Lead is not magic. Arguably the best protection against radiation in space would be borated concrete, or borated water. Light nucleii give you more cross section per kg and nucleii such as boron and zirconium have anomalously high appetites for particular kinds of radiation. Lead is good for gamma rays, IIRC, not for cosmic rays or solar wind.
But you need a lot of shielding. The atmosphere at sea level is equivalent to 10m of water above us, and yet we still have significant radiation. Boron is much more effective, but you still need a lot of it and it does not stop all kinds of radiation.
So, if you want to go into space you need chips which are inherently resistant to radiation, which is not as difficult as you might think. Most radiation events are transient ionization with no structural damage, so what you want is a device which is not perturbed by radiation. This is why DRAM (with thousands of electrons per cell) is much less bothered by radiation than SRAM (where you can flip a junction with a much smaller electron cascade). So long as your device will not lock up or fuse on such electron bursts your chip will do pretty well. As for redundancy, in many cases what you need is just error detection and retry of transactional functionality. This can be as simple as the ECC in an SRAM or as complex as the retry mechanism in a database. The hard part is finding a general solution to detecting the error. Memory errors are easy - but how do you know if your multiplier just glitched? In practice the solution may be a combination of a tough process along with the usual case that logic gates are less susceptible to radiation than SRAM and spend most of their time in idle states which do not contribute to results, so the error rate will be very low.
In cases where it really, really matters and where there are real-time constraints on delivering safe results, that is where schemes like triplicated voting are essential.
