I'm not a system or network admin but this seemed like the best place to ask this question. If not please direct me to where because I have been banging my head against the wall trying to figure this issue. The issue being the server with faster CPUs is running my processing code in double the time (slower)...
I have 2 servers.
1st(older server):
- 2x 7402 EPYC CPU
- Gigabyte Motherboard (R182-Z92-00)
- 16x 16GB ECC DDR43200 (SK-Hynix)
- 1TB NVMe drive
- Ubuntu 20.04 - 5.4-107 (or something like that, sorry its not right in front of me right now)
- BIOS(2.18.1264 American Megatrends Inc)
2nd(newer server):
- 2x 7F72EPYC CPU
- SuperMicro Motherboard (H11DSU-iN)
- 16x 16GB ECC DDR4 3200 (Samsung)
- 1TB NVMe drive
- Ubuntu 20.04 - 5.4-107 (or something like that, sorry its not right in front of me right now)
- BIOS (2.20.1275 American Trends Inc)
What settings are available in the BIOS differ a bit but I tried to match as best I could:
Spreadsheet Image
For both servers I made sure to have all the same installations following the answers here. On both servers I'm using g++/gcc 10. With mkl for lapack, openblas-openmp for blas. In my code I'm using FFTW 3.3.10 and Armadillo 10.8, which were both installed using the same configure commands. Are there other checks I can make? Is it just because the 2 systems have different motherboards and one is just that much slower? Any advice would be greatly appreciated.
