Because benchmarks are benchmarks, has better multithreaded performance than a current i3 and it cleans the floor of any iGPU Intel has on any of its chips.
Oh, and its almost four(!) times(!!) cheaper(!!!) than that old Core 2 Quad.
http://www.amazon.com/Intel-Q9650-Pr...e+2+Quad+Q9650
With that price I could build a whole system...with a crappy looking case -.-''
I only recommended the APU cause I like iGPU's just in case the dGPU dies or something.
But, if Intel is the choice. Wait for Haswell and get an i7 version. Then you will be "set" for quite awhile.
The processor we are talking has part of its name rhyme with "Hi Five", which even AMD advocate vampirr was able to pick up.
I wasn't talking about price or suggesting he buy it. I was talking about performance(aka speed) and how Trinity barely surpasses an old Intel CPU. Assuming Intel kept improving their processors in later generations, that means that there is no way an APU can surpass an i5.
Similarly, a nerfed 6670 does not compare to a higher end Radeon, as the nomenclature itself dictates very distinct performance tiers. Never mind that the architecture is different and the difference between the number of stream processors in the 7660D and 7850 is vast.
There is no closer approximation to processor speed prior to purchase than benchmarking. Of course, one benchmark is too specific to one application, hence you throw an entire suite at it. Yes, sometimes there are inaccuracies, in which the you simply just penalize the score that you get by some reasonable amount to account for variance and whether the number is overstating or understating something.
The benchmark gap(subtract the larger from the smaller) and the benchmark ratio(divide the larger by the smaller and subtract by 1) between two processors is what matters more when comparing two processors. If there is a big enough difference between the scores, then the hypothesis that one chip is superior to the other and not being just "noise in the line" becomes stronger(see the Pentium 4 and early Core 2 Duo scores vs anything reasonably recent).
Between the i5-2400 and A10-5800K the gaps in fps in games, frames encoded per second for the video encode, Cinebench R10 scores in both multi- and single-threaded performance, the A10 cannot compare to the i5-2400.
But let's not use benchmarks then(your way). Then, there is no way to tell what performance you are getting until you assemble the box. Hence, you are trying to justify suggesting to a man to piss away a good 300-500 dollars for a complete downgrade in all components on the basis that "benchmarks are benchmarks" and hence all processor are indistinguishable and everything is luck-based.
There is the very real possibility that he is actually sacrificing much of the performance he already had. So, if we throw benchmarks out of the window, one can claim the 6670 is better than the 7850, and no reprocussions will occur until some poor fellow gives it a whirl and sees crappy the crappy fps.
Between approximating or not approximating at all, I think it's quite clear that using a reasonable estimate is better buying like a chicken with its head cut off.
Facebook
Twitter