The Soviet 1801VM2 LSI-11 Processor | Hacker News Reader

The Soviet 1801VM2 LSI-11 Processor

cpushack.com

145 pointspicture92 comments

The CMOS military version ran at 5 MHz and consumed 25 milliwatts which is pretty amazing. A CMOS 8088 would take about 10 times that.

Source: https://www.renesas.com/us/en/products/space-harsh-environme...

They should have copied ARM, but who knew at the time what could be done with 25k transistors?

https://www.theregister.com/2012/05/03/unsung_heroes_of_tech...

The West normally lacks understanding what hard work, innovation and intelligence is a universal value. During my years at Stanford I had the opportunity to learn about the North Korea nuclear program and how the West never imagine the advancement they produced with very limited resources and without Russian help. Couple of professors visited the nuclear facilities as part of the UN investigation and were shocked with the intelligence, and innovation of Korean engineers. The rest is history and we have now a North Korea with nuclear capabilities

If you want a world where nations aren’t scrambling to acquire nuclear weapons, then maybe we ought to reign in a hegemonic superpower which has been invading and bombing countries since the end of WW2.

It's simpler than that, really: ban all nuclear weapons and bomb the shit out of anyone who attempts to create them.

This two-tired world, where the rich guys and their cronies have nuclear weapons, while the second grade countries must submit to their wishes, will clearly leave many nations wanting to balance the scale. History has shown that all countries with security problems develop nuclear weapons as soon as they get to a suficient wealth level.

Nuclear weapons are the only thing preventing WW3, WW4 and so on. We're living in an extremely peaceful world, thanks to nuclear weapons.

survivor bias; also cf Leibnitz' "best of possible worlds"

What nuclear weapons are fantastic at preventing are the regular border skirmishes that counties have fought since the invention of territory. It used to be that counties were constantly gaining or losing land to their neighbors and sometimes even ceasing to exist or appearing out of an existing country. Maps would be constantly in flux. These days situations like that are rare. The Russian annexation of Crimea and the breakup of Yugoslavia being the notable exceptions. All of which did not involve nuclear armed powers (at least on the losing side).

Pakistan and India would like to disagree.

Have the maps in Kashmir region changed since the late 90s?

Banning nuclear weapons would be suicidal with how aggressive NATO nations are in the global south. Having nuclear weapons is like having a nice sharp dagger on the US throat.

If you really want to make the world a saver place the US should give up its first strike policy. That will calm down the escalation calculation China or Russia have to make.

Given the behavior of both, Russia and China, i can kinda understand why the US doesn't want to give up that policy.

IDK, I believe a first-strike policy is inherently destabilizing. Never quite understood the idea. You don't have to abolish nukes to cancel such a policy.

Russia and China are the only two great powers seeking to conquer and annex territory in the world today.

China just recently annexed territory the size of France in the South China Sea. They annexed Hong Kong. Next they want Taiwan. They eagerly want several of Japan's island territories. They very obviously won't stop there, as their military power increases dramatically over the next 10-20 years; they're just getting warmed up on territorial conquest in Asia.

Russia is dead set on destroying Ukraine as it exists today and will proceed to, at a minimum, annex to the Dnieper River. They already annexed Crimea and a large chunk of East Ukraine. They'll attempt to take all of the Black Sea territory of Ukraine, further impoverishing and cutting off the remaining chunk of broken Ukraine post annexation. Further it has its eyes on taking Moldova, either directly or by forced alliance. Taking Ukraine's Black Sea territory will specifically make Moldova easy to conquer; there is no scenario where they don't pursue that combo. It has begun the (likely longer term) process of annexing Belarus, Lukashenko has capitulated into Putin's arms. Nearly any territory in the region that is not in NATO is up for grabs for conquest as far as Russia is concerned. Russia realistically destroyed Georgia, annexing part of the nation, and it's likely to eventually get around to finishing the rest. It would be a big task for the weaker modern Russia, however they'd also really like to figure out how to sink their claws into Kazakhstan and reclaim some of that territory as part of the empire. For Russia ideally there would be some convenient chaos, civil war in Kazakhstan, and Russia would then use border security as an excuse to de facto invade and claim territory that they'd never give back.

Meanwhile, the US has spent most of the post WW2 era trying to maintain global border security, including in Europe, the Middle East, Africa, Latin America and Europe. The US was in Vietnam? It sure was, trying to protect the South from a conquering North; when the US left, the South promptly got conquered (and yet the US is the villain in that story? yeah right). The same fate awaited South Korea, except the US saved them from the conquering North. How much territory did the US annex in Iraq? Zero. How much territory did the US annex in Afghanistan? Zero. It didn't try to take those nations, their territory, or their natural resources.

When was the last time the US tried to conquer and annex Canada or Mexico? Canada sits right next to the world's primary superpower, with an almost entirely unguarded border. Check out Canada's tiny military spending some time. Ask India how their border context is going with China these days; everyone that shares a border with Russia or China is terrified of their territorial ambitions.

While the US has been a superpower during the post WW2 years, how many nations has it annexed? Right.

>Russia and China are the only two great powers seeking to conquer and annex territory in the world today.

Well, I would respectfully disagree with the "conquer" part. Since the Afghanistan and Iraq wars, that sounds really hollow.

Oh, but that’s not “conquering”. That’s “liberating”, the same way the US sponsored puppet violent dictatorships in Latin America that “saved” those countries from becoming “communist dictatorships”.

>The US was in Vietnam? It sure was, trying to protect the South from a conquering North

Sorry but you are dead wrong about the Ukraine, and especially about Vietnam, the US feared the rise of Communism and by that loosing control over Vietnam.

You're pretty wrong about Vietnam. The US got involved to avoid the communists, the most legitimate entity in Vietnam, the one that fought the Japanese and French and which liberated the country from taking control. If the US doesn't intervene, there is never a South Vietnam.

What you described is roughly how things went down in Korea, so maybe you just confused the two.

US sponsored coups and wars, often in favour of downright terrible people (like Pinochet) weren't for "global border security", but for power, resource and against anything vaguely leftist.

Nobody has to make any escalation calculation, that's bad reasoning.

The know how is hardly a secret. Most countries can build a nuclear weapon provided they want to, they are let to do it and they have the raw materials.

But you are right, innovation is not a monopoly of the West.

During the Soviet times USSR had many technological advances, their population was quite literate in sciences, schools were very good and USSR had many teams winning math and physics Olympics.

Having great schools resulted in a large number of scientists and researchers.

The situation repeats with China. They invested a lot in scientific education and it is starting to pay off. Meanwhile, in the West, higher education means discussing politics and being woke.

While the core point holds, that scientific literacy pays off, more discrimination is required about the other side of liberal arts:

-- the full realm of human development should be cultivated, and

-- there is a crisis in some broad instances of higher education.

For example, with reference to the parent: you /should/ discuss politics: only, productively.

You do not want your society to be well trained in engineering but underdeveloped in other human qualities fundamental in civilization.

> For example, with reference to the parent: you /should/ discuss politics: only, productively.

I would argue that the scope of education is to improve and spread knowledge.

Politics might be important for society but it is not within the scope of education. Unless you make education about politics and forget knowledge.

> the scope of education is to improve and spread knowledge. Politics might be important for society but it is not within the scope of education

Why is knowledge of politics not knowledge?

>Why is knowledge of politics not knowledge?

It is. But is not the only knowledge needed by a country. Or the most important knowledge.

The purpose of education is the mental refinement of the individual, it scope is dealing with knowledge to that purpose. You deal with knowledge to provide a form of wisdom. Notion and ability go together through education. (Notion without ability would be a dead encyclopedia.)

Political science (say, the comparative advantages of an organizational, or administrative, or voting system vis-a-vis philosophical principles and quantitative methods etc.) is a discipline. It is surely part of education.

When individuals treat political discussions like "hooligans" could treat "Barcelona vs. Arsenal", that is a fault in mental refinement, hence also in education.

>Political science (say, the comparative advantages of an organizational, or administrative, or voting system vis-a-vis philosophical principles and quantitative methods etc.) is a discipline.

Studying politics doesn't mean doing politics. And studying politics is best done solely as a discipline, not instead of other disciplines.

Yes. You wrote «discussing politics»: the expression leaves interpretation of "discussing politics with a skewed bias towards that discipline" - still research, investigation, inquire, study, learning -, it is not necessarily read as "doing politics" ("party taken, now let us be active, or loud, or hooligans").

Now: in theory, if your students spent time being loud instead of studying, that should reflect on the marks, should not it? They finally should not pass if doing politics replaced studying. But it is reported of prestigious Athenaeums in USA, and it is also seen and told in Europe, that the "feedback" from the institutions is being lower and lower: even the worse will get the degree. There are horror stories directly told by professors. So, there is a confirmed crisis in education.

But you want your citizens to be fully developed, not """able to count and unable to read""".

Just look at the incredible amount of knowledge around working Titanium that came from Russia, even to do things that the West had decided were impossible.

They may hot have had Russian help, but they did get help from Pakistan.

"By 2002, Pakistan had admitted that North Korea had gained access to Pakistan's nuclear technology in the late 1990s."

https://foreignpolicy.com/2021/10/11/aq-khan-pakistan-north-...

ARM1 shipped in 1985, and it wasn't immediately obvious at the time that it was the way forward. Meanwhile, this LSI-11 clone dates to 1982. Cloning ARM three years in advance would have been a neat trick!

Precisely. Initial ARM work started in '83 and it wasn't exactly well published until the revealing of the Acorn Archimedes ('87), the BBC Co-processor ('86) was spotted rarely in the field (mostly due to its price).

But: nice to reflect on that back in the 80's designing a new CPU could be done with a very small team and modest funding in 2 years time going from drawing board to first silicon (assuming that kind of talent, of course).

ARM only properly started in '83, with first working silicon in '85, likely by then the predecessor to this project was already well underway. Besides, I think that they should get credit for making their own design rather than cloning someone else's.

Considering it was made decades after first Western CMOS, it's no big wonder

That was the only CPU certified for military use in USSR. I know guy who worked on Navy Surface to surface missile, and they had to put 32 cores to achieve required performance. Even with low consumption, control module overheated and melted down in 10 minutes. Annoying for testing, but doesn't matter for real rocket. Flight time was much less.

Love it. Know your requirements. (Well, love it from a very specific angle...)

> These 1806VM2 are still being made by Angstrem, if you need to build a PDP-11 computer to run Tetris on, or repair a Buran shuttle you may have laying around.

wow!

PDP11's were pretty great machines. Using RSTS/E (resource sharing time sharing... Eeeee!) you could really stretch things out.

The process memory was limited to 32kb (maybe 64kb... can't recall) and the OS's executable system supported seamless overlays. Basically as needed the OS would swap in different code segments (min 4k?). With careful planning you could build some pretty big applications... so long as you could get descent separation. You *could* swap overlays over and over and over... but this would kill any sort of performance... best to swap in a "AR" overlay, and when the user switches to "GL" swap that overlay in.

The machine language was also pretty nice. Several addressing modes were program counter relative which was amazing (eg. JMP to 1500 words from current PC) . When I first learned that that windows loads an executable and then goes through and fixes up memory addresses so JMP instructions (et al) would work I could hardly believe it. If only the x86's had decent support for a PC relative addressing mode :)

And the Digital Equipment Corp documentation! Glorious. A wall of shelves, 10 feet long, 7 feet tall, filled with 3 inch binders explaining everything in detail! Shivers!!

Microsoft's online documentation is the closest thing in quality to the old DEC stuff.

Jumps have always been PC relative on x86. It's calls that can optionally not be. It's mainly the indirect calls that are troublesome.

DEC documentation was something else though, agreed. My KD11-A came with everything you'd expect to be documented these, plus full schematics, plus microcode listing in source and flow chart form.

PDP-11's (LSI-11's) were originally limited to 32k 16-bit words (64kb) but were exanded to 18bit addressing allowing up to 256Kb. The PC is only 16 bit though so 64k of code directly addressable at any one time (ignoring the reserved area for memory mapped hardware). The MMU was pretty simple so easy enough to access the extra memory without relying on the OS (simpler under RT-11 than RSTS/E).

I still have shelves of DEC documentation (and a few PDP-11's).

Thinking about a PiDP-11-like project but with an actual PDP-11 inside it now…

interesting. can't find a place to get them on first search though.

Man, I wonder what that shuttle went for when the Soviet Union collapsed...

For the retro/breadboard builder, this is sort of interesting.

PDP-11 clones are nice but I've always wanted more information on the Russian balanced ternary designs, how far they got going along that path.

https://en.wikipedia.org/wiki/Balanced_ternary

Apparently pretty far, I've found various sources for why funding was cut from "The west was going binary so that was the order of the day" to "The space race". What was interesting was the hardware description I could find wasn't vacuum tubes but rather ICs and allegedly some sort of logic based on "Ferrite cores and diode ladders" which was apparently a major thing as it was vastly cheaper than transistors (super expensive at the time) or vacuum tubes (expensive and unreliable). How much of this is true or just what's on the russian computing museum's website I don't know. But the concepts aren't dead:

https://hackaday.com/2014/10/07/hackaday-10th-anniversary-no...

Maybe we will see some of that in a future version of Thunderbolt with PAM-3 encoding:

https://www.anandtech.com/Show/Index/16858?cPage=3&all=False...

OMG. Sweet memories. I do not really play games but I spent countless amount of hours back in USSR Academy of Science playing Tetris on that DVK-1. Tetris to me is a form of meditation. My fingers know how to play and my mind can just float aimlessly ;) This was a pleasant payback as that DVK-1 had drank a healthy dollop of my blood when suddenly resetting in the middle of me writing a program when I was just about to save it.

Wow, the die shot! It looks like an eerie caricature of then-contemporary chip designs. I have seen blocky hand layouts, and modern (presumably) auto layouts that look like bacteria cultures. But I have never seen warped and slanted blocks.

I think it’s a mask diagram not a die shot, in any case these features are common for that era especially for the metal layers / ic traces look at die shots of a Z80 and it’s clones.

Ah, yes. What do I know... but this Russian CPU still looks much more "weird" than a Z80! The Z80 has 90 and 45 degree angles, this thing also has like 10 degree angles, and wavy structures, and big curves, and triangles.

I'm curious to see where you see the odd angles. When I zoom in sufficiently all the shapes seem to resolve to segments of 90 and 45 degree angles.

Well OK, there is only one triangle - in the upper left corner. The other things I mostly see in the center and right of center. Especially in the blue lines - power and clock or so?

Here is an 8080 the topology isn’t that much different https://lh3.googleusercontent.com/-Z4RzZ24yNqU/WF4ffuFFN4I/A...

I could not find it on ddg/google right now, but I remember seeing some absolutely beautiful plant/vine-like trace routes done by the soviet auto-routing tools of past ages. I feel a bit sad that the 45degree aesthetic is the only one that survived today.

I also thought 45 degrees is used to reduce ESD, provide the best use of die space, and reduce stress on the metal with high current. Maybe all of that is a myth? Any other angle doesn't have the same benefits.

Any-angle routing may actually be better for use of space and reducing stress. I sometimes get more traces packed around a bend with any-angle. And most polymide PCBs are any-angle or at least super rounded with teardrops, having sharp transitions of any type would act as weak points.[1] ESD really doesn't make sense, maybe you meant EMI or EMC? There's nothing particularly special about 45 degrees other than the ease of modeling

[1]: Some macbook trackpad, note the polymide/flex substrate and rounded tracks. This isn't fully any-angle like what TopoR does though https://commons.wikimedia.org/wiki/File:Apple_MacBook_Pro,_m...

https://en.wikipedia.org/wiki/TopoR

Parent is downvoted for only providing a link, but : TopoR is a free-angle PCB autorouter ( it has other tricks too ).

It is of Russian origin and commercial software!

It is not a question of aesthetic.

In old integrated circuit manufacturing processes, with large minimum dimensions, it was possible to draw arbitrary curves without problems.

Some of the software tools were limited to straight lines or even to straight lines only at certain angles, just because the limitation simplified the algorithms, but if one really wanted, it was possible to draw anything on the masks.

On modern manufacturing processes, with extremely small minimum sizes, it has become more and more difficult to predict which will be the differences in geometry between what you draw on the mask and what will be made on the surface of the semiconductor wafer, after photolithography and etching of the layer that must be patterned.

In order to make the photolithography and etching processes more predictable, the design rules have become more and more restrictive, e.g. by allowing only vertical, horizontal and 45 degree straight lines, or only vertical and horizontal lines, or even on some layers only traces parallel to a certain direction.

Moreover there are a lot of extra constraints, e.g. the need to place dummy patterns around those that are useful, because the same pattern in a different neighborhood will be distorted in a different way, which could be hard to predict.

I believe this Bad Apple Demo ran on a similar Russian PDP-11 clone.

https://youtu.be/8Q1vN51o-Dg

HN thread: https://news.ycombinator.com/item?id=17965269

The PDP-11 ISA is one of the cleanest and most orthogonal there ever was, and served as the basis for the 68K family's design. This would have made for an incredibly fun micro.

There was a modern greenfield Russian CPU that was in the news a while back. Did it get abandoned or eaten by a GRU?

Probably the former. In either case, it would be interesting to see computer chips evolve independently and try to compete.

There's the various Elbrus chips that are only really used in defense applications. The best stuff Russia can make domestically is on 90nm process tech:

https://en.m.wikipedia.org/wiki/Elbrus-2S%2B#Technology

They have some stuff done on TSMC 28nm too:

https://www.anandtech.com/show/15823/russias-elbrus-8cb-micr...

We had a couple of БК-0010 computers in a club back in late 80th which were also based on 1801ВМx chipset. Those were quite popular home computers in USSR.

https://en.wikipedia.org/wiki/Electronika_BK

Reply on news.ycombinator.com