tag:support.primatelabs.com,2011-01-31:/discussions/geekbench/33189-multi-core-vs-single-core-scoresshould-the-multi-core-score-be-roughly-single-core-x-of-coresPrimate Labs: Discussion 2019-07-02T07:58:44Ztag:support.primatelabs.com,2011-01-31:Comment/471464212019-07-02T07:50:42Z2019-07-02T07:50:44ZMulti Core vs Single Core scores....Should the multi core score be roughly single core x # of cores?<div><p>Some things that can significantly negatively affect multicore score:<br>
- L3 cache contention, both time and space. Large amounts of the silicon are assigned to dealing with this. - Main memory contention, time and space. More channels helps but this one will always be expensive to improve. - Thermal throttling. Could be blocked vents but these days is mostly designed in for cheapness or lightweight so happens all the time.</p>
<p>Any one of those can be the chief bottleneck. Sometimes workload dependant.</p></div>joe9000tag:support.primatelabs.com,2011-01-31:Comment/471464212019-07-02T07:58:41Z2019-07-02T07:58:43ZMulti Core vs Single Core scores....Should the multi core score be roughly single core x # of cores?<div><p>What would be helpful to making sense of the discrepancy is a second number paired to each score. This extra number would be the average clock rate of the cores during testing.</p></div>joe9000