[simen]

They're comparing to an in-order CPU. Given that most CPUs are out-of-order (at least of the non-embedded variety, and GC is less used in such applications anyway), it would be better and more intellectually honest to actually compare to a typical CPU that performs GC. They kind of address this in the paper but only in a short aside: "Note that previous research [1] showed that out-of-order CPUs, while moderately faster, are not the best trade-off point for GC (a result we confirmed in preliminary simulations)." So they don't quantify what any of this means.

I think it's an interesting idea, but it doesn't bode well when they seemingly choose the wrong target for comparison and hand-wave away the difference as insignificant.

[reitzensteinm]

The comparison at least in the abstract is energy efficiency. It's quite likely that a small in order CPU is very good at chasing dependent pointers around the heap for its power consumption.

Imagine a linked list. Each pointer access is likely to miss to main memory, and no concurrency is possible. Both the highest and lowest end cores will sit around making a single request every 80ns.

They claim that the comparison was to the best alternative and I'd probably take them at their word barring any specific evidence.

[klodolph]

I think an interesting comparison here is GPU cores, where a core will get blocked on a memory access and it will switch out its state for another. It looks like this is the approach here, which is a bit more aggressive than ordinary out-of-order approach. It's less ordered.