unums

Alas, a lot of examples were lifted from the unums book and sort of square peg/round holed into posits.

This transcript is very interesting, and in particular I didn't know about Unums (Universal Numbers), which look amazing.

You'd need to unpack them (into fixed size unums?) first.

Are there any more details on unums available online (i.e.

> Unums have proven difficult to build efficient HW implementations for Valid point, but not quite true anymore.

Would working with unums side-step the issue for math problems?

Similarly in primitive types, perhaps some day with have unums.

Perhaps unums are the best choice for some problems.

I don't see unums replacing floats for, say, video game graphics.

But for numerical computation, it seems like the only real flaw with unums compared to doubles is the nonexistence of a hardware implementation, and the existing popularity of doubles.

How compatible are unums with modern processors?

Maybe I don't fully understand the issues at play here but I remember reading about so-called 'unums'[0], an alternative binary number representation that seems quite amazing.

There is a whole chapter on "fixed-size unums" and the issues of making them just as easy to index as arrays as are floats.

For hardware without a FPU, you can do floats in software when accuracy is more important than speed (vs fixed point), one can imagine that if this thing actually works then for hardware without a "UPU" one can just do unums in software when needed.

I would argue that unums are a "superior replacement" for doubles in many cases, though: in the case that you support unums that are "wide" enough, you can represent doubles exactly, plus you have additional values, plus some nice rules about when approximation error occurs/is propagated and not as many bits need to be stored or moved around on buses.

The encoding is easy to understand, and less common floating point formats such as bfloat16 and posits/quires (Unums) are supported.

Unums shouldn't be seen as better than ints, longs, floats, or doubles anymore than a double is seen as better than an int.

Unums do nothing to address that issue, except possibly by virtue of not having multiple implementations in the first place.