Originally Posted by Levelog
I can't believe people are still using passmark to gauge gaming
performance. You use games or game engine benchmarks to gauge gaming performance. Just like you'd use render based benchmarks to gauge render performance. Passmark for gaming is about as helpful as looking at spec sheet stats.
Anyone who's doing that, it's a very bad idea. Passmark tries to max out every available resource on a CPU and then generates a performance rating based on that. Even in the single-core test, it maxes out everything it can on one core. It's valid as a broad-based performance benchmark, but gaming is a specific task that only uses a subset of a processor's cores and instruction sets. In nearly all cases, Intel CPU's are stronger in that particular subset than AMD.
We discussed this before on the AMD 860K thread. If you OC an 860K to 4.5 GHz, it delivers Passmark ratings better than an i5-4460. If you actually game on it, it performs more like an i5-2400, which isn't awful, but it's about 15 percent below a 4460. And that's using my 6970. On a newer card with higher performance, I suspect the gap might be a little bigger, not that anyone's likely to be using an R9 290 or a GTX 970 GPU with an 860K.
Just posted this for the information of anyone who isn't familiar with the Passmark benchmark and what it does. If you already know this, carry on.