WCG in numbers - power

In this post we provide some behind the scene numbers about WCG


Have you ever wondered how big a computer should be to equate the World Community Grid computing power? For all of 2020, the average power of WCG was the equivalent computing power of a cluster running at 100% capacity 24x7 for 365 days with 11,999 nodes with 1 Intel Core i7-9700K CPU @ 3.60GHz. Indeed, GPUs could substantially increase this power, as shown in OPN1 March 8 post.

How is this amount distributed across volunteer device configurations? Table 1 shows the relative amount of computing power produced by computers with different RAM and CPU configurations. This calculation does not differentiate between CPU and GPU computing power, and about 35% of the computing power at the time of this report comes from GPUs

Top 5 percentages are marked in BOLD (which together are 56.9% the power of WCG): 10.4% of our computing power comes from computers with 64GB or more RAM and 32 or more cores, 35.0% from computers with 32GB-63GB RAM and 16-31 cores, 68.6% from computers with 16GB-31GB RAM and 8-15 cores and 93.1% from computers with 8GB-15GB RAM and 4-7 cores. 

What platforms are more frequently used? 71.11% of the computing power is provided by Windows OS with X86/64 processors, 25.51% by Linux OS (25.06% with X86/64 processor and 0.45% with ARM processor), 0.01% by FreeBSD OS with X86/64 processor, 2.83% by Mac OS (2.78% with X86/64 processor and 0.05% with M1 processor) and 0.54% by Android OS with ARM processor.