Originally Posted by moustang
There are LOTS of things to explain the differences.
I find it odd that so many people on such an enthusiast site as this seem to not understand benchmarks. When you run a benchmark what do you think you are benchmarking? Your GPU, your CPU, or every single component in your system? The answer is you're benchmarking every single component in your system. Differences in motherboards will show in a benchmark. Differences in RAM speed and RAM timing will show in a benchmark. Differences in memory bus speeds, PCI-e bus speeds, and I/O speeds will show in a benchmark. Even something as simple as using a separate sound card rather than an onboard sound chip will show in benchmarks. You're never benchmarking just your CPU and/or GPU. If you were just benchmarking your GPU then your benchmark would show perfectly linear scaling as resolution changed, which never happens.
As I said, they're using the same CPU as I am... which means they're using the same chipset as I am, have the same PCI-e bus speeds, the same I/O speeds, etc. They're clocked the same as well and as I also said, their memory speeds are if anything higher (and the timings barely suffer for it; 3400 MHz vs 3000 MHz, 16-18-18-38 vs 15-17-17-35) plus they've got the same amount (16 GBs), therefore they've got better memory. In fact, their motherboard is a higher-end board from the same manufacturer's Z170 lineup. And I've neither touched cache overclocking nor the speeds of anything else on the board. Their CPU, memory, chipset, and board are all-around superior or equal at minimum.
The GPUs were also matched in both claimed core and memory clock speeds and beyond that, there is nothing that Firestrike could test. It is a primarily processing speed benchmark that's low on local memory usage (and therefore does not in any way benefit from storage drive speed) and has two distinct measurements: CPU physics processing and graphical rendering. That's it.