one. CTC's Datapoint 2200 intelligent computer/terminal sold well, and the company later changed its name to Datapoint.
Meanwhile, having lost a major customer, Intel now owned the copyright to the 8008 microprocessor and decided to sell it commercially. Although many online articles and references use April 1972 as the launch date for the Intel 8008, the company released the microprocessor on March 13, 1972, just four months after releasing the Intel 4004. Some online citations say that was 1974, apparently confusing the Intel 8008 and 8080 microprocessors. However, thanks to extensive research by Ken Shirriff, the official unveiling of the Intel 8008 appears to have been a 1-page article written by Stephen William Fields titled "8-bit Parallel Processor Offered on a Single Chip" published in the March 13, 1972 issue of Electronics magazine.
Intel now sold not one, but two single-chip microprocessors: the 4004 and the 8008. They had a clear lead in this new market.
The Intel 8008 microprocessor had a 16Kbyte address space (using 14-bit addressing, which was considered huge at the time, four times larger than the 4004 microprocessor) and a two-phase 800KHz clock (for the fastest speed grade). According to the datasheet, an 8008 instruction fetch/execute cycle required at least five processor states, or 10 clocks. This was a peak instruction execution rate of 80,000 instructions per second.
By today's microprocessor standards, or even by the standards of the past 30 years, Intel's 8008 was indeed weak and slow. But it was the first commercially available 8-bit single-chip microprocessor that you could build useful systems with. System designers began integrating the Intel 8008 into many new products, including embedded systems, such as the first two versions of Hewlett-Packard's legendary and long-lived 2640 series of intelligent CRT terminals, and some early microcomputers. In addition, the introduction of the 8008 microprocessor helped Intel sell more of its main products, DRAM and EPROM, by supporting and even encouraging the design of systems that required semiconductor memory.
The Intel 8080 and 8085, the 8-bit Zilog Z80, and all the x86 microprocessors developed by Intel and other processor vendors over the past half century carry some reptilian DNA from the original Intel 8008. If you don't believe me, look closely at the register sets of these microprocessors.
Later, Federico Faggin left Intel and founded Zilog in 1974. He put some 8008 microprocessor DNA into the extremely successful 8-bit Z80 microprocessor, which was released in 1975. But that's another story.
★ Click [Read original text] at the end of the article to view the original link of this article!
*Disclaimer: This article is originally written by the author. The content of the article is the author's personal opinion. Semiconductor Industry Observer reprints it only to convey a different point of view. It does not mean that Semiconductor Industry Observer agrees or supports this point of view. If you have any objections, please contact Semiconductor Industry Observer.
Today is the 2973rd content shared by "Semiconductor Industry Observer" for you, welcome to follow.
Recommended Reading
★ Understanding semiconductors from three dimensions
★ Heterogeneous computing, is it about to explode?
★ Compound semiconductor manufacturing landscape changes
Semiconductor Industry Observation
" The first vertical media in semiconductor industry "
Real-time professional original depth
Scan the QR code , reply to the keywords below, and read more
Wafers｜ICs｜Equipment ｜Automotive Chips｜Storage｜TSMC｜AI｜Packaging
Reply
Submit your article
and read "How to become a member of "Semiconductor Industry Observer""
Reply Search and you can easily find other articles that interest you!
Click to read the original text to view this article
Original link!
Featured Posts
京公网安备 11010802033920号