David Kanter (dkanter.delete@this.realworldtech.com) on August 22, 2012 3:45 pm wrote:
> Suppose you had a TX that required simultaneously
> accessing 5 variables that map to the same set. In a scheme that just used a
> 4-way L1, it would always fail. OTOH, the same TX would be able to succeed on a
> MOB-based implementation.

Thanks, now I get it. We look at TX very differently.

You think about how to eliminate the limits in hardware, I
care about how to use it in software under the guarantees
given. (And that leaves me with 1 cacheline and no nesting.)

It was hard to understand what you wanted to show in your
article. The introduction didn't mention the MOB as goal or
vs. a cache based implementation at all and I didn't see
your view ("We want bigger transactions because... 70 writes
is ideal").