Ryzen 7 1700 Verses Xeon E5-2650v4 for CFD-ACE+ VoF

These are not exactly the newest processors around but currently have a microfluidics job to run with CFD-ACE+ VoF and was interested to see which processors we had sitting around the office would run the best.  The VoF model was running in the order of 900k cells with high order surface reconstruction, surface tension and varying contact angles.

The results are still quite interesting as they show the interplay of CPU clock speed verses memory bandwidth.  At the lower core counts the Ryzen is some 30% faster than the Xeon, which is what you would expect given the differences in the clock speeds for the two chips.  However, as the number of cores in use increases the memory bandwidth and the supporting architectures become increasingly dominant the the Xeon caps out at 30% faster than the Ryzen.

And yes, there are eight (8) cores in the Ryzen but this particulare ACE+ job really did not like to run on all eight cores, hence the stop at seven.

Raw Run Times as a function of cores
Relative Run Time as a Function of Cores
Times Speed Up as a Function of Cores
Run Time Differences in Seconds as a Function of Cores
Percentage Run Time Differences as a Function of Cores

 

 

Site Footer