how come jboy has 50 more points then me in single core even though we are same speed? is that something I need to worry about?
edit: nm i just OC'd to 5.1 and now surpass him by 10 points, so i guess everything is fine.
There are a few factors at play;
1) First of all, the most obvious as
@jboydgolfer has already said; no two processors are identical, even with identical specifications. Impurities in silicon will result in varying levels of voltage, current, capacitance and resistance. Overclocking does rely on what is known as the 'silicon lottery' whereby a better 'draw' will yield better overclocking results. This somewhat ties into a technique known as binning, which is essentially how all the different models are made. For example, if one chip can achieve 4.00 GHz, but the other can only achieve 3.80 GHz, they will likely be turned into two different models. (Look back to Haswell, and see just how many Pentium, Core i3 and Core i5 models there were.) Binning also takes into consideration cache, memory controller, integrated graphics and hardware-level features such as simultaneous multi-threading (Intel markets this as Hyper-Threading). However, where the 'silicon lottery' differs from binning is that the 'silicon lottery' applies to multiple chips of the exact same model.
2) While unknown and probably not the case since he didn't showcase it, it may be possible that
@jboydgolfer altered settings beyond the processor frequency, which would aid with better performance.
3) The benchmark itself, I have noticed from my own testing, will give you completely different numbers over a number of runs. With modern, quad-core processors, the multi-threaded score can easily jump by 50-100 points because it's only a small percentage of the overall score. You're best to run it several times in succession and to draw an average, if you really care about it. And I think it might be worth mentioning that successive runs may yield higher scores for desktop processors, and lower scores for mobile processors.
4) Is Windows doing anything in the background?
5) And finally, margin of error. 10% is agreeably too high for a margin of error in this case, but I wouldn't worry about, say, 3-5%.
[...] I wouldnt worry too much about it though, its just a CPUz bench test, i dont put too much stock in them. I'd put more faith in a cinebench result, the CPUz tool (as i understnd it) ismore of a utility, than a benchmark. unless im mistaken.
Pretty much. As with all benchmarks, the result only really means anything if you will be doing the exact same workload as the benchmark uses. For CPU-Z, that's essentially scalar SSE2, and nothing more. So unless you plan on using software that only uses SSE2 instructions, I wouldn't put much thought into the numbers it spits out at all.
don't be mad at me mods, I want to share my golden chip
.
