Talk:IBM 7090

"Daisy" anachronism ??

Our "Notable applications" section says "Daisy" was first perform on a 7094 in 1961, but the earlier section says the 7094 was first installed in September 1962. A YouTube video here also claims a 7094 did it, in 1961. I presume a 7090 might have done it, as they were essentially compatible, assuming speed was not a problem. But can anyone resolve the issue? We could remove the inconsistency from the article by changing 7094 in the daisy item to "7090-series", but that still leaves the possibility that the Sept 1962 date is incorrect for the 7094. Wwheaton (talk) 01:30, 8 September 2010 (UTC)

I see now that our IBM 704 article claims the song was done on a 704, not a 7094, in 1962 at Bell Labs, based on this 1997 www source at Bell Labs. Our 7090 article here has a www external link to Decca Records DL 9103 here, but it claims it was done at Bell Labs in 1960, on a 7090. Wwheaton (talk) 01:51, 8 September 2010 (UTC)

I see 1961 for James Tenny in the Decca page, but I don't see a date for Decca recording.

I see conflicting dates for the 704; https://web.archive.org/web/20140401034716/http://www.bell-labs.com/news/1997/march/5/2.html claims 1962 but https://www.loc.gov/item/prn-10-116/national-recording-registry-adds-25/2010-06-23/ claims 1961. Shmuel (Seymour J.) Metz Username:Chatul (talk) 16:45, 22 March 2021 (UTC)

All we know is that "Daisy Bell" was the first song sung by a computer. After that, everything is unknown, including the date and the machine. This is annoying and has been an ongoing problem for more than 20 years. Bell Labs sources says it was the IBM 704 in 1962; but the Decca recording that was released in 1962 (Decca Records DL 9103) says it was made on an IBM 7090. (liner notes) Reddit thinks the discrepancy arises because the 704 used a vocoder and the 7094 was direct synthesis. As if this couldn't get any stranger, according to Clarke himself, he said he heard the song sung on an ILLIAC at Bell Labs in the early 1960s (in Murray Hill, New Jersey), not on an IBM 704 or a IBM 7094, but I don't think Bell Labs used an ILLIAC. Viriditas (talk) 20:05, 4 February 2026 (UTC)→

Right, I'm pretty sure that the ILLIAC stayed at University of Illinois. Bubba73 ^{You talkin' to me?} 20:19, 4 February 2026 (UTC)

Yeah, I know, it's weird that he said it. His comments appear in "Foreword: The Birth of HAL" in the book HAL's Legacy: 2001's Computer as Dream and Reality (1997), on p. xiv. One alternate interpretation is that he was using the term as a proprietary eponym to refer to supercomputers in general, just like we use "Google it" today. Viriditas (talk) 20:43, 4 February 2026 (UTC)

The movie did say that HAL was from the computer lab in Urbana, Illinois, which IS where the ILLIAC was. Bubba73 ^{You talkin' to me?} 21:09, 4 February 2026 (UTC)

Yeppers. Anyway, I believe I may have figured out the entire problem discussed here and below. I'm putting something together to resolve it finally, once and for all, but I may not have anything finalized until tonight or tomorrow. Viriditas (talk) 21:16, 4 February 2026 (UTC)

Note, it turns out Bell Labs may have used an ILLIAC on their 1960 recording of computer music. Very odd. See the bottom comment on this page. Viriditas (talk) 23:04, 4 February 2026 (UTC)

Never mind. It looks like they just published a song that was recorded on an ILLIAC, not that they used it in their work. Viriditas (talk) 23:11, 4 February 2026 (UTC)

There was only one ILLIAC I and only one ILLIAC II. They both resided in Urbana, Illinois. The ILLIAC II wasn't functional until 1962, but the 1960 recording could have been made on ILLIAC I. Bubba73 ^{You talkin' to me?} 01:01, 5 February 2026 (UTC)

See this video: https://www.youtube.com/watch?v=l381_ho8KR8 - it plays "Daisy Bell", Then a little after 27:00, there is an explanation of how it was done. I think it implies a 7090 - it says work was started on a 704. Bubba73 ^{You talkin' to me?} 04:26, 12 February 2026 (UTC)

Nomenclature in IBM 7094 registers table

@Loadmaster: The IBM 7094 registers table in § IBM 7094 has some issues:

1. The bit numbering does not match that in the 7090 and 7094 manuals
2. The S, Q and P bits of AC are not labelled.
3. The manuals use the term instruction counter (IC), not PC.

I'm not sure what the best way is to lay out the table: IBM labels bits as

AC

S, Q, P, 1...35

IC

3...17

Index register

3...17

MQ

S, 1..35

SI

0...35

Storage, including instructions

S, 1..35

For some instructions

Prefix

S, 1, 2

Decrement

3...17

Tag

18...20

Address

21..35

The numbering of the bits in the IC and index registers matches the numbering of the bits in the decrement. The same issues exist for table IBM 704x/709x registers in IBM 700/7000 series § Later scientific architecture (704/709/7090/7094). One possible arrangement is

IBM 7094 registers

Data registers S Q P 1 2 3 ... 17 18 20 21 ... 35 (bit position) Accumulator AC S   Multiplier/Quotient MQ 0 1 2 3 ... 17 18 20 21 ... 35 (bit position)   Sense Indicators SI Index registers 3 ... 17 (bit position)   Index Register 1   XR1   Index Register 2   XR2   Index Register 3   XR1   Index Register 4   XR4   Index Register 5   XR5   Index Register 6   XR6   Index Register 7   XR7 Instruction counter 3 ... 17 (bit position)   Instruction Counter   IC

but is an alternative preferable? -- Shmuel (Seymour J.) Metz Username:Chatul (talk) -- revised 17:57, 15 July 2024 (UTC)07:41, 14 July 2024 (UTC)

These are good suggestions. I had forgotten that IBM labeled the bits backward from the now-common convention (where bit n corresponds to value 2ⁿ within a word). The diagram looks good, the only changes I would make is to include two digits for all the bit numbers (i.e., bit 1 is displayed as ⁰₁), for consistency with other CPU infoboxes for other processors. Note: I went ahead and applied this change to the article. (A worthy goal is to have a register diagram for every CPU article in WP.)

The problem remains as to how to best align the shorter registers with the longer ones. On the one hand. the reverse bit numbering makes this somewhat problematic. On the other hand, showing the registers and their relative sizes is probably more important to the article than their visual alignment. Whatever the result, the same changes should be applied to the infobox in the IBM 709 article. — Loadmaster (talk) 15:51, 28 July 2024 (UTC)

I'm not sure whether the term reversed is appropriate in general, and certainly not for IBM hardware. Even after the 7000 series, IBM numbered bits from 0 on the S/360 through z Systems, on S/38 through iSeries, and on PC/RT through IBM Power Systems.

I couldn't find a 709 manual on bitsavers. Can someone confirm that it used the same numbering as the 7090/7094? -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 13:31, 29 July 2024 (UTC)

I could not either: So, I think there is one on archive.org, and another on a very weird place:

https://justine.lol/sectorlisp/ibm709.pdf "PROGRAMMING AND CODING THE IBM 709-7090-7094 COMPUTERS" (1963)

and

https://archive.computerhistory.org/resources/text/Fortran/102663112.05.01.acc.pdf "Reference manual 709 7090 Fortran programing system" (Jan 1961)

There was one on bitsavers:

http://www.bitsavers.org/pdf/ibm/7090/D22-6508-2_709_7090_General_Information_Manual_Aug1960.pdf ( Aug '60)

I remember the programming forms numbering the bits from 00 to 35,

but the documentation ... oh my... had the memory locations in Octal! 135.180.103.252 (talk) 04:39, 23 August 2024 (UTC)

The first is not IBM's.

The second one is about FORTRAN, not about the 709, 7090 or 7094.

The third one matches the 7090 and 7094 principles of operation.

Real programmers do it in octal; cue battle to the death between ones' complement and two's complement, with sign magnitude and decimal computers looking on in amusement. ;-) -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 13:01, 23 August 2024 (UTC)

You would not have know about *any* of that, unless you were there. Thanks for the good laugh. Really. Now I need an Octal to IBM BCD chart...and...

From the Console numbering, it was not a leading zero, or super/sub script. 158.51.81.86 (talk) 21:48, 23 August 2024 (UTC)

Well, I've never done ternary ;-)

There is a slight anomaly in the 7090 principles of operations manual; although for words in storage the bit label "S" is inline with the labels 1-35, for the AC and MQ registers the sign is shown in a box above and to the left of Q, P and 1-35. The 7094 PoOps has S inline for AC and MQ. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 12:38, 27 August 2024 (UTC)

Nice job on the CPU register infobox. Now can someone apply the same fixes to the register infobox in the IBM 709 article? In general, it would be nice to see register infoboxes for all CPUs, especially now that we have a nice format and bit-numbering layout for IBM CPUs.

My preference for the bit position numbers is the ¹₂ format instead of the 12 form, on the theory that it is slightly more horizontally compact. It's also the style used in many other existing CPU infoboxes (8080, i386, 68000, etc.). — Loadmaster (talk) 01:25, 24 September 2024 (UTC)

Now can someone apply the same fixes to the register infobox in the IBM 709 article?

I can't find a 709 principles of operation. Can anybody confirm that the 709 has the same 36-bit SI register as the 709?

My preference for the bit position numbers is the ¹₂ format instead of the 12 form, on the theory that it is slightly more horizontally compact.

That's the way that I originally wrote it, but 135.180.103.252 (talk · contribs · IP contribs · WHOIS) changed it in permalink/1242250376; the changed version more closely matches the manual. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 13:50, 25 September 2024 (UTC)

Rendering of bit numbers in register table

When I originally revised the register table, I marked up the bit numbers with <sup>...</sup> and <sub>...</sub>, a style used for tables in other articles. 135.180.103.252 (talk · contribs · IP contribs · WHOIS) changed it in permalink/1242250376 to drop leading zeros and to put the entire number in <sup>...</sup>, a style more aligned with the IBM documentation. I believe that a case can be made for any of the "⁰₁"-"³₅", "₁"-"³₅" and "¹"-"³⁵" styles, so I'm soliciting opinions on which is best.

A secondary issue linked to the first is how to render the special bit labels S, Q and P. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 14:11, 25 September 2024 (UTC)

Based on the above #Nomenclature in IBM 7094 registers table discussion, it would appear you have wide support for your original ⁰₁ format. As far as the original format not matching the manual, the bits in the other processor manuals do not match that format either. RastaKins (talk) 18:18, 10 February 2026 (UTC)

Show or suppress leading zeros? Does anybody know if the 709 had sense indicators? -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 09:16, 11 February 2026 (UTC)

I'd like to post an RFC to make all of these consistent:

1. IBM 704#Registers
2. IBM 709#Registers
3. IBM 7090#Registers
4. IBM 700/7000 series#Later scientific architecture (704/709/7090/7094)

This would include bit numbering, number style and instruction counter nomenclature.

Should that start with {{rfc|sci}} and be in Wikipedia:Requests for comment/Maths, science, and technology? -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 14:48, 15 February 2026 (UTC)

RFC seems to be used for dispute resolution. I don't see much pushpack here against your idea. How about WP:BEBOLD and go ahead and implement your bit numbering. And while you're at it, determine if the 7090 instruction counter should be bits 21-35. RastaKins (talk) 15:29, 15 February 2026 (UTC)

Recent Daisy Belle edit

Recent edits, e.g., permalink/1307995768, corrected an anachronism by changing 7094 to 7090. However, the article still conflicts with sources giving primacy to the earlier 704 playing Daisy Belle.

Probably the IBM 704 article should cite one or more of these. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 12:47, 27 August 2025 (UTC) -- Revised 13:23, 27 August 2025 (UTC) -- 16:32, 27 August 2025 (UTC)

Some good research, Shmuel. I think some of the confusion might be attributed to the 704 being the first platform to implement the program, MUSIC. But even the MUSIC-N article has an obvious anachronism: It states "MUSIC II was developed by Mathews on an IBM 7094 at Bell Labs in 1958." RastaKins (talk) 13:49, 27 August 2025 (UTC)

@RastaKins: Late enough for the 709, but not for the 7090 :-(

The guinness article mentions a paper, but gives no detail. -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 16:39, 27 August 2025 (UTC)

This doesn't solve anything. See the above. We still have conflicting sources for the 704, the 7090, and the 7094. We do know that the Decca recording was made on the 7090. Viriditas (talk) 20:58, 4 February 2026 (UTC)

Take a look at this: Viriditas (talk) 21:25, 4 February 2026 (UTC)

That's two independent and verified points indicating the 7090. The initial demonstration in 1961 and the Decca recording released in 1962 (possibly recorded earlier). I'm working on establishing several more. Viriditas (talk) 21:29, 4 February 2026 (UTC)

Kelly & Gerstman demonstrated music on the 7090 at the Acoustical Society in Philadelphia in 1961, but the news coverage doesn't specify that they played "Daisy Bell" specifically. However, it is confirmed that Bell Labs released two different recordings of computer music and voices, one in 1960, which does not appear to contain "Daisy Bell", and a second edition in 1962 which was released under the Decca label. It is that later release that contains a recording of "Daisy Bell" on the 7090 as the third track by Max Mathews, a "Bicycle Built for Two". Something interesting: the original 1960 release contains a 24 page booklet of liner notes that could further inform this topic. Additional supporting research on the 7090 is confirmed in the following papers: Kelly, J. L., Jr., and Gerstman, L. J. (1961). "An artificial talker driven from a phonetic input". J. Acoust. Soc. Am. 33: 835A; Kelly, J. L. Jr. and Lochbaum, C. C. (1962). "Speech Synthesis," Proceedings of the Speech Communications Seminar, Royal Institute of Technology (KTH), Stockholm, Sweden, September 1962. The paper is available to read in the compilation Speech Synthesis (benchmark Papers In Acoustics), edited by James L. Flanagan and Lawrence R. Rabiner. All of the work in both papers indicates it was done on an IBM 7090. None of them mention using the 704 for programming the voice. I did find some sources saying the voice was programmed on the 704, but could not reliably confirm any of them. The Guiness Book of World Records contains an unusual entry that I've also been unable to confirm. To summarize, contrary to what many sources say, there is zero reliable evidence showing that "Daisy Bell" was performed live before 1962 (but may have been recorded earlier, we don't know). In several different places, it is alleged that Arthur C. Clarke heard it for the first time in 1962 at Bell Labs, although Clarke's own recollections differ on this. For example, Clarke said he heard it on an ILLIAC at Bell Labs, which of course makes no sense. Strangely, the 1960 first edition release of Music From Mathematics from Bell Labs did contain track number eight by Lejaren Hiller, "Excerpt From Illiac Suite For String Quartet". See for example, Illiac Suite. Viriditas (talk) 23:07, 4 February 2026 (UTC)

Okay, I see the National Recording Registry indicates it was recorded in 1961 via the LOC. This seems like it's open and shut now. Viriditas (talk) 23:42, 4 February 2026 (UTC)

Update: Just confirmed that "Daisy" was performed in public in Philadelphia in 1961 on an IBM 7090 at the 61st annual Acoustical Society of America meeting. The original article about this failed to mention the specific song, but a second article published by the Australian Associated Press about the same event a month later in The Sydney Morning Herald mentions the song. I've added the sources to the article on the song if anyone wants to add them here. Viriditas (talk) 00:20, 6 February 2026 (UTC)

Good work! Bubba73 ^{You talkin' to me?} 01:04, 6 February 2026 (UTC)

I do not believe Daisy Bell could be performed live. To allow plenty of compute time, the audio waveforms were certainly spooled to digital tape and later played back in real-time. The 7-track tapes of the era could be read at 15,000 characters per second, plenty-fast for real-time playback. Using this method, there would be no limitation on the complexity of the algorithm. RastaKins (talk) 15:59, 6 February 2026 (UTC)

The 7-track tape drives on the 7090 did not include a drive that slow. The available drives ran at 22,500 C/S, 62,550 C/S and 90,000 C/S. Were you thinking of the 727, used on the 7xx machines? -- Shmuel (Seymour J.) Metz Username:Chatul (talk) 20:32, 6 February 2026 (UTC)

There are many stories, covering many years, of picking up the RF emissions of processors on AM radio receivers. I suspect that would be done in program loops. It wouldn't surprise me to know that was done on the 704, 709, and 7090. The output of an AM radio tuner could be direct connected to a tape record, to record generated music. One might be able, given the cycle time and instruction cycle counts, to figure out what those might have been. (And yes that would be WP:OR, but that is allowed in talk.) Gah4 (talk) 23:52, 6 February 2026 (UTC)

Several of the tracks on the Decca recordings are far more advanced than a simple buzz that might be heard through an AM radio. For the era, these recordings are quite impressive. Their voicing is far more advanced than the Processor Technology Music System that ran on 8080 processors in the 1970s. A 7090 tape drive that was too-fast could still be used for real-time music playback by padding with null records. RastaKins (talk) 16:42, 7 February 2026 (UTC)

Extended content

References