Synthetic testing
Synthetic and real-world testing
On this page, we'll grab a couple of benchmarks from the regular test suite. Please , you will only need to focus on the grey bars in the charts, the rest in there is for your understanding of performance scaling, otherwise, they serve no purpose.
For the first test I figured, wPrime serves it all, a huge load on both the CPU and memory subsystem. The post-patch was slower, but overall it isn't even one second. A marginal difference at best would describe it. I also added the Core i7 8700K on the Maximus X Formula in a few test runs, (this one has the new 1003 BIOS).
Above the Mozilla Kraken test, it simulates really heavy browser usage. The result is listed in Milliseconds, and that would be 3 of them. Again, a completely irrelevant difference that could very well be a random difference.
Compression then, here you typically have a lot of file IO, CPU and memory dependency. We use a threaded test, and yes, there is a 2MB difference between the two patches. This is a 5% differential, and that makes it significant. Would you in real-life ever notice that though? No, not very likely.
Decompression is much simpler as a task, an algorithm that likes memory, here there was no difference spotted/measured.
Cinebench R15 then. Single threaded we drop a point or two. We can confirm that all results for the patched OS are slower, but a result like this is simply not significant whatsoever. I also included the Core i7 8700K in here (ASUS Bios 1003).
Cinebench R15 multi-threaded, yes we drop a few points again. But again this is not even remotely significant. The generic consensus, however, remains that all tests show slightly slower performance, that is a fact.
For the last synthetic test, I opted Frybench, frybench would probably be a perfect match benchmark to test the patch as it uses CPU and memory heavily and is a drag on the entire PC. Seen from overall results, it again is small in performance differences. But the difference is 3 seconds, and that is significant in the sense that this is over 1 maybe 2% of a perf difference. Let's run some game related benchmarks though.