The Elbrus 2, which was the first Soviet supercomputer, didn't come until 1977. Earlier systems, like the BESM, were mainframes - designed more for data handling than compute power. It looks like 355 BESM-6:s were made, ending in 1987. There were also Minsk mainframes.
There were also clones of US hardware, like the ES EVM clone of the IBM 360 (production didn't start until 1972), and PDP-11/VAX clones. Those were after 1970.
I'll use 50 BESM-6:s as my shot-in-the-dark estimate. Those seemed to be 1 megaflop (see https://en.wikipedia.org/wiki/BESM ), so the USSR could probably do about 50 MFLOPS in 1970.
Or no more than three iPhones.
For 1985 it's much harder to even guess. Is it even reasonable to include the aggregate sum of the microcomputers of that era? Not if you want to simulate an H-bomb. Only you can answer that.
>> Is it even reasonable to include the aggregate sum of the microcomputers of that era? Not if you want to simulate an H-bomb
I think so. Since no one's going to be simulating an H-bomb on an iPhone anyway. If we are going to argue maybe there are soldiers trapped in a nuclear-proof bunker with a nuclear bomb and before they launch it they want to simulate it, and all they have is an iPhone, we can argue the same case for a microcomputer.
I was being a bit facetious. My underlying point was that the network overhead on a microcomputer likely overwhelms a "Beowulf cluster" (to use a 1990s term) of them, so 10,000 microcomputers in practice won't be anywhere near 10,000x the performance of a single microcomputer.
What then is the goal of estimating the aggregate raw MFLOPs for the USSR?
As a side point, Apple doesn't allow a compiler or language interpreter on their iPhone, so unless they either had the app beforehand, or access to a development machine, they won't be able to anything with it. While if they had an Apple ][ they could start right away in Apple BASIC on the ROM.
The Elbrus 2, which was the first Soviet supercomputer, didn't come until 1977. Earlier systems, like the BESM, were mainframes - designed more for data handling than compute power. It looks like 355 BESM-6:s were made, ending in 1987. There were also Minsk mainframes.
There were also clones of US hardware, like the ES EVM clone of the IBM 360 (production didn't start until 1972), and PDP-11/VAX clones. Those were after 1970.
I'll use 50 BESM-6:s as my shot-in-the-dark estimate. Those seemed to be 1 megaflop (see https://en.wikipedia.org/wiki/BESM ), so the USSR could probably do about 50 MFLOPS in 1970.
Or no more than three iPhones.
For 1985 it's much harder to even guess. Is it even reasonable to include the aggregate sum of the microcomputers of that era? Not if you want to simulate an H-bomb. Only you can answer that.