The short answer is yes—core count matters. The longer answer is that computer core count matters, and is particularly beneficial when you have software that is written to take advantage of multiple cores and if you have a sufficiently taxing workload. How do we know? Because we’ve seen the benchmarks. We share the test results in this post.
Core Count Matters with ANSYS Mechanical
ANSYS is a leader in computer-aided engineering (CAE) simulation software. Its product, ANSYS Mechanical, is written to take advantage of multi-core systems. ANSYS Mechanical is a finite element analysis (FEA) tool that allows designers and engineers to perform structural analysis of physical systems in a software simulation.
As part of this structural analysis, FEA calculations can take hours or days to perform. Because these delays are a real hit to engineers’ productivity, engineers have a vested interest in accelerating these calculations. But how can they get ANSYS Mechanical to perform complex calculations faster?
By upgrading to a system with more than two cores. Benchmark tests performed by ANSYS show a steady increase in speed as more cores are added. Here’s the scoop.
ANSYS engineers ran five benchmark tests on a two-socket workstation with 22 cores per socket (using the Intel Xeon processor E5-2699 v4). They ran a first round of testing with all but two of the CPU cores disabled, because two-core systems are the most common configuration on workstations used for FEA, according to experts at ANSYS. The engineers then re-ran the tests with 8, 16, and 32 cores enabled.
The baseline two-core configuration ran a simulation in 17 minutes, 9 seconds. The 32-core system ran the same simulation in 2 minutes, 23 seconds—seven times faster. At those rates, a complex analysis in ANSYS Mechanical that might require 7 hours to solve on a 2-core system would require less than 1 hour to solve on a 32-core system.
Think of it another way: two busy engineers are working side by side. One uses a legacy two-core system and the other uses a modern workstation with 32 cores. In the time it takes the first engineer to run one simulation, the second could run up to seven.
Sure, in the real world of CAE, nobody runs simulations back to back like that. But the performance difference is still striking. Why wait hours for a simulation that could be done in minutes with a processor upgrade? For more benchmark results and details about the testing, read the full paper: “Engineer Productivity Boosted by Higher-Core CPUs.”
ANSYS makes available a suite of benchmark tests that allow decision-makers to measure hardware performance data by using problems that are intended to represent typical usage. Find more test results from ANSYS.