Milestone-Proposal talk:MPD7720 DSP

Evaluation from Consulted Experts on Docket # 2018-14 (“World’s first commercial digital signal processor chip, μPD7720, with a 16x16-bit multiplier, 1980" -- Naohisa (talk) 02:29, 8 August 2019 (UTC)

Evaluation of Docket # 2018-14 (“World’s first commercial digital signal processor chip, μPD7720, with a 16x16-bit multiplier, 1980”) for IEEE Milestone

Consulted expert: Naohisa Ohta

Around 1980, I was engaged in research and development of speech coding systems for digital communications. In order to optimize the coding scheme, the algorithm was always modified in the research process for evaluation. A high-speed and flexible digital processing device was the key to success for such real-time applications. I witnessed how the digital signal processor μPD7720 had an impact in the actual situations as described in the proposal and in the advocate recommendation as well.

The μPD7720 was the first commercially available digital signal processor chip that had capability of high-speed digital signal processing with flexibility (programmability). The chip had a unique processing architecture such as a 16x16-bit parallel multiplier and a novel data bus connections that made it possible to efficiently perform multiply-and-add operations that are essential in signal processing.

Most impressively, μPD7720 became commercially available from a very early stage (1980) and it was the only digital signal processing chip commercially available for 4 years. During that period, μPD7720 (and its family chips) stimulated a variety of signal processing applications not only in research but also in product development, leading to improved cost-performance. It made an impact to a wide variety of signal processing equipment in communication and broadcasting, especially in voice-band data modems and ADPCM speech codecs, leading to the digital signal processing era.

As a consulted expert in the research of signal processing and telecommunication applications, I strongly support this IEEE Milestone proposal.

Naohisa Ohta--Naohisa (talk) 02:29, 8 August 2019 (UTC) IEEE Fellow, Life Member Media Design Laboratory Keio University Graduate School of Media Design, Japan

Re: Evaluation of Docket # 2018-14 (“World’s first commercial digital signal processor chip, μPD7720, with a 16x16-bit multiplier, 1980”) for IEEE Milestone -- Takao Onoye (talk) 01:22, 27 August 2019 (UTC)

Consulted Expert: Takao Onoye

In 1980, NEC presented the world first commercial single chip DSP, uPD7720, at ISSCC1980, which are 3 years ahead of the introduction of Texas Instruments' TMS320. Around that time, the state-of-arts processors are i8086, Z8000, and MC68000, and none of PCs were available.

Tthe uPD7720 fixed-point DSP, which integrates a high-speed multiplier on a chip, expanded the market of DSPs. It is notable that the DSP was developed by a consumer electronics company and hence was used not only in a telephony but also in various applications including data processing (e.g. modems and disk drive controller) and media processing (e.g. speech processing and graphics controllers).

In this sense, the uPD7720 laid the first foundation of current wide spread of DSPs and I am sure that it can be regarded as the good candidate of an IEEE milestone.

Takao Onoye, Dr.Eng. Executive Vice President for Research, Osaka University, 2019-2020 Chair, IEEE Japan Council, 2019 Vice Chair - Professional Activities, IEEE Region 10

Advocate Recommendation of Docket # 2018-14 (“World’s first commercial digital signal processor chip, μPD7720, with a 16x16-bit multiplier, 1980”) for IEEE Milestone -- Shirakawa (talk) 04:42, 25 September 2019 (UTC)

IEEE technical field: Signal Processing, Solid-State Circuits Consulted experts in the technical field: - Prof. Naohisa Ohta, Graduate School of Media Design, Keio University, Yokohama, Japan - Prof. Takao Onoye, Graduate School of Information Science and Technology, Osaka University, Suita, Japan

Recommendation of Docket # 2018-14 for IEEE Milestone: The world’s first “commercially available” digital signal processor (DSP) chip, μPD7720, was developed by engineers represented by Dr. Takao Nishitani and Mr. Yuichi Kawakami of NEC Corporation and release in 1980. It was presented at both IEEE ISSCC1980 and IEEE ICASSP1980, where AT&T Bell Labs. also presented at the same sessions a similar DSP chip which, however, was not commercially available. Texas Instruments (TI), recognized as the leader in the DSP market, announced its first DSP, TMS32010, in 1982, but released it in 1984. Thus NEC’s μPD7720 had been the only DSP commercially available until TI’s TMS32010 was released in 1984. It should be pointed out here that compared to AT&T’s DSP, μPD7720 was four times faster in multi-plication due to single-clock 16x16-bit multiplication within 250-nsec, leading to the true real-time processing for complex systems. On the other hand, μPD7720 greatly helped establish the standard algorithm of ITU-T ADPCM (adaptive differential pulse code modulation) Standard for the modern telephone and VoIP networks as well as Bluetooth connection through its software programmable capability. This μPD7720 was extended to a new DSP, μPD7730, with an 8x8 bit multiplier and a barrel shifter to implement floating-point multiplication essential to the ITU-T ADPCM Standard. Moreover, μPD7720 as well as its family chips, including a shrunk CMOS version with an 8 MHz clock and an extended RAM, μPD77C25, were used in a wide variety of applications, such as digital transmission signal modulation, speech, audio, and video coding, speech recognition/synthesis, signal enhancement/interference cancellation, etc. In addition, telecommunication equipment such as switches and transmission terminals, multimedia players such as Walkman, mobile phone handset including smartphones, digital cameras and camcorders, digital TV receivers, car navigation systems, DVD recorders/players, PCs, and tablet PCs are implemented by DSPs or DSP cores. It should be added here that the development of μPD7720 received the 2017 IEEE Donald O. Pederson Award, and that μPD7730 and the contribution to ITU-T ADPCM Standard were recognized as the 2018 Industry Innovation Award by IEEE Signal Processing Society. Thus this advocate, strongly supported by two world experts, would like to heartily recommend the proposal of Docket # 2018-14 for IEEE Milestone.

Check List: 1) Is the proposal for an achievement rather than for a person? Yes, the proposal is for an achievement.

2) Does the location being proposed for the milestone plaque have a direct and logical connection with the work (e.g. where the achievement was developed, tested, demonstrated, or installed)? Yes, the proposed location has a direct access and logical connection with the work. Is the location truly publicly-accessible? Yes, it is publicly accessible.

3) Can the plaque site physically support the weight of the plaque? Yes, the plaque site can physically support the weight of the plaque. Is there space to fit the plaque’s dimensions? Yes, there is space to keep the plaque.

4) Is there an IEEE organizational unit(s) willing to sponsor the milestone? Yes, Tokyo Section, IEEE Japan Council, is willing to sponsor the milestone.

5) Is the work truly a significant achievement vs. an incremental improvement to an existing technology? The work is a significant achievement as is clear from the fact that no smartphone works without this achievement.

6) Were there prior or contemporary achievements of a similar nature? Yes, there was a similar achievement by AT&T Bell Labs. They announced a similar but a four-times slower chip at ICASSP80 by coincidence. If so, what sets this achievement off from them? The Bell Labs. chip was never available as a commercial product, which makes the nominated achievement far more important.

7) Whether the achievement (or the particular version of the technology being proposed) truly led to a functioning, useful, or marketable technology. Yes, the achievement was a commercially available digital signal processor chip as shown by the accompanying proof.

8) Is the proposal adequately supported by significant references and citations (minimum of five, but as many as needed to support the milestone), such as patents, contemporary newspaper articles, journal articles, or citations to pages in scholarly books? Yes, the proposal is supported by conference records, journal articles, newspaper articles, data sheets, and patents.

9) Are the scholarly references recent? Some are recent and some are not. Many of them are from the days of achievement made public for the first time. If not, has there been more recent scholarly research which ought to be taken into account? Yes, some of the references are from 2008, 2012, and 2017.

10) If the names of individual persons are mentioned in the citation, care should be taken that those persons are the engineers (as opposed to, say, the program managers) who designed the technology. It was confirmed that those individuals are all engineers who designed the technology.

11) The citation’s word length must not exceed 70 words (the maximum which will fit on the plaque). 60-65 words are preferred for aesthetic reasons. The citation word length is 68 words, which does not exceed 70 words.

12) Is the citation understandable by the general public (e.g. spell out acronyms, avoid jargon)? Yes, the citation is understandable by the general public.

13) Milestone plaques are permanent and will be read by future viewers. Is the citation phrased in a way that does not depend on present perspective? Yes, the citation is phrased in a way that does not depend on present perspective. Will it be understandable in the future?" Yes, it will be understandable in the future.

14) After the process has been completed, the proposal will be either rejected, approved, or approved only after required changes are made. The citation is in the final form, with agreement among the proposers and advocate.

Support MIlestone but Citation needs improvement -- John Vardalas (talk) 19:42, 8 January 2020 (UTC)

Given the positive reviews from the external referees, the achievement in this proposal deserves to be a an IEEE Milestone. But, in my view, the citation needs improvement.

1) The beginning of the citation is ambiguous. What does "Engineers represented by Dr. Takao Nishitani and Mr. Yuichi Kawakami" mean?

2) I do not believe that the citation adequately describes the uniqueness of this DSP. Even though this product was later surpassed by other DSP products, the proposers and reviewers make a point of noting that this DSP opened the door to the development of new and important technologies.

3) I also feel that, whenever feasible, the citation should create a sense of place between the reader and the achievement, i.e. in this location is where it happened, as opposed to simply in this organization.

4) I feel that a concluding sentence is needed that would make the impact resonate with the general public.

My proposed citation below addresses all but my last comment. I leave it to others to suggest a concluding sentence, if they think, as I do, that one is needed.

"On this site, in 1980, engineers at the NEC Corporation developed a software programmable digital signal processor (DSP), μPD7720, with a 250-nsec, 16x16-bit. Being the first commercially available DSP, it helped accelerate the development of a wide variety of signal processing equipment in communication and broadcasting such as voice-band data modems and ADPCM speech codecs, and opened the door to the digital signal processing era."

Re: Support MIlestone but Citation needs improvement -- Ebruton (talk) 12:01, 10 January 2020 (UTC)

Again, I agree with John. I think the final sentence including acknowledgement that it was the first commercially available DSP helps make this more accessible to the non-expert.

Proposal Review -- Jason.k.hui (talk) 19:23, 11 January 2020 (UTC)

Some comments:

1) Suggest rewording "Engineers represented by Dr. Takao Nishitani and Mr. Yuichi Kawakami of NEC Corporation" to "The Nippon Electric Company, Limited, later NEC Corporation, developed..."

2) What does "digital signal processing era" mean?

Revised citation | Policy on naming of individuals -- Dmichelson (talk) 19:24, 15 January 2020 (UTC)

John V's citation is much better than the original version. If not including names of individuals is a defacto History Committee policy, it should be included in the Milestone Proposal guidelines.

Suggested edits to citation to emphasise what made this achievement different -- Administrator4 (talk) 15:17, 20 January 2020 (UTC)

At its teleconference of 15 January 2020, the IEEE History Committee discussed the proposal, but did not take any action on it. Members said they were unsure of what made this achievement different from similar chips. In order to make the plaque citation stronger, I have made edits by taking wording from several places: 1) The supporting write up in the proposal, and also from the advocate's and expert's reviews. I propose the following, and welcome the proposers' thoughts. Note: From what I can tell, the official name prior to 1983 was Nippon Electric Company, although it now appears as NEC Corporation.

First Commercial Digital Signal Processor Chip, μPD7720, 1980

NEC's (Nippon Electric Corporation) μPD7720 was the first commercially-available software-programmable digital signal processing chip. Its novel bus structure, 250-nsec instruction cycle, 16x16-bit multiplier, and faster FIR (finite impulse response) filtering provided true realtime processing for complex systems. It accelerated development of communication and broadcasting signal processing equipment such as voice-band data modems and ADPCM speech codecs, and helped initiate the digital signal processing era.

(69 words)

Proposer's edits to citation posted on proposer's behalf -- Administrator4 (talk) 14:42, 30 January 2020 (UTC)

We carefully considered your recommendation, and propose the following citation. There are two changes in the citation. In the last sentence, we changed "opening" to "and opened" and deleted "ADPCM" to reduce 1 word.

First Commercial Digital Signal Processor Chip, μPD7720, 1980

The first commercially-available software-programmable digital signal processor chip, μPD7720, was developed at NEC (formerly Nippon Electric Company). Its novel bus structure, 250-nsec instruction cycle, and 16x16-bit multiplier, enabling fast finite impulse response filtering, provided true realtime processing for complex systems. It accelerated development of communication and broadcasting signal processing equipment such as voice-band data modems and speech codecs, and opened the digital signal processing era. (70 words)

Re: Proposer's edits to citation posted on proposer's behalf -- Jason.k.hui (talk) 16:55, 27 February 2020 (UTC)

A couple of suggested minor edits to the above citation:

1) Change "realtime" to "real-time"

2) Substitute "opened" in the last sentence with "inaugurated"

Re: Re: Proposer's edits to citation posted on proposer's behalf -- John Vardalas (talk) 22:07, 27 February 2020 (UTC)

Subject to Jason's two minor edits, I support this Milestone.

Re: Re: Re: Proposer's edits to citation posted on proposer's behalf -- Amy Bix (talk) 23:02, 27 February 2020 (UTC)

Could I suggest rewriting the first sentence to avoid passive voice? (as per good writing practice):

NEC (formerly Nippon Electric Company) developed the first commercially-available software-programmable digital signal processor chip, μPD7720, in 1980.

Re: Re: Re: Re: Proposer's edits to citation posted on proposer's behalf -- Ebruton (talk) 17:40, 28 February 2020 (UTC)

As ever, I support Amy's suggested amendments.

Ebruton (talk) 17:40, 28 February 2020 (UTC)

citation -- Juan Carlos (talk) 21:41, 1 March 2020 (UTC)

Amy's suggestion is an excellent start; I would also simplify from 16x16bit to just 16bit multiplier. Vardala also corrects that DSP is used to develop communications and broadcasting equipment, and I don't think we need to repeat "signal processing" in the second phrase. For the general public, "coders" is easier to understand than the more technical "codecs" and also we can drop the "data" qualifier for modems.

NEC (formerly Nippon Electric Company) developed here the first commercially-available software-programmable digital signal processor chip, the μPD7720 in 1980. Its novel bus structure, 250-nsec instruction cycle, and 16-bit multiplier enabled fast finite impulse response filtering and provided true real-time processing for complex systems. It accelerated the development of communication and broadcasting equipment such as voice-band modems and speech coders, and lead to wide use of digital signal processing.

Dave Kemp comments

Dave Kemp - I support this milestone proposal. I suggest the Middle sentence "Its novel...complex systems" be removed.

Re: Dave Kemp comments -- Administrator7 (talk) 22:31, 3 March 2020 (UTC)

Mr. Kemp reminds us all that the Milestones are intended for a public audience that will not comprehend the technical importance of the components of such a device. The technical details rightly belong in the narrative history based on Historical Significance, technical obstacles, or "What features set this work apart from similar achievements". The briefer and simpler the citation, the more significant the accomplishment. To that end, the following citation is suggested to commemorate the accomplishment:

NEC (then Nippon Electric Company) developed here the first software-programmable, digital signal processor chip, the μPD7720. Its commercial availability accelerated the global development of communications and broadcasting equipment such as voice-band data modems and speech codecs, and inaugurated the widespread adoption of digital signal processing of information. (47 words)

It is not necessary to repeat the Milestone's title in the first sentence; therefore "commercial" can be applied in the second sentence where it helps explain the global importance of NEC's work.

The nominators need to provide PDFs of at least the minimum number of required references if they have not emailed them to ieee-history@ieee.org.

Alex Magoun, IEEE History Center

Citation for this Milestone -- Radent (talk) 22:29, 4 March 2020 (UTC)

I believe the citation is totally correct as written. It is also not understandable by most of the public including me. Is a 250-nsec instruction cycle faster than before? The 16X16-bit multiplier seems important, but to the public? Why are these details, although important, necessary in the citation?

I want know how this breakthrough affected us in future applications.

BTW, what is a codec?

Michael Polis - Comments -- Polismpp (talk) 19:16, 7 March 2020 (UTC)

I have no problem with this proposal.