More anticipated than the sixth iPhone

Great news came out last Friday. Izhevsk Radio Plant released the first Russian PCs.

The fact that they were put into production was written six months ago. But the fact that they can be "felt" has become known only now.

For me, this is the IT news of the year. And that's why.

First, you have to start somewhere. Learning to do finished high-tech things is an achievement in itself. Once started, you can incrementally move on. By the way, the price of 200-400 thousand per PC is cheap. For an experienced batch, this is ridiculous money.

Secondly, it seems to me that now is the time when it is time to finish whining about the "element base". I do not want to be misunderstood. The elemental base is important. But another thing is much more important.

Element base is now available as a service. You can "print" anything you like in China. Or even with us, as you see. Even in smaller volumes and outdated technologies.

And this is enough for many applications. Because now super-duper is a modern element base - and even microarchitecture! - does not matter as much as 15 years ago. Technologies even 10 years ago are acceptable for solving a large number of modern tasks.

This is all because software began to rule the world. Good software can compensate for the disadvantages of "bad" hardware. Everything around is gradually becoming “software-defined”, software-defined. Today we are already moving towards software-defined networking, software-defined storage, software-defined data center. Ultimately, we will have software-defined everything. If you want, there will be software-defined microprocessing. By the way, we almost have it already. So far in the laboratories, but also on the market, he will certainly break through.

However, where does the PC based on Elbrus? In my opinion, Elbrus is on the right track (under the current conditions, of course). They make architecture that Russia can afford. This is an architecture with static planning (a wide team, the so-called). It allows you to have simple hardware, but compensates for this simplicity with software.



Russia now cannot afford superscalar architecture. This is too complex iron. But in Russia they know how to write software! Repeatedly proven.

I think that building the first Russian PCs on the basis of a broad team is logical and logical.

True, static planning will have to be avoided in the future. I apologize to all the fans of the broad team, but this is a dead end. The machine must be dynamic.

The right thing about a wide team is that it is software-defined. (By the way, the use of binary translation in Elbrus to ensure portability of applications from the x86 architecture is also part of a software-defined approach to the construction of microprocessor architectures).

Software is what every hardware developer should think about in Russia now, however paradoxical it may sound. If we recognize this and accept it, we will have a greater future in computing.

The UPDATE . From the comments I see that I was misunderstood by many people. I didn’t mean that "you can take bad hardware, write good user applications for it, and thereby get good performance." Speaking of software, I mean not user applications, but middleware, which, in a sense, is part of the architecture of the machine. In relation to a wide team, such software is the compiler. The program execution planning is completely shifted to it. Iron is not involved in planning.
In my opinion, as many functions of iron as possible should be sought to shift to similar middleware. Ideally, it should be designed in conjunction with hardware (the so-called software-hardware co-design). This approach has many advantages. But this is beyond the scope of the post ...