We haven’t heard anything form the developers for a pretty long time, since late 90s. But nevertheless the project didn’t die out. Moreover, it is even close to the final stage now, as we read over here!
According to the Chief Executive of the Russian Elbrus MCST (Moscow Center of Sparc Technology), Eugenyi Babaian, the first working prototype of the processor might be ready by the end of the year already. The company plans state that the first sample processor will be manufactured with 0.18micron technology (supposedly on TSMC’s fab!) and will work at about 400MHz. Frankly speaking, it seems to be not very high for today, eh? However, the next versions should work at 1.2GHz and will be called E2K. It is about to happen some time in 2004. Later one when they move to finer manufacturing technology, the working frequency may increase up to 3GHz. The performance of this solution is expected to reach around 500 and 1200 points in SPECint95 and SPECfp95 tests respectively.
For a better comparison see: the regular Athlon 1GHz achieves around 42.9 in SPECint95 and 29.4 in SPECfp95. You can have a look at the comparative testtables here. Note that the 3GHz version is very unlikely to come out soon, however, even at lower frequencies the new CPU will prove not bad at all.
I would also like to say a few words about the architecture of the future CPU. As is known, the early processor versions were clones of Sun Microsystems solutions, however, later on the developers decided to take another way by working out a new architecture and instructions set. From this viewpoint, Elbrus reminds us a bit of Transmeta’s processors, as it uses "binary compilation" technology similar to "code-morphing" used by Transmeta to ensure x86 compatibility of its products. Due to "hyperparallel architecture" the CPU will be able to perform more instructions within a certain time slot than the processors from other manufacturers, even despite its lower working frequency. I would also like to stress that the situation with Intel Pentium 4 is just the opposite.
The new processor will be able to perform up to 24 instructions per clock, which is much better than by any of the 6 existing analogues from other processor developers. According to Boris Babian, the head of Elbrus Developer Team, the competitors are very-very far away from any close results.
However, things are not so cloudless for the Russian project. In the first place it is about the financing. Even if the Russian government could have provided $100 million for the new architectures development, the marketing campaign would have required over 1 billion dollars. That is why the only way out for the Russian Company is to cooperate with the abroad partners, which might be interested in investing into this project.
Among Elbrus companions already are such well-known companies as: Avant!, Cisco Systems, Infineon Technologies, Sun Microsystems, and Transmeta Corp. the officials from Sun and Transmeta believe that this project will live and that the developer team is strong enough, however the success anyway depends on the company’s ability to gain finance. Anyway, the support of famous companies like that has already increased the chances of the Russian project.
Elbrus Company in its turn doesn’t comment in any way on the matter.
It looks as if things were not so hopeless for Russia in the computer field and if everything goes the right way, the Russian product may come into life one day and become a worthy alternative to the existing solutions. We only hope that Elbrus will be doing better and we will have the right to say that there is something we are proud of in the semiconductor market.
Note that an analyst from Mercury research thinks E2K has very high chances to gain public recognition, as Elbrus MCST are known for their excellent achievements in the past.