Talk:BESM-6

I find it strange that there is no mention of the 48-bit computer by Control Data Corporation, the 1604. Didn't the Soviets build the BESM-6 to be binary compatible with the 1604? Jim Bowery (talk) 13:46, 2 June 2019 (UTC)

Definitely not. These two machines have the same word size and instruction size, but the instruction set is completely different, the floating-point numbers representation is different, the number of index registers is different, BESM-6 does not have a separate set of integer arithmetic instructions (the programmer should add a special instruction to disable the normalization of floating-point numbers in ALU for their execution). (BESM-6 ALU Description CDC1604 Manual)

I would say that some concepts were taken from IBM Stretch (look ahead, the pipeline, the ability to use intermediate results of calculations without writing them into RAM). A number of publications on Stretch were translated in the USSR and published in 1965 as a separate book.

However, since the USSR bought the CDC-1604 in 1968, the developers of the Fortran translator for BESM-6 were tasked with ensuring compatibility with the CDC-1604 programs.(About software developing for BESM-6) Avivanov76 (talk) 18:07, 21 June 2019 (UTC)

The fastest supercomputer at the time, the CDC 6600, achieved 2 MIPS using one central and ten peripheral processing units.

This is misleading.

The CDC 6600 PPUs did no computing so they had no effect on the MIPS rating. There is no point in mentioning them.

The CDC 6600 used 60-bit words, so each operation did more than a BESM-6 operation on 48-bit words.

The CDC 7600 was "released" (whatever that means) in 1967, before the BESM-6's 1968 release. It was about 10 times faster than the 6600. All this is from the Wikipedia entry. — Preceding unsigned comment added by DHR (talk • contribs) 19:54, 2 July 2020 (UTC)

During the 1975 Apollo-Soyuz Test Project the processing of the space mission telemetry data was accomplished by a new computer complex comprising a BESM-6. Apollo-Soyuz mission’s data processing by soviet scientists finished one half hour earlier than their American colleagues from NASA.

I think this is totally fake story. First of all, it is based only on a book written by the Soviet computer engineer Boris Malinovsky in 1995 — 20 years after the Soyuz-Apollo mission, while Malinovsky himself was 74 year old at that time. There're no other proofs of this especially from the American side. Also this doesn't correspond with BESM-6 characteristics, even according to Russian Wikipedia article, BESM-6 was about 3 times slower than CDC 6600, while American Apollo team used a system of several IBM/360s that were comparable to CDC 7600 and CDC7600 was faster than CDC 6600. So BESM-6 simply couldn't be so dramatically faster than the American IBM/360-based system. And this could be also disproved by the indirect evidence — if the Soviet computer system would have been so faster than the American one, the Soviet propaganda would have used this fact and gathered maximum public attention to this for sure. The second indirect evidence is that USSR de facto closed the BESM project and turned to IBM 360 and DEC clones instead.

1995 (when Malinovsky wrote his book) was a hard time for Russia, when all the power and respect USSR used to have were gone. And that anger and frustration led to many fakes and myths that meant to restore people's honor. Malinovsky personal feelings are perfectly understandable — he was in his 80s and the sphere he worked in laid in ruins, so that was probably his attempt to restore honor to the Soviet computer science. And there's one thing more — in 1980s the Soviet leaders decided to abandon all native computer projects and turn to making IBM and DEC clones. This was a tragedy for computer scientist and engineers - their work was simply thrown in a trash bin, they saw it as a betrayal and got fixed on proving that the systems they developed were better. This affected the computer culture in Russia in general — for many people it became a matter of patriotism and self-respect.

This seems true, just going by the MIPS numbers and the CDC being a wider machine. But the way this is worded in the article needs work. I might take a stab at rewriting it, if I get around to checking the relevant sources. It might also be that the Soviets' code and/or organization was more efficient that the yanks', so that the BESM-6's lower performance mattered less. KetchupSalt (talk) 16:47, 22 November 2021 (UTC)

Transistor

In the lead, it says that this was the USSRs first transistor computer. But the second paragraph on the overview section says "Like its predecessors, the original BESM-6 was transistor-based", meaning that its predecessors were also transistor computers. What is correct? Bubba73 ^{You talkin' to me?} 07:55, 14 November 2023 (UTC)

List of transistorized computers shows the BESM 3 and 4 as transistor computers. Bubba73 ^{You talkin' to me?} 07:57, 14 November 2023 (UTC)

Good question. The first source doesn't seem to mention transistors, at least as best my translation facilities can tell. The second source is in English but has no mention of transistors. Finally I can't access the book cited. ru:БЭСМ doesn't seem to mention transistors for any BESM before BESM-6. Perhaps someone confused the M-20 for the BESM-3? KetchupSalt (talk) 00:09, 15 November 2023 (UTC)

Except for the second sentence in the first paragraph of this article, other information indicates that it was not the first USSR transistor computer. Bubba73 ^{You talkin' to me?} 00:55, 15 November 2023 (UTC)

References