Ubuntu 16.04 vs. 18.04 Performance On Six Systems

[phoronix]

Phoronix: Ubuntu 16.04 vs. 18.04 Performance On Six Systems

Continuing on with our benchmarking of the recently released Ubuntu 18.04 LTS, here are some reference benchmarks on a total of six systems with AMD and Intel hardware while looking to see how the out-of-the-box performance compares to the previous Long Term Support release, Ubuntu 16.04.

http://www.phoronix.com/vr.php?view=26251

[squash]

I can see why you'd do it, but the Go benchmarks feel a bit weird. Go as a platform isn't the sort of thing people tend to just use wha apt installs. Since Go compiles to a binary, it isn't dependant on the expectation of a complex runtime installed by the OS packagers. So while the benchmark results are true, their usefulness seems questionable. More than anything you're comparing Go 1.6 vs Go 1.10. Go 1.6 is a distant memory for people who actually use the language, much more useful would be the performance of Go programs running under each OS having been compiled with an identical (current) version of the suite.

[bbiiggppiigg]

Can you shed some light on why the performance of SQLite on Core-i7 7800k for Ubuntu 16.04 is such an outlier ?

If it is a hardware-related issue, Ubuntu 18.04 seems to dealt with it just fine.
If it is a os-related issue, Ubuntu 16.04 seems to have a stable performance (around 30-40s ) for all other generations of CPU.

[Guest]

Thanks a lot. I appreaciate the Go tests. Interesting to see Go performance improving this much.

[Royi]

On the test, do you compile the files?
If so you compare the different compilers on each platform and not the OS itself.

[randomizer]

On the test, do you compile the files?
If so you compare the different compilers on each platform and not the OS itself.

And what is the OS itself?

[Royi]

Take application as is and run it on the OS.
The binary file should be identical if you are to compare OS.
OS is about services, not different binary code.