What about non-reclaimable performance losses?
What about non-reclaimable performance losses?
Posted Mar 20, 2025 10:23 UTC (Thu) by paulj (subscriber, #341)In reply to: What about non-reclaimable performance losses? by NYKevin
Parent article: Better CPU vulnerability mitigation configuration
Not impossible in practice at all. Indeed, it is sufficiently practical that Sun Microsystems built a product line around such a CPU - SPARC T1 to T3. Sucked for single-thread performance, but they were fairly good at highly-parallel serving workloads, particularly on perf/W.
Posted Mar 21, 2025 18:46 UTC (Fri)
by NYKevin (subscriber, #129325)
[Link]
> and every program whose performance mattered was able to use all of the extra cores.
What about non-reclaimable performance losses?
Sorry, I should have quoted even more selectively than I already did. I specifically meant that *this* is impossible: