Ok, so I´ve been using the "wrong" version too? Ok, I´m going to try 2.78b now. Edit: Holy Cow, it´s a difference between day and night... 00:46.34 The same settings I used for my first benchmark: Intel i5 6600k @ 4,5Ghz 32GB 3000 Mhz RAM @ 2666 Mhz 14-14-14-32 Now this looks a lot better.
As did I, back in the day. TBH though, when small update can produce that much of a difference, I still question Blender's suitability as a benchmark.
Given how varied the results were on my Pentium I'd be inclined to agree. I'd expect some variation, but not that much.
Not utterly useless - a learning experience The next time the various elements will be specified more fully.
You are right, blender is not really designed to be a benchmark. Also its quite possible that different compiler options (or even a different complier) can give better or worse results. So it was fun, but easy to have an apples versus oranges result.
Gaming results will make or break ryzen is a honest assessment. As the amount of people who can use 8 cores and 16 thread is very limited. It has to match the 6600k equivalent chip across a wide range of games. If it does this AMD have a winner. ( if priced correctly) if not it will be largely ignored, the early leaked screens had it well down on this cpu though in games. 6600k is quicker than the 6900k they are benching against in all games tested on most websites.
What I would really like to see is Geekbench result for Ryzen, which would show single thread performance. Perhaps we should start a third Geekbench thread after Ryzen hits the shelves.
More interesting in that test is how well it scores vs the old AMD chip. I would of expected more gains in truth. As the 8370 I assume is stock always did show well on this. 15k gain compared to this is low side? Considering the extra cores is less than 2k per thread. That old chip scores 3031 points per thread. New chip down at 2479.5. Which compares to the 6600k which is at 2571.2 per thread.