The Soviet 1801VM2 LSI-11 Processor (2021) | Hacker News Reader

The Soviet 1801VM2 LSI-11 Processor (2021)

cpushack.com

100 pointsthrowup23846 comments

I happen to have three of those: one standalone NOS chip and two on Q-Bus SBCs. One however has a crack in its cerdip package. Here's what one of the boards looks like:

http://blog.funcall.org/hadrware%20vintage%20pdp/2017/03/14/...

Boards like these were used both in DVK systems and in different CNC controllers.

I used DVK for a while and yes good chunk of that time was used to play Tetris ;)

Yep they weren't exactly uncommon but generally were installed at various institutions and professional schools: there was no legal way to obtain one privately. They were also (due to peculiarities of militarized planned economy) ridiculously overbuilt and costly to make. It's rather hard to find a complete DVK on the market now as most of them were reprocessed for their very substantial content of precious metals.

A factory in my hometown used to have CNC mills with these systems as controllers and they were regularly pilfered by the socialist workers. The backplane had insane amounts of gold plating (grams!) on edge connectors and boards themselves used platinum/palladium capacitors for bypass! These were stripped on the SBC specimen in my blog post above too.

Arh.. nostalgia. I learnt to programme in PDP-11 assembly on ДВК-1 and БК-0010 somewhere in 89-90. That was my second assembly though. The first one was MOS 6502 on Atari 800s, they were quite popular in USSR in late 80th.Funny thing is that you still can purchase КМ1801ВМ3 chips in Russia for less than a $1 in bulk. I once considered building a ДВК clone to play with and to teach my kids some proper assembly, but then found this DVK-FPGA project on Github: https://github.com/forth32/dvk-fpga

You should resell them on eBay. IM6100s (PDP-8 CPU on a chip) are going for $50 / each.

Possibly they wouldn't fit (metric pitch at 2.5 mm spacing for many Russian made chips).

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

also pdp-11 and pdp-8 aren't compatible

> I once considered building a ДВК clone to play with and to teach my kids some proper assembly

I actually built a custom board with a K1801VM1 a while back to teach my son programming basics from the ground up. It's slow as molasses, but great at this purpose. The PDP-11 ISA is simple and comprehensible, as well as the 1801 microarchitecture. Underclocking and a few LEDs in the key points of the board can easily illustrate the entire process; it's very direct and you can almost physically see it "thinking" under the hood. We used that board to drive our Christmas lights and for simple games before moving to something more high-level.

This sounds interesting. Are your schematics available ?

I used to write all my code for this CPU in plain machine codes. Granted they weee very easy to remember- in octal form.

Eg 10102 - mov R1, R2

Very elegant and human readable machine code.

Btw, the same processor was used in the MK85 “calculator” which in essence was a mobile computer (ca 1990!). I owned one not knowing that fact, and was so amazed to realize it was actually a 1801vm1/vm2 architecture when I unsoldered ROM from it, connected to my parallel port and read its content - it had that familiar code there!!

wow, that's awesome! i've only ever looked at pdp-11 machine code a little bit

There was this joke at Soviet time, mimicking the usual Soviet propaganda: Soviet microprocessors - the most large microprocessors in the world!

They made many clones of Western designs but had also number of original designs https://en.wikipedia.org/wiki/List_of_Soviet_microprocessors

I remember reading that some version of the MicroVAX CPU had etched into it, in broken Russian, something approximating "VAX: When you care enough to steal the very best". I'm not sure how many Soviet engineers would have been familiar with the Hallmark commercials to get the joke, though.

I think they would have gotten the point without it, given that they were into stealing its design - of course the subtle point might have been not familiar for them.

Yes, on CVAX: https://micro.magnet.fsu.edu/creatures/pages/russians.html

The broken Russian reads more like an advertisement to buy the real thing. Until you have supplied that quote I had no idea, even though I saw this picture.

Hm, indeed, can be interpreted this way, if you imagine commas at right places.

If you care, enough to steal, the real best.

(The word "настоящий" used as the translation of "very" means "real". For the rest we can leave the original english words and with commas even have some similar broken language effect)

The Russian is too broken to be comprehensible, so the burn doesn't work as well as it could. The equivalent broken English would be something like: "SVAX: when you whaaant enough steal genuine best". Yes, it's just as difficult to parse the Russian.

DEC celebrated this in their "VMS at 20" anniversary book in 1997: https://vtda.org/docs/computing/DEC/VMS/VMSat20Book.pdf

... with 16 pins and two carrying handles!

Systime, a UK based DEC var and terminal maker was busted copying VAX boards to on-sell to the SU when DEC couldn't under ITAR restrictions. They sold clones with DEC boards and decided to up the profit by cloning the DEC boards.

Systime was doing fine selling VT100 compatible terminals but went seeking bigger bucks. I think the VAX was a sweetspot for them while it lasted. They went bust in the end.

Their headquarters in Leeds was a futuristic (for the 1980s) building with magstripe guided parts moving robots, a fantastic works cafeteria, the owners helicopter parked outside, and featured in a nuclear missile thriller TV series called "edge of darkness" -I worked at Leeds Uni on a systime funded research project and spent time there. Nice people in the research labs, smart, motivated.

I've heard Latvia had a fisherman kolkhoz dedicated to smuggling dual purpose tech from Sweden (presumably with their benevolent neutrality).

It interesting, because VAX had never been used much in the USSR. The main architectures were PDP-derived or IBM-derived.

Probably sold mostly to the sphere, not the core. Systime might have been in the pdp-11 business too but I was there working Vax, and the story was told to me about Vax (by systime employees I might add)

I remember using Soviet-made VAX clone from СМ ЭВМ series:

https://en.m.wikipedia.org/wiki/SM_EVM

Your list misses SM-1700 ("СМ-1700"), a full clone of VAX-11/730. But all the VAX-compatible computers were produced only during the last several years of the USSR, so just a few thousand had been built in total.

well, except one of the most famous vaxen of all time, kremvax

The table on that page says some chips used risc-v architecture. But made by the Soviet Union? Risc-v was introduced 10 years ago. What on earth?

The table includes post-USSR chips too:

> Newer devices from Russia, Belarus, and Ukraine are listed here if they are labelled according to the Soviet integrated circuit designation.

Derp, sorry I missed that. Thank you for pointing it out for me.

Is there a datasheet for these Soviet PDP-11 chips? I mean assuming I want to make something new with it now..

https://gid.pdp-11.ru/a/mpss/mpss86-04.html

https://gid.pdp-11.ru/a/mpss/mpss84-04.html

I don't understand the reset sequence of the KM1801VM1 and VM2, but it's clear for the VM3 (it starts at 0173000 or 026 depending on WO pin).

It looks like it has status signals so that program and data memory spaces can be split, the English translation has this funny comment:

"When working with weapons of mass destruction, system memory is “hidden”, i.e. its addresses in program mode do not occupy RAM address space. "

Can you paste the original russian and i'll try to translate better? i suspect that " weapons of mass destruction" was a mistranslation of an acronym to WMD, which got then expanded

Yeah, it should be "single-chip microprocessor" (ОМП / однокристальный микропроцессор).

Many PDP-11s had "split I&D" (Instruction & Data) capability built into the MMU so you could address 128k memory instead of only 64k. See: https://gunkies.org/wiki/PDP-11_Memory_Management

elbrus 2000 is real power

The Elbrus 2000 is rather fascinating, but they can't even make them anymore. Perhaps the older models, but the newer ones where all manufactured by TSMC, and Russia doesn't have the ability to create anything below 90nm.

There was an article about some Russian bank testing out the e2k, but they reported that the Elbrus-8C, was not acceptable. The 8C is 28nm, so made by TSMC, so even if it could perform it would be available.

I'm still curious about the chip though.

the slow death of elbrus is a very sad story, similar to the usa's efforts to revive its interplanetary space travel program

Or Russia's efforts to revive it's expansionist military. Yes, very sad indeed.

the only way this makes sense is if you hate both elbrus and the us space program, since someone who wasn't being sarcastic about russia's military wouldn't term it 'expansionist'

Keep trying.

you're just posting hate in order to troll

Sberbank testing seems to be blown out of proportion by Wikipedia-studying analysts. Or someone else, for some other reason. It's easy to guess that certain “patriotic” circles had decided to make Sberbank support the production by signing some big long term contract, and they decided to NOPE out of it, and quickly made a critical assessment in comparison with top of the line commercial offers. Effective performance of Elbrus in various tests was neither a secret nor a surprise.

The problem isn't performance, it's the architecture. Supposedly, it's intended use case is being a processor for radar control systems with lots of floating point number-crunching/DSP/whatever power to process input data in real-time and also run some general purpose code based on results, all in a single package. Other tasks rarely need that combination of features, it's either top general purpose performance on x86 servers, or top number crunching performance on supercomputers.

Random links:

http://www.elbrus.ru/files/521dd0/256487/1a773c/000000/2014-...

More detailed on page 100 in this book (with many details on other pages, obviously):

http://mcst.ru/doc/book_121130.pdf

Random paper on optimizations:

https://doi.org/10.14529/mmp200109

I remember seeing some benchmark results some time ago in which optimized x86 code running via Elbrus software+hardware translation was faster than native Elbrus compile for the generic C code. Unfortunately, VLIV compilers must be very smart, and ingest all the software and hardware optimizations from others to compete with common architectures.

Just adding the link to the Toms Hardware article on the Sberbank tests: https://www.tomshardware.com/news/russias-biggest-bank-tests...

thanks for all this great information!

Reply on news.ycombinator.com