Anon (no.delete@this.spam.com) on September 18, 2020 3:25 am wrote:
> Andrei F (andrei.delete@this.anandtech.com) on September 18, 2020 1:04 am wrote:
> > I just want to add on that sentence as that's not what I wrote
> > in the piece: I don't think the memory subsystem is to blame.
> >
> > It's significantly stronger than ICL and showcases *much* better DRAM latency and significant
> > single core bandwidth uplift. 429.mcf showcases great scaling well beyond clocks, showing
> > that latency for example is not to blame. In my opinion it's a regression *because* of the
> > reworked cache, as essentially the L3 is now 20% slower per clock versus ICL.
>
> Maybe, SPEC2006 dataset size had too much influence.
> https://images.anandtech.com/graphs/graph16084/117493.png

In what way? Too big or too small? Generally 2017 is less sensitive to memory at the extreme cases, but it has larger footprints on average.

2006 has a more varied characteristic to memory pressure, for example 429.mcf uses a 1.8GB active memory region while 505.mcf_r is something along the range of 800MB if I remember correct, the 2006 test actually stresses memory a lot more.