It sounds like the Russians are doing just fine. The expectation that they would create an equivalent or superior processor to established technologies is kind of humorous.
Being within 2 -3 times slower than the top of the line is quite reasonable for a new chip.
The Java results being so much slower are likely the difference in optimizing for new arch, bot neseccarily the chip itself.
Given that many apps lice quite nicely on VM and containers (using a lot less capacity then is available from the hardware), I think this is respectable to start with.