I was curious if anyone has tried implementing the GC mark phase with SIMD instrs?
If so, what were the results, and why didn't it get accepted?

At first thought, it seems like a simple project that could improve performance of big-memory programs by a significant margin, right?

On 02/09/2017 1:42 PM, Márcio Martins wrote:
> I was curious if anyone has tried implementing the GC mark phase with SIMD instrs?
> If so, what were the results, and why didn't it get accepted?
> 
> At first thought, it seems like a simple project that could improve performance of big-memory programs by a significant margin, right?


I'm not so sure it can be.
https://github.com/dlang/druntime/blob/master/src/gc/impl/conservative/gc.d#L1955