AMD Ryzen 9 3900X vs. Intel Core i9 9900K Performance In 400+ Benchmarks

[phoronix]

Phoronix: AMD Ryzen 9 3900X vs. Intel Core i9 9900K Performance In 400+ Benchmarks

Given the recent AMD "ABBA" Ryzen 3000 boost fix, the upcoming release of Ubuntu 19.10 powered by Linux 5.3, here is a fresh round of AMD Ryzen 9 3900X vs. Intel Core i9 9900K benchmarks in a side-by-side matchup . It's just not any comparison but our largest i9-9900K vs. 3900X comparison ever: 112 gaming benchmarks and 321 system/CPU benchmarks carried out for our most extensive look yet at how these ~$500 CPUs are competing in this fierce race.

http://www.phoronix.com/vr.php?view=28314

[Niarbeht]

Do I want to know how many hours this took?

[Michael]

Do I want to know how many hours this took?

Over a week... Not too bad with PTS automating it all, but still time consuming coming up with the set of tests to use, going through and verifying all the results, etc.

[Michael]

P.S. the 300+ benchmarks are also being re-run as well with "no mitigations" on each processor as a follow-up comparison....

[Niarbeht]

Over a week... Not too bad with PTS automating it all, but still time consuming coming up with the set of tests to use, going through and verifying all the results, etc.

And I skipped right to the summaries and I was done in under a minute! :P

[GruenSein]

While I appreciate these benchmarks, their key finding is less about the CPU itself and more about the question whether or not the particular software is threaded efficiently. This shows in benchmarks know to scale well with cores, in which Intel gets blown out of the water.

Also, I like the separate summary for gaming and other benchmarks. Another interesting distinction would be synthetics (interesting but not noticeable by the user) vs. real work loads like compiling, rendering etc.

[SyXbiT]

Excellent benchmark Michael. Glad to see real AAA game benchmarks vs just the open source games.

[schmidtbag]

Pretty comprehensive stuff. That's a lot of data.

Over a week... Not too bad with PTS automating it all, but still time consuming coming up with the set of tests to use, going through and verifying all the results, etc.

Well, now that you came up with the tests, I suppose you could just have them pre-scripted so you don't have to do that a 2nd time around.

Out of curiosity - how does the geometric mean work when it comes to comparing the "lower is better" vs "higher is better" graphs? Does it account only for the percent improvement over the other product? Because otherwise I imagine those results could start counteracting each other since they're "graded" inversely.

[Michael]

Out of curiosity - how does the geometric mean work when it comes to comparing the "lower is better" vs "higher is better" graphs? Does it account only for the percent improvement over the other product? Because otherwise I imagine those results could start counteracting each other since they're "graded" inversely.

For the geo means, the LIB results are 1/result, in order to have them all be higher is better.

[caligula]

While I appreciate these benchmarks, their key finding is less about the CPU itself and more about the question whether or not the particular software is threaded efficiently. This shows in benchmarks know to scale well with cores, in which Intel gets blown out of the water.

It's also quite likely that the future versions of applications will focus more on threading, not the other way.