4 ms·
I verified Geekbench results to be very tightly correlated with my use case and workloads (JVM, Clojure development and compilation) as measured by my wall time
by jwr 2y ago
I verified Geekbench results to be very tightly correlated with my use case and workloads (JVM, Clojure development and compilation) as measured by my wall times. So yes, I consider it to be a very reliable indicator of performance.
- stogot 2y agoCurious how you verified that? I should possibly do thesame
- qeternity 2y agoRun Geekbench on a sample of hardware. Run your workload along same hardware. Regress.
- m00x 2y agoThat's not very scientific at all. With how close the CPUs are, how would you compare the tiny differences?
- jwr 2y agoRun my compilation on a CPU, note down the time it took and the Geekbench score for that CPU. Run the compilation on another CPU, note down the time it took and the Geekbench score for that CPU. Now look at the ratios — if Geekbench scores implied the faster CPU was, say, 20% faster, is my compilation 20% faster? I'm looking at my notes and without digging too much, I can see two reasonably recent cases: 30% faster compilation (Geekbench said 30%), and another one: 40% faster compilation (Geekbench said 38%). So yes, I do consider Geekbench to be a very reliable indicator of performance.